EDITOR-IN-CHIEF GUEST EDITOR-IN-CHIEF DIGITAL CONTENT MANAGER

Candice K. Silversides, MD, Michael J. Landzberg, MD, Kara McDermott, BA,

University Health Network, Adult Congenital Heart Disease, American College of Cardiology, Washington, DC, USA
University of Toronto, Palliative Care Cardiology, DIGITAL PRODUCT MANAGER
Toronto, Ontario, Canada Boston Children’s Hospital, Boston, MA, USA
Taryn Myers, PMP,
DIVISIONAL VICE PRESIDENT, PUBLISHING
American College of Cardiology, Washington, DC, USA
Justine Varieur Turco, MA,
DEPUTY EDITOR MARKETING MANAGER
American College of Cardiology,
Paul L. Douglass, MD, Washington, DC, USA Colleen Whipple-Erno, BA,
Interventional Cardiology, Health Equity, EXECUTIVE MANAGING EDITOR American College of Cardiology, Washington, DC, USA
Wellstar Health System, Inc., SOCIAL MEDIA COORDINATOR
Monica R. Payne-Emmerson, MS,
Wellstar Health System Center for American College of Cardiology, Sarah Khalaf, BA,
Cardiovascular Care, Washington, DC, USA American College of Cardiology, Washington, DC, USA
Marietta, GA, USA MANAGING EDITOR EDITORIAL SUPPORT
STATISTICAL EDITORS Denise Kuo, BA, Ania Bukowski, BAS,
Larissa I. Stanberry, PhD, American College of Cardiology, American College of Cardiology, Washington, DC, USA
Minneapolis Heart Institute Foundation, Washington, DC, USA Adam Etkin, MA,
Minneapolis, MN, USA DIGITAL PUBLISHING DIRECTOR American College of Cardiology, Washington, DC, USA
Zhenyu Zhang, PhD, Ron Schmelzer, Jr., MSPM, PMP, CSM,
Peking University, American College of Cardiology,
Beijing, China Washington, DC, USA

EDITORS-IN-CHIEF ASSOCIATE EDITORS Anju Bhardwaj, MD, SENIOR CONSULTING

JACC Gerhard-Paul Diller, MD, PhD, University of Texas/ EDITORS
Digital Health, Adult Congenital McGovern Medical School
Valentin Fuster, MD, PhD,
Heart Disease, Houston, TX, USA Heidi M. Connolly, MD,
Mount Sinai School of Medicine,
University Hospital Muenster, Adult Congenital Heart Disease,
New York, NY, USA
Muenster, Germany SOCIAL MEDIA Mayo Clinic,
JACC: Cardiovascular Interventions
Rochester, MN, USA
David J. Moliterno, MD, Uri Elkayam, MD, CONSULTANTS
University of Kentucky, Lexington, KY, USA Heart Failure, CardioObstetrics, Bailey Ann Estes, BSN, Marc Fisher, MD,
JACC: Cardiovascular Imaging USC School of Medicine, Neurocardiology,
RN-BC, RNFA, CNOR, RCIS,
Y. Chandrashekhar, MD, Los Angeles, CA, USA Harvard Medical School,
Hendrick Medical Center,
University of Minnesota/VAMC, Dipti Itchhaporia, MD, Abilene, Texas, USA Beth Israel Deaconess
Minneapolis, MN, USA Innovation in Digital Health, Medical Center,
JACC: Heart Failure Global Health, Kanwal M. Farooqi, MD, Boston, MA, USA
Biykem Bozkurt, MD, PhD, Hoag Memorial Hospital Presbyterian, Columbia University Irving
David R. Holmes, Jr., MD,
Baylor College of Medicine, Newport Beach, CA, USA Medical Center,
Neurocardiology, Interventional
DeBakey VA Medical Center, Patrick R. Lawler, MD, MPH, New York, NY, USA
Cardiology,
Houston, TX, USA Critical Care, General Cardiology, Richard Ferraro, MD, M.Ed, Mayo Clinic, Rochester, MN, USA
JACC: Clinical Electrophysiology Peter Munk Cardiac Centre, Johns Hopkins Hospital,
Kalyanam Shivkumar, MD, PhD, University Health Network, Baltimore, MD, USA Jane W. Newburger, MD, MPH,
UCLA Cardiac Arrhythmia Center, UCLA University of Toronto, Pediatric Cardiology,
Health System, Los Angeles, CA, USA Toronto, Ontario, Canada Michel Ibrahim, MD, Boston Children’s Hospital,
JACC: Basic to Translational Science Temple University Hospital, Boston, MA, USA
Luc Mertens, MD, PhD,
Douglas L. Mann, MD, Pediatric Cardiology, Imaging, Philadelphia, PA, USA Samuel C. Siu, MD, SM, MBA,
Washington University School
The Hospital for Sick Children, CardioObstetrics,
of Medicine, St. Louis, MO, USA Claudio Montalto, MD,
University of Toronto, University of Western Ontario,
JACC: Case Reports Toronto, Ontario, Canada De Gasperis Cardio Center, London, Ontario, Canada
Julia Grapsa, MD, PhD, Niguarda Hospital,
Garima Sharma, MD,
Barts Health NHS Trust, London, UK Milan, Italy Steven R. Steinhubl, MD,
CardioObstetrics, General Cardiology,
JACC: CardioOncology Digital Health,
Johns Hopkins
Michael D. Spartalis, MD, Scripps Translational
Bonnie Ky, MD, MSCE, University School of Medicine,
National and Kapodistrian Science Institute,
Perelman School of Medicine at the University of Baltimore, MD, USA
University of Athens, La Jolla, CA, USA
Pennsylvania, Philadelphia, PA, USA SOCIAL MEDIA EDITORS Athens, Greece
JACC: Asia Sadeer Al-Kindi, MD, Sean van Diepen, MD, MSc,
Case Western Reserve University Jonathan Windram, MBChB, Critical Care Cardiology,
Jian’an Wang, MD, PhD,
School of Medicine, Mazankowski Heart Institute, University of Alberta,
Second Affiliated Hospital, Zhejiang University
Cleveland, OH, USA Edmonton, AB, Canada Edmonton, Alberta, Canada
School of Medicine, Hangzhou, China
Updated March 2023
Nanette K. Wenger, MD, Aditya Bharadwaj, MBBS, Carolina A. Escudero, MD, MSc, Eric M. Horlick, MDCM,
Geriatric Cardiology, General Cardiology, Pediatric Cardiology, Interventional Cardiology,
Emory University Interventional Cardiology, University of Alberta, Toronto Congenital Cardiac
School of Medicine, LomaLindaUniversityMedicalCenter, Stollery Children’s Hospital, Centre for Adults, Peter Munk
Atlanta, GA, USA Loma Linda, CA, USA Edmonton, Alberta, Canada Cardiac Centre, University
Health Network,
Christopher J. White, MD, Nico A. Blom, MD, PhD, Daniel E. Forman, MD, Toronto, Ontario, Canada
Neurocardiology, Pediatric Cardiology, Geriatric Cardiology,
Ochsner Health System, Amsterdam University University of Pittsburgh, Nkechinyere N. Ijioma, MBBS,
New Orleans, LA, USA Medical Center, VA Pittsburgh Healthcare System, General Cardiology,
SECTION EDITORS Amsterdam, the Netherlands Pittsburgh, PA, USA Interventional Cardiology,
Mayo Clinic Health System,
M. Edip Gurol, MD, Erin A. Bohula, MD, DPhil,
Andrew M. Freeman, MD, La Crosse, WI, USA
Neurocardiology, Critical Care Cardiology,
Massachusetts General Hospital, Brigham and Women’s General Cardiology,
Nino Isakadze, MD, MHS,
Boston, MA, USA Hospital, Harvard Medical School, Preventative Care,
Health Technology and
Boston, MA, USA National Jewish Health,
Kathryn J. Lindley, MD, Implementation Science,
Denver, CO, USA
CardioObstetrics, Joan E. Briller, MD, Johns Hopkins School of Medicine,
Kevin G. Friedman, MD, Baltimore, MD, USA
Washington University School CardioObstetrics, Pediatric Cardiology,
of Medicine, University of Illinois at Chicago, Boston Children’s Hospital, Jacob C. Jentzer, MD,
St. Louis, MO, USA Chicago, IL, USA Harvard Medical School, Critical Care Cardiology,
Ariane J. Marelli, MD, MPH,
Craig S. Broberg, MD, MCR, Boston, MA, USA The Mayo Clinic,
Adult Congenital Heart Disease,
Adult Congenital Heart Disease, Parag Goyal, MD, MSc, Rochester, MN, USA
McGill University Health Center,
Oregon Health & Science Geriatric Cardiology,
Montreal, Quebec, Canada
University, Weill Cornell Medicine, Navin K. Kapur, MD,
Thomas S. Metkus, MD, Portland, OR, USA New York Presbyterian Hospital Critical Care Cardiology,
Critical Care Cardiology, New York, NY, USA The Cardiovascular Center,
Johns Hopkins University Jack Colman, MD,
Tufts Medical Center,
School of Medicine, Adult Congenital Heart Disease,
Sonia M. Grandi, PhD, Boston, MA, USA
Baltimore, MD, USA Toronto General Hospital/UHN,
Epidemiology,
Mount Sinai Hospital/SHS,
Justin T. Tretter, MD, Hospital for Sick Children, Jason N. Katz, MD, MHS,
University of Toronto,
Pediatric Cardiology, Toronto, Ontario, Canada Critical Care Cardiology,
Toronto, Ontario, Canada
Cleveland Clinic, Barry Greenberg, MD, Duke University,
Cleveland, OH, USA Jennifer Co-Vu, MD, Heart Failure, Durham, NC, USA
Rajesh Vedanthan, MD, Pediatric Cardiology, University of California
Global Health, University of Florida Medical Center, Paul Khairy, MD, PhD,
New York University Grossman College of Medicine, Sulpizio Cardiovascular Center, Electrophysiology,
School of Medicine, Gainesville, FL, USA La Jolla, CA, USA Montreal Heart Institute,
New York, NY, USA Matthias Greutmann, MD, Université de Montréal,
Abdulla A. Damluji, MD, PhD, MPH,
Mary N. Walsh, MD, Adult Congenital Heart Disease, Montreal, Québec, Canada
Geriatric Cardiology,
General Cardiology, University of Zurich,
Johns Hopkins School of Medicine,
St. Vincent Heart Center, Zurich, Switzerland Safi U. Khan, MD, MS,
Baltimore, MD, USA
Indianapolis, IN, USA Jasmine Grewal, MD, General Cardiology, Health Disparities,
Curt J. Daniels, MD, Adult Congenital Heart Disease, DeBakey Heart and Vascular Center,
EDITORIAL CONSULTANTS
Adult Congenital Heart Disease, St. Paul’s Hospital, Houston Methodist Hospital,
Jamil A. Aboulhosn, MD, Houston, TX, USA
The Ohio State University University of British Columbia,
Interventional Cardiology,
College of Medicine, Vancouver, British Columbia,
David Geffen School of Yuli Y. Kim, MD,
Columbus, OH, USA Canada
Medicine at UCLA, Adult Congenital Heart Disease,
Afshan Hameed, MD,
Los Angeles, CA, USA Melinda B. Davis, MD, Hospital of the University
CardioObstetrics,
Karen P. Alexander, MD, CardioObstetrics, of Pennsylvania,
University of California Irvine,
Geriatric Cardiology, University of Michigan, Philadelphia, PA, USA
Irvine, CA, USA
Duke Clinical Research Institute, Ann Arbor, MI, USA
Graeme J. Hankey, MD, James N. Kirkpatrick, MD,
Duke University Medical Center, John A. Dodson, MD, MPH, Neurocardiology, Geriatric Cardiology,
Durham, NC, USA Geriatric Cardiology, The University of Western Australia, University of Washington
Mohamad Alkhouli, MD, NYU Langone Health, Perth, Western Australia, Australia Medical Center,
Neurocardiology, New York, NY, USA Colleen M. Harrington, MD, Seattle, WA, USA
Mayo Clinic, Rochester, MN, USA
Stavros Drakos, MD, PhD, CardioObstetrics, Adrienne H. Kovacs, PhD,
Brett R. Anderson, MD, MBA, MS, University of Massachusetts
Critical Care Cardiology, Clinical Psychology,
Pediatric Cardiology, Medical School,
University of Utah Health & President, International Society for
NewYork-Presbyterian/Columbia Worcester, MA, USA
School of Medicine, Adult Congenital Heart Disease,
University Irving Medical Center,
Salt Lake City, UT, USA Toronto, Ontario, Canada
New York, NY, USA Steven M. Hollenberg, MD,
Craig J. Beavers, PharmD, Alexander C. Egbe, MD, MPH, Critical Care Cardiology, Ashok Krishnaswami, MD, MAS,
Pharmacology, Adult Congenital Heart Disease, Hackensack University Geriatric Cardiology,
University of Kentucky, Mayo Clinic, Medical Center, Kaiser Permanente Medical Center,
Lexington, KY, USA Rochester, MN, USA Hackensack, NJ, USA San Jose, CA, USA

Updated March 2023

EDITORIAL C ONSULTANTS CONTINUED

Min Ji Kwak, MD, MS, DrPH, P. Elliott Miller, MD, MHS, Bimal R. Shah, MD, MBA, Hector O. Ventura, MD,
Geriatric Cardiology, Critical Care Cardiology, Health Technology and Heart Failure,
University of Texas Yale University School of Medicine, Implementation Science, Ochsner Cardiovascular Institute,
Health Science Center at Houston, New Haven, CT, USA Duke University School of Medicine, New Orleans, LA, USA
Houston, TX, USA Durham, NC, USA
Ana Olga Mocumbi, MD, PhD, Dominique Vervoort, MD, MPH, MBA,
Adam B. Landman, MD, Chrisandra L. Shufelt, MD, Global Health,
Global Health,
Health Technology and Universidade Eduardo Mondlane & MS, NCMP, University of Toronto,
Implementation Science, Instituto Nacional de Saúde Women’s Health, Toronto, Ontario, Canada
Brigham and Women’s Hospital, Maputo, Mozambique Cedars-Sinai Medical Center,
Harvard Medical School, Los Angeles, CA, USA Salim S. Virani, MD, PhD,
Boston, MA, USA Philip Moons, PhD, RN, General Cardiology,
Clinical Psychology, Adnan Siddiqui, MD, PhD, Baylor College of Medicine,
Abbi Lane, PhD, Neurocardiology, Houston, TX, USA
KU Leuven,
Cardiovascular Epidemiology, SUNY University at Buffalo,
Leuven, Belgium Shadi Yaghi, MD,
Arnold School of Public Health, Buffalo, NY, USA
University of South Carolina, David A. Morrow, MD, MPH, Neurocardiology,
Columbia, SC, USA Critical Care Cardiology, Alexis Simpkins, MD, PhD, Brown University,
Brigham and Women’s Hospital, MSCR, Providence, RI, USA
Magalie Ladouceur, MD, PhD, Boston, MA, USA Neurocardiology,
Adult Congenital Heart Disease, Ali N. Zaidi, MD,
University of Florida,
Georges Pompidou Krishnakumar Nair, MBBS, Adult Congenital Heart Disease,
Gainesville, FL, USA
European Hospital, Icahn School of Medicine at Mount Sinai,
MD, DM,
Paris, France Shashank S. Sinha, MD, MSc, New York, NY, USA
Electrophysiology,
Critical Care Cardiology,
Zachary Laksman, MD, MSc, University of Toronto, Dominica Zentner, MBBS, PhD,
Inova Heart and Vascular Institute,
Electrophysiology, Toronto, Ontario, Canada CardioObstetrics,
Falls Church, VA, USA
University of British Columbia, Tien M.H. Ng, PharmD, Royal Melbourne Hospital,
Vancouver, British Columbia, Heart Failure, Karen Sliwa, MD, PhD, University of Melbourne,
Canada University of Southern California CardioObstetrics, Melbourne, Victoria, Australia
School of Pharmacy, University of Cape Town,
Sandra J. Lewis, MD, Cape Town, South Africa Liesl Zuhlke, MBChB, PhD,
General Cardiology, Cardiovascular Los Angeles, CA, USA
Pediatric Cardiology,
Disease in Women, Michael A. Solomon, MD, MBA,
Erwin Oechslin, MD, University of Cape Town,
Oregon Health and Critical Care Cardiology,
Adult Congenital Heart Disease, Cape Town, South Africa
Science University, National Institutes of Health
University of Toronto, Clinical Center,
Portland, OR, USA
Toronto, Ontario, Canada Bethesda, MD, USA
ETHICS COMMITTEE
Jennifer S. Li, MD, MHS, Alexander (Sasha)
Kenneth V. Snyder, MD, PhD, Chair: Richard L. Popp, MD,
Pediatric Cardiology, Opotowsky, MD, MMSc,
Neurocardiology, Stanford Health Care,
Duke University Medical Center, Adult Congenital Heart Disease,
University at Buffalo, Palo Alto, CA, USA
Durham, NC, USA Cincinnati Children’s Hospital,
Buffalo, NY, USA
Vaikom S. Mahadevan, MD, Cincinnati, OH, USA Holly Atkinson, MD,
Interventions, Nicole Beaton Sur, MD, Mount Sinai Health System,
Ki Park, MD,
University of California, Neurocardiology, New York, NY, USA
CardioObstetrics,
San Francisco, CA, USA University of Miami Miller
University of Florida
School of Medicine, Lawrence S. Cohen, MD,
Doff B. McElhinney, MD, College of Medicine,
Miami, FL, USA Yale University School of Medicine,
Pediatric Cardiology, Gainesville, FL, USA
New Haven, CT, USA
Stanford University Mehmet Akif Topcuoglu, MD,
School of Medicine, Gregory Piazza, MD, MS, Kim Fox, MD,
Neurocardiology,
Palo Alto, CA, USA General Cardiology, National Heart and Lung Institute,
Hacettepe University,
Vascular Medicine, Imperial College,
Alexandre Mebazaa, MD, PhD, Ankara, Turkey
Brigham and Women’s Hospital, Royal Brompton Hospital,
Critical Care Cardiology, Harvard Medical School, Harriette G.C. Van Spall, MD, MPH, London, UK
Université de Paris, Boston, MA, USA
Paris, France Health Technology and
Robert Frye, MD,
Ileana L. Piña, MD, MPH, Implementation Science,
Rowlens Melduni, MD, MPH, Mayo Clinic Rochester,
Heart Failure, McMaster University,
General Cardiology, Rochester, MN, USA
Central Michigan University, Hamilton, Ontario, Canada
Mayo Clinic,
Cleveland, OH, USA Philip J. Landrigan, MD,
Rochester, MN, USA
Poonam Velagapudi, MD, MS, Mount Sinai,
Thomas S. Metkus, MD, Nandita S. Scott, MD, General Cardiology, New York, NY, USA
Critical Care Cardiology, CardioObstetrics, Interventional Cardiology,
Johns Hopkins University Massachusetts General Hospital; University of Nebraska Eric N. Prystowsky, MD,
School of Medicine, Harvard Medical School, Medical Center, The Care Group, LLC,
Baltimore, MD, USA Boston, MA, USA Omaha, NE, USA Indianapolis, IN, USA

Updated March 2023

 2023-2024 OFFICERS
B. Hadley Wilson, MD, FACC, President, Atrium Health’s Sanger Heart & Vascular
The American College of
Institute, Charlotte, NC, USA
Cardiology’s (ACC) Board of Cathleen Biga, MSN, RN, FACC, Vice President, Cardiovascular Management of
Trustees (BOT) is the Illinois, Woodridge, IL, USA
centralized governing authority Edward T.A. Fry, MD, MACC, Immediate-Past President, Division of Cardiology,
for the ACC and responsible for overseeing
Ascension-IN Medical Group, Ascension St. Vincent Heart Center of Indiana,
decisions related to
Indianapolis, IN, USA
the College’s mission and strategic direction.
The president serves Christopher M. Kramer, MD, FACC, Treasurer, Chief, Cardiovascular Division,
as the head of the BOT. University of Virginia Health System, Charlottesville, VA, USA
Nicole L. Lohr, MD, PhD, FACC, Secretary and Board of Governors Chair, Zablocki VA
Medical Center, Milwaukee, WI, USA
Himabindu Vidula, MD, MS, FACC, Board of Governors Chair-Elect, Penn Medicine,
University of Pennsylvania, Philadelphia, PA, USA
Cathleen C. Gates, MA, American College of Cardiology (ACC) Chief Executive
Officer, Washington, DC, USA

Fred M. Kusumoto, MD, FACC, Chair, Department of Cardiovascular Disease, Mayo

2023-2024 SCIENTIFIC
PUBLICATIONS COMMITTEE (SPC) Clinic, Jacksonville, FL, USA
John A. Bittl, MD, FACC, Department of Interventional Cardiology, AdventHealth
The mission of the Scientific Publications Ocala, Ocala, FL, USA
Committee (SPC) of
Brian Alan Boe, MD, FACC, University Hospitals Cleveland Medical Center and
the American College of Cardiology (ACC) is
Nationwide Children’s Hospital, Columbus, OH, USA
to provide oversight
and direction for all scientific Marc P. Bonaca, MD, FACC, UCHealth University of Colorado Hospital, Aurora, CO, USA
peer-reviewed journals of the Mark A. Creager, MD, FACC, Heart and Vascular Center, Dartmouth-Hitchcock
ACC and to advance the mission Medical Center, Lebanon, NH, USA
of the ACC to transform
Mary M. McDermott, MD, Northwestern Memorial Hospital, Chicago, IL, USA
cardiovascular care and
Glaucia Maria Moraes De Oliveira, MD, MSc, PhD, FACC, Federal University of Rio de
improve heart health.
Janeiro (UFRJ), Rio de Janeiro, Brazil
Andriana Nikolova, MD, Cedars-Sinai Medical Center, Los Angeles, CA, USA
Sandra M. Oliver-McNeil, DNP, ACNP-BC, AACC, Beaumont Hospital, Royal Oak, MI, USA
Kian Keong Poh, MB BCh, FACC, Department of Cardiology, National University Heart
Center, Singapore

CORRESPONDENCE FOR Karin R. Sipido, MD, PhD, Department of Cardiovascular Sciences, Division of
AMERICAN COLLEGE OF Experimental Cardiology, KU Leuven, Leuven, Belgium
CARDIOLOGY
Gilbert H. Tang, MD, MSc, MBA, FACC, Mount Sinai Health System, New York, NY, USA
All correspondence for the College, Brittany Weber, MD, Department of Medicine, Brigham Women’s Hospital, Harvard
other than that related to Medical, Boston, MA, USA
JACC: Advances should be
Janice Sibley, MS, MA, ACC Executive Vice President, Education and Publishing
sent to Resource Center,
Divisions, Washington, DC, USA
American College of Cardiology,
2400 N Street, Justine Varieur Turco, MA, ACC Divisional Vice President, Publishing, Washington,
NW, Washington, DC 20037 DC, USA

Updated March 2023

JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ORIGINAL RESEARCH

GERIATRIC CARDIOLOGY

Hierarchical Development of Physical

Frailty and Cognitive Impairment and
Their Association With Incident
Cardiovascular Disease
Abdulla A. Damluji, MD, PHD,a,b,c Naila Ijaz, MD,a Shang-En Chung, SCM,d Qian-Li Xue, PHD,b,d
Rani K. Hasan, MD, MHS,c Wayne B. Batchelor, MD, MHS,a Ariela R. Orkaby, MD,f Ajar Kochar, MD,e
Michael G. Nanna, MD, MHS,f David L. Roth, PHD,b Jeremy D. Walston, MD,d Jon R. Resar, MD,c
Gary Gerstenblith, MDc

ABSTRACT

BACKGROUND Frailty and cognitive impairment (CI) are geriatric conditions that lead to poor health outcomes among
older adults with cardiovascular disease. The association between their temporal patterns of development and
cardiovascular risk is unknown.

OBJECTIVES This study aims to examine the 5-year cardiovascular outcomes by the pattern of development of frailty
and CI in older adults without a history of coronary artery disease.

METHODS We used the National Health and Aging Trends Study, linked to Medicare data. Frailty was measured using
the physical frailty phenotype. CI was measured using the AD8 Dementia Screening Interview, measured cognitive
performance, or self-report by patient or caregiver for a diagnosis given by a physician. The primary outcome was incident
major adverse cardiovascular event at 5 years.

RESULTS Of a total 2,189 study participants aged 65 and older, 38.5% were male. In this study population, 154 (7%)
participants developed frailty first, 829 (38%) developed CI first, and 195 (9%) participants developed both simulta-
neously (frail-CI group). Those who developed frailty and CI simultaneously were older, more likely to be female, and had
multiple chronic conditions. The frail-CI group had the highest risk of major adverse cardiovascular event (hazard ratio
[HR]: 1.81; 95% CI: 1.47–2.23) followed by frail first (HR: 1.46; 95% CI: 1.17–1.81) and CI first (HR: 1.31; 95% CI: 1.15–1.50).
Frailty first was associated with the greater risk of stroke (HR: 1.49; 95% CI: 1.06–2.09) compared to the intact group.

CONCLUSIONS The simultaneous development of frailty and CI is associated with an increased risk of adverse
cardiovascular outcomes including death compared with the development of each syndrome alone. Diagnostics to
detect frailty and CI are critical in assessment of cardiovascular risk in the older population. (JACC Adv 2023;2:100318)
© 2023 The Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an open
access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

From aThe Inova Center of Outcomes Research, Inova Heart and Vascular Institute, Falls Church, Virginia, USA; bJohns Hopkins
Older Americans Independence Center and the Center on Aging and Health; cDivision of Cardiology, Department of Medicine,
Johns Hopkins University School of Medicine, Baltimore, Maryland, USA; dDivision of Geriatrics and Gerontology, Department of
Medicine, Johns Hopkins School of Medicine, Baltimore, Maryland, USA; eThe Brigham and Women’s Hospital, Harvard Medical
School, Boston, Massachusetts, USA; and the fSection of Cardiovascular Medicine, Yale University School of Medicine, New Haven,
Connecticut, USA.

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100318

2 Damluji et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Frailty and Cognitive Impairment in Older Adults With CVD MAY 2023:100318

I
ABBREVIATIONS n the developed world, the older adult 5 years of follow-up beyond visit 2 in the NHATS
AND ACRONYMS population is expanding at a rapid linked to Centers for Medicare & Medicaid Services
pace. According to the U.S. Census Bu- data.
ADL = activities of daily living
reau, 16.5% of the United States population
AMI = acute myocardial
are older than 65 years of age,1 and that METHODS
infarction
figure is projected to grow to 20.3% by
CAD = coronary artery disease
2030.2,3 Several geriatric conditions preva- THE SOURCE AND STUDY POPULATIONS. We
CI = cognitive impairment
lent among older adult patients can impact analyzed data from the 2011 NHATS baseline cohort.7
CVD = cardiovascular disease
the assessment of cardiovascular risk. 4 Phys- Funded by the National Institute on Aging
IADL = instrumental activities
ical frailty is defined as the loss of physio- (U01AG032947), NHATS is a prospective cohort study
of daily living
logic reserve, which leads to decreased that focuses on functioning of older patients. The
MACE = major adverse
cardiovascular event
resistance to stressors, increased vulnera- source population is derived from a probabilistic
bility, and a progression to poor overall sample of Medicare beneficiaries aged 65 years and
PVD = peripheral vascular
disease health outcomes. Cognitive impairment older who were interviewed in 2011 during their
(CI), a progressive neurocognitive disorder, baseline visit (visit 1) and annually reinterviewed to
is commonly experienced in old age and is character- document the changes that occurred in their func-
ized as declines in brain functions such as memory, tional status as they grow older. Geriatric risks such
language, and problem-solving. The spectrum of as frailty, physical and cognitive functioning, ability
both frailty and CI can greatly impact the quality of to carry out activities of daily living (ADL), and
life and management of cardiovascular disease environmental characteristics of each participant
(CVD) in older patients. In this study, the spectrum were collected. African Americans and very old par-
of CI spans from age-related mild cognitive impair- ticipants were oversampled from the Medicare
ment to major neurocognitive disorder including de- enrollment file. The data for each participant in the
mentia were studied. From a clinical perspective, NHATS repository are linked to their Medicare data
dementia can have the most significant impact on that were available to the investigators for analysis.
clinical outcomes and management. The term “cogni- The study population included participants
tive frailty” has been used to define the coexistence aged $65 years at enrollment who had no history of
of frailty and CI and the addition of CI to the frailty CAD prior to their 2011 NHATS baseline visit. Partici-
criteria increases the predictive validity of the defini- pants were assessed for frailty and CI at each subse-
tion for adverse health outcomes in geriatric popula- quent visit and categorized into the following groups:
tions. 5 While these 2 syndromes can coexist in a 1) intact: those who did not develop either frailty or CI
minority of cases, the development of one condition during the first 2 years of follow-up, ie, visit 1 or visit
can increase the risk of the development of the 2; 2) frail first: those who had physical frailty and not
other.6 CI at visit 1 or those who were not found to have
Previous research from the NHATS (National frailty and CI at visit 1 but had developed frailty at
Health and Aging Trends Study) has shown that the visit 2; 3) CI first: those who developed CI and had no
incidences of death and major adverse cardiovascular frailty at visit 1 or those who were not found to have
events (MACEs) are increased in individuals classified frailty and CI at visit 1 but had developed CI at visit 2;
as frail. 7 However, the incidences of CI, frailty, and and 4) frail-CI: those who were found to have both
their patterns of development over time among older frailty and CI on the same visit (either visit 1 or visit
adults at risk for coronary artery disease (CAD) are not 2). Prior work addressed the temporal association
yet understood. In this study, we aimed to: 1) eval- between physical frailty and incident CVD in the
uate the incidence of CI and frailty in an older adult NHATS study.8 In this study, the same study subjects
population without known CAD; 2) examine the were used but measures of cognitive function were
incidence and patterns of the temporal development added to provide an additional layer of insight in the
of each geriatric syndrome during visit 1 and visit 2 of complex interplay between cognitive function,
the study; and 3) assess their influence on MACE at physical function, and incident CVD.9

The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received September 8, 2022; revised manuscript received February 3, 2023, accepted February 24, 2023.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Damluji et al 3
MAY 2023:100318 Frailty and Cognitive Impairment in Older Adults With CVD

DEFINITIONS OF PHYSICAL FRAILTY AND COGNITIVE mobility disability. The Katz scale was used to assess
IMPAIRMENT. Physical frailty was defined using the for disability in: 1) self-care (ADL: bathing, dressing,
framework provided by Fried et al3 as a clinical syn- eating, toileting); 2) household activities (IADL: doing
drome characterized by increased vulnerability to laundry, preparing meals, shopping for groceries and
stressors due to decline in reserve and function across for personal items, medication management,
multiple physiologic systems that occurs with age, handling bills and banking); and 3) mobility (inside
leading to a compromised ability to manage everyday the home, going outside, getting out of bed).11
acute stress. Frailty in each patient in the NHATS Dementia status was determined using: 1) physician
study was assessed using the 5 domains of the Fried reports with a diagnosis of dementia or Alzheimer’s
physical frailty phenotype: exhaustion, low physical disease; 2) a scoring administered to proxies indi-
activity, weakness, slowness, and shrinking.10 If 3 or cating the participant has probable dementia; and
more of these 5 criteria were present based on pre- 3) results from cognitive testing using the AD8 De-
viously published criterion,11 the individual was mentia Screening Interview that evaluates memory,
categorized as frail and those with 1 or 2 of the 5 were orientation, and executive function. 13 Disability was
categorized as ‘prefrail.’ Those without the presence measured using the American Community Survey
of any criteria were categorized as ‘robust.’ If the grip Disability Questions. Loss of independence was
or walking test was not done because of health/safety defined as participants reporting never or rarely ever
concerns, a value of 0 was assigned to indicate going outside or having to use devices to go outside. 12
worst performance. DEMOGRAPHIC CHARACTERISTICS, MEDICAL CONDITIONS,
CI refers to a state of cognitive vulnerability often AND HEALTH CARE UTILIZATION. Each participant in the
associated with vascular risk factors and with a higher study was asked whether their physician had ever
risk of developing overt dementia. 3 Similar to Chu diagnosed them with the following medical condi-
et al,9 CI was assessed using measures of cognitive tions: high blood pressure, diabetes mellitus, stroke,
performance testing or by proxy reports.9 Dementia any cardiac disorder, arthritis, lung disease, and CI or
was defined as a time independent variable, and was dementia. Hospitalization information for the past
coded 1 when it was observed during visit 1 or 2, and 12 months was also obtained. 8,12
0 otherwise.9 CI was determined by using 3 sources of
STATISTICAL ANALYSIS. In a nationally representa-
information. Participants were classified as having CI
tive sample, NHATS participants with a history of
if they had at least one of the following: 1) impair-
CAD were excluded at baseline. Similar to work by
ment in either 1 of 2 cognitive domains: executive
Chu et al,9 participants were categorized according to
function or memory; 2) if participant or proxy re-
the hierarchical development of frailty and CI during
ported a diagnosis of CI or dementia given to them by
the first 2 years of the study, as follows: 1) frailty
their physician; or 3) a score $2 on the AD8 instru-
onset 1 year or longer before CI; 2) CI onset 1 year or
ment. The AD8 Dementia Screening Interview is an 8-
longer before frailty; 3) CI-frailty co-occurring within
item battery of queries regarding an individual’s
the same year; and 4) neither CI nor frailty occurred
cognitive status by assessing memory, temporal
9 by visit 2.
orientation, judgement, and functioning. The main
Demographics, smoking status, comorbidity, hos-
exposure in this study is the order of onset of physical
pitalizations, emergency department visits, falls, self-
frailty, CI, namely physical frailty first, CI first or co-
care, mobility, household activities, depression,
occurrence within 1 year of each other in Medicare
9 anxiety, and CI at visit 2 were reported for the pat-
participants over 65 years of age.
terns of onset, namely: no frailty or CI, frailty onset
CARDIOVASCULAR OUTCOMES. The primary outcome
before CI, CI onset before frailty, or CI-frailty co-
of this study is the time to first incident MACE, occurrence. Percentages were calculated for categor-
defined as a composite of death from any cause, acute ical variables and mean
SD for continuous
myocardial infarction (AMI), any subsequent coro- variables. Data on self-care, mobility, and household
nary heart disease, stroke, or peripheral vascular activities are presented as cumulative proportions at
disease, whichever came first. 8,12 The length of 5 years for the 4 groups. Likelihood ratio chi-square
follow-up for MACE was between visit 2 and visit 8 and Kruskal-Wallis tests were used to assess the dif-
(ie, 5 years of follow-up). ference in sociodemographic health factors and inci-
GERIATRIC CONDITIONS. Each participant’s geri- dence of cardiovascular outcomes among the CI/
atric risks were assessed during follow-up visits frailty groups. Proportional hazard models were used
which included level of functioning, ADL, instru- to assess the unadjusted association between the
mental activities of daily living (IADL), disability, and patterns of onset of frailty and CI on cardiovascular
4 Damluji et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Frailty and Cognitive Impairment in Older Adults With CVD MAY 2023:100318

outcomes among older adults at 5-year follow-up. Overall, those who had co-occurrence of frailty and
Patients were censored if they developed the cardio- CI simultaneously were older, more female, and
vascular outcomes of interest or if they were lost to belong to ethnic minority groups (Table 1). They had
follow-up. To address confounding by age, de- more help in self-care, mobility, household activities,
mographics, and other risk factors, we performed and have a higher rate of disability. The presence and
additional multivariable Cox models. Model 2 extent of disability were lower in the CI first group, as
adjusted for age, gender, sex, race/ethnicity, census compared to the frailty-first group and CI-frailty
division, residence, income, body mass index, groups. In addition, the proportion of those able to
smoking status, diabetes, hypertension, number of perform household activities and ADL or IADL
comorbid diseases, and dependency status (as a sur- without impairment were higher in the CI first group
rogate measure for composite functional status). as compared to the frailty-first and CI-frailty groups.
Those with prior stroke were excluded from the The incidence of cardiovascular outcomes over the
model because the rate of CI can be impacted by prior 5-year follow-up period is presented in Table 2.
cerebrovascular events. The assumption for Cox Overall, 25% of the participants died by 5-year follow-
proportional hazard models was checked by plotting up and 59% suffered from a MACE, with CAD being
the Schoenfeld residuals against survival time for the most common (30%). The frail-CI group had the
each primary and secondary cardiovascular outcome highest incidence of all-cause mortality (60%) and
by each of the 4 groups. We then applied a competing developed more MACEs (87%) (Figure 1). The frail-CI
risk analysis to account for the potential impact of group also had the highest incidence of peripheral
death on each component of MACE. We used the Fine vascular disease (PVD) (40%) and CAD (40%); how-
and Gray proportional subdistribution hazard model, ever, the frail first group had the highest incidence of
which is widely utilized in practice to analyze AMI (11%) and stroke (29%) (Table 2). In unadjusted
competing risk data.14 The model allows for the esti- Cox proportional hazards models, all 3 patterns of
mation of subdistribution of hazard while accounting geriatric syndrome development, as compared with
for competing risk of death. 14 All tests are 2-sided, the intact group, were associated with incidence of
and the statistically significant level is set at P < 0.05. death, MACEs and each individual component of
Data analyses were conducted using SAS (version 9.4, MACEs: AMI, stroke, PVD, and CAD (Table 3). Frail-CI
SAS Institute Inc). The Johns Hopkins Medicine had a greater risk of MACEs, death, acute MI, PVD,
Institutional Review Board approved this study. and CAD as compared to frailty first and CI first.
After adjusting for age, gender, race/ethnicity,
RESULTS census division, residence and income, body mass
index, smoking status, diabetes, hypertension, num-
In the NHATS study, 2,189 community-dwelling par- ber of chronic diseases, and dependency status, the
ticipants had no history of CAD at baseline and hier- development of either frailty, CI or both simulta-
archical data on frailty and CI during visit 1 and 2 of neously remained associated with a greater risk of
follow-up. Of those, a total of 1,011 (46%) remained MACE and death, as compared with the intact group,
intact without frailty or CI; 154 (7%) had frailty-first; over the 5-year follow-up period (Table 3). When
829 (38%) had CI-first; and 195 (9%) had frailty-CI co- compared to the intact group, both geriatric syn-
occurrence. The baseline demographic data and dromes and, regardless of the order of development,
functional characteristics based on these patterns of were associated with a greater risk of PVD and CAD.
hierarchical development of these geriatric syn- Frail-CI was associated with a greater risk of AMI
dromes are presented in Table 1. The mean age was (hazard ratio [HR]: 1.93; 95% CI: 1.07-3.47), whereas
77.5 years and 61% of the cohort were $75 years of age. development of frailty first was associated with a
Female participants constituted 62% of the cohort and greater risk of stroke (HR: 1.49; 95% CI: 1.06-2.09),
the majority enrolled were Whites or Caucasians. On and CI first and frail-CI were no longer statistically
average, the majority were overweight, and 48% of associated with a risk of stroke when compared to
the cohort smoked at least 1 cigarette per day. More the intact group. Competing risk analysis is pre-
than one-half of the study population had at least 2 sented in Table 4. We found that after adjusting for
coexisting chronic medical conditions, and 11% of the potential confounders, the subdistribution HRs for
cohort had 4 or more chronic comorbidities. The most frailty-first was associated with development of AMI
prevalent medical conditions were hypertension, during follow-up after accounting for the competing
arthritis, and osteoporosis. Approximately 20% of the risk of death. The association between the CI-frailty
study population was living with diabetes mellitus group with MACEs was mostly explained by all-
and 9% had history of stroke at baseline. cause mortality during follow-up.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Damluji et al 5
MAY 2023:100318 Frailty and Cognitive Impairment in Older Adults With CVD

T A B L E 1 Characteristics of the Study Population of Participants Without a History of Coronary Heart Disease Enrolled in the NHATS by
Patterns of Frailty and CI

Total Intact Fraila First CI First Frail-CI

(N ¼ 2,189) (n ¼ 1,011) (n ¼ 154) (n ¼ 829) (n ¼ 195) P Value

Age, mean 77.5 74.6 78.3 79.3 83.9 <0.001

Age, y <0.001
65-69 18.6 26.8 15.6 12.3 4.6
70-74 21.0 27.3 18.8 17.1 6.2
75-79 20.8 21.3 18.2 21.5 17.4
80-84 20.2 16.9 26.0 23.4 19.0
85-89 12.0 6.0 14.9 14.2 30.8
90þ 7.5 1.7 6.5 11.5 22.1
Sex 0.009
Female 61.5 59.4 68.2 60.9 70.3
Male 38.5 40.7 31.8 39.1 29.7
Race <0.001
Non-Hispanic White 74.5 82.0 73.9 68.4 62.4
Non-Hispanic Black 19.2 14.6 18.3 22.3 30.4
Hispanic 4.1 2.1 5.2 6.1 5.2
Other 2.2 1.4 2.8 3.3 2.1
BMI, mean 27.1 27.7 27.6 26.4 26.5 <0.001
Smoking status 0.011
Smoke at least 1 cigarette/d 47.8 49.2 52.0 48.0 36.9
Self-reported disease
Arthritis 53.5 48.4 76.0 49.6 79.0 <0.001
Diabetes mellitus 20.0 17.2 27.3 19.9 28.7 0.0004
Hypertension 62.3 60.0 72.1 61.1 71.8 0.0006
Lung disease 12.5 11.9 22.1 11.0 14.4 0.0035
Osteoporosis 20.7 18.2 29.2 19.8 30.9 <0.001
Stroke 8.9 4.7 11.7 9.8 24.6 <0.001
Number of chronic diseases <0.001
0-1 39.4 47.3 14.9 40.4 13.9
2-3 49.5 46.4 66.2 49.7 51.3
4þ 11.1 6.3 18.8 9.9 34.9
Hospital stay past 12 mo 15.4 10.1 25.5 16.7 29.9 <0.001
Any fall past month 29.9 23.9 43.5 29.5 51.3 <0.001
Self-care disability <0.001
No difficulty 76.7 89.0 58.4 76.1 29.7
Difficulty but no help 11.2 7.6 23.4 11.7 18.5
Help 12.1 3.4 18.2 12.2 51.8
Mobility disability <0.001
No difficulty 70.5 83.9 44.2 70.8 20.5
Difficulty but no help 16.5 13.1 33.1 16.9 20.0
Help 13.0 3.1 22.7 12.3 59.5
Household activities disability <0.001
No difficulty 64.7 81.6 36.4 61.5 12.8
Difficulty but no help 12.5 12.5 26.0 10.7 9.7
Help 22.8 5.9 37.7 27.7 77.4
Overall disability level <0.001
No difficulty 53.8 69.9 29.2 49.8 6.7
Difficulty but no help 20.0 21.6 27.3 18.9 10.3
Help 26.2 8.5 43.5 31.2 83.1
Depression
PHQ2 score $3 12.8 7.1 19.6 13.4 34.4 <0.001
Anxiety
GAD2 score $3 10.5 5.6 17.0 11.4 26.6 <0.001
Continued on the next page
6 Damluji et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Frailty and Cognitive Impairment in Older Adults With CVD MAY 2023:100318

T A B L E 1 Continued

Total Intact Fraila First CI First Frail-CI

(N ¼ 2,189) (n ¼ 1,011) (n ¼ 154) (n ¼ 829) (n ¼ 195) P Value

No. of ED visits <0.001

0 77.3 82.5 68.8 76.0 62.1
1 15.8 13.5 23.4 16.8 17.4
$2 7.0 4.1 7.8 7.2 20.5
No. of hospitalizations <0.001
0 89.0 92.4 83.8 89.6 73.3
1 8.8 6.4 12.3 8.3 20.5
$2 2.2 1.2 3.9 2.1 6.2
Total LOS in hospital, mean 0.93 0.47 1.69 0.92 2.77 <0.001
Number of physician visits, mean 6.99 6.42 9.14 7.00 8.22 <0.001
Number of ADL impairment <0.001
0 66.3 81.0 43.5 64.5 15.4
1-2 20.7 15.4 28.6 24.1 26.7
$3 13.1 3.6 27.9 11.3 58.0
Number of IADL impairments <0.001
0 65.9 81.9 38.3 63.0 16.9
1-2 20.9 15.8 37.7 23.3 23.6
$3 13.3 2.3 24.0 13.8 59.5

Values are % unless otherwise indicated. aFrailty was assessed by the physical frailty phenotype paradigm that is grounded in 5 criteria: exhaustion, low physical activity,
weakness, slowness, and shrinking (www.nhats.org). Claims-based frailty index is a validated frailty tool with the physical frailty phenotype that utilizes claims data.
ADL ¼ activities of daily living; BMI ¼ body mass index; CI ¼ cognitive impairment; ED ¼ emergency department; GAD2 ¼ generalized anxiety disorder 2-item;
IADL ¼ instrumental activities of daily living; LOS ¼ length of stay; NHATS ¼ National Health and Aging Trends Study; PHQ2 ¼ patient health questionnaire-2.

DISCUSSION 2) participants who developed both CI and frailty

simultaneously were older and had a high preva-
We examined the patterns of development of frailty lence of multiple chronic conditions and baseline
and CI and their associations with cardiovascular disability as compared with the other groups; and 3)
outcomes among older adults in NHATS without prior as compared with participants who developed
known CAD during a 5-year follow-up. The major frailty first or CI first, participants who developed
findings are as follows: 1) The highest proportion of simultaneous CI and frailty had a higher risk of
participants developed cognitive frailty first (37.8%), developing MACEs, mostly attributed to mortality at
followed by those who developed a combination of 5-year follow-up, even after adjusting for de-
physical and cognitive frailty (8.9%), and then those mographic characteristics, traditional cardiovascular
who developed only physical frailty (7.0%); risk factors, and multimorbidity at baseline (Central
Illustration).
In a previous study, we found that frailty in those
without prior CAD was associated with MACE and
T A B L E 2 The Incidence of MACEs by Patterns of Frailty and CI Among
Older Adults Without a History of Coronary Heart Disease in the NHATS mortality. 8 In addition, frail patients more frequently
During 5-year Follow-Up suffered from CI at baseline (26.8%) as compared with

Total Intact Fraila First CI First Frail-CI

non-frail (2.6%), and 40% of those who were frail had
(N ¼ 2,189) (n ¼ 1,011) (n ¼ 154) (n ¼ 829) (n ¼ 195) P Value probable dementia compared with just 3.5% of non-
MACE 59.1 46.1 71.4 66.2 86.7 <0.001 frail.8 Previous research has shown that with the
Death 24.7 9.4 30.5 34.1 59.5 <0.001 great degree of frailty, there is an increased for
AMI 7.6 5.6 11.0 8.6 10.8 0.008
development of CI.15 This suggested that the under-
Stroke 19.7 17.5 29.2 19.9 22.6 0.007
lying mechanisms of CI and frailty are interrelated,
PVD 23.9 16.5 33.8 27.5 39.5 <0.001
and these may also increase the risk for cardiovas-
Any CAD 29.7 24.3 37.7 32.5 40.0 <0.001
cular events.
Values are % unless otherwise indicated. aFrailty was assessed by the physical frailty phenotype This study confirmed that CI, both with frailty and
paradigm that is grounded in 5 criteria: exhaustion, low physical activity, weakness, slowness, and
shrinking (www.nhats.org).
independent of frailty, was associated with a higher
AMI ¼ acute myocardial infarction; CAD ¼ coronary artery disease; CI ¼ cognitive impairment; incidence of adverse cardiac events and mortality
MACE ¼ major adverse cardiovascular event; NHATS ¼ National Health and Aging Trends Study;
PVD ¼ peripheral vascular disease.
than those without CI. Outcomes were worse when CI
and frailty developed simultaneously. These findings
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Damluji et al 7
MAY 2023:100318 Frailty and Cognitive Impairment in Older Adults With CVD

F I G U R E 1 Kaplan-Meier Survival Curve of MACE at 5 Years by the Hierarchical Development of Frailty and Cognitive Impairment

1.00

0.75
MACE−free Probability

0.50

0.25
Intact
Frail−First
CI−First
Frail−CI
0.00
0 1 2 3 4 5
Time (Years)

Kaplan-Meier survival curve illustrating major adverse cardiovascular event–free survival over a 5-year follow-up by pattern of frailty and
cognitive impairment development in the National Health and Aging Trends Study participants without a history of coronary artery disease
(log-rank P < 0.001). Major adverse cardiovascular event was defined as acute myocardial infarction, stroke, peripheral vascular disease and
coronary artery disease, and excluded mortality. CI ¼ cognitive impairment; MACE ¼ major adverse cardiovascular event.

may be explained by CI causing inability to adhere to domains of cognitive function can result in inability
guideline directed medical therapy, difficulty man- to follow care plans and follow-up with clinical rec-
aging cardiovascular risk factors, poor dietary habits, ommendations, resulting in an increased likelihood
and overall worse physical health. Memory, atten- for adverse cardiovascular outcomes and mortality.
tion, ability to learn, and follow executive func- The combination of frailty and CI also exacerbates
tioning are all necessary to adhere to medical other clinical conditions because of impairment in
recommendations, care for oneself, and maintain health literacy and ability to follow recommendations
good health.16 Impairment in one or more of these resulting in worse health outcomes.16

T A B L E 3 Proportional Hazards Regression Model Evaluating the Influence of Patterns of Frailty and CI on 5-Year Cardiovascular Outcomes
Among Older Adults Without a History of Coronary Heart Disease in the NHATS

MACE Death AMI Stroke PVD CAD

Model 1a
Frailty firstc 1.99 (1.62–2.46) 3.65 (2.58–5.18) 2.27 (1.32–3.89) 1.91 (1.38–2.65) 2.38 (1.75–3.26) 1.88 (1.41–2.51)
CI first 1.75 (1.55–1.98) 4.24 (3.36–5.35) 1.80 (1.27–2.55) 1.25 (1.01–1.55) 1.98 (1.62–2.42) 1.58 (1.33–1.88)
Frail-CI 3.34 (2.80–3.99) 9.56 (7.27–12.5) 2.99 (1.81–4.94) 1.77 (1.27–2.46) 3.68 (2.80–4.82) 2.56 (1.98–3.30)
Model 2b
Frail firstc 1.46 (1.17–1.81) 2.62 (1.83–3.76) 1.73 (0.98–3.05) 1.49 (1.06–2.09) 1.74 (1.26–2.41) 1.43 (1.06–1.92)
CI first 1.31 (1.15–1.50) 2.65 (2.06–3.40) 1.43 (0.98–2.09) 1.01 (0.80–1.27) 1.52 (1.23–1.89) 1.35 (1.12–1.63)
Frail-CI 1.81 (1.47–2.23) 4.45 (3.23–6.13) 1.93 (1.07–3.47) 1.05 (0.71–1.54) 1.97 (1.43–2.70) 1.68 (1.25–2.26)

Values are HR (95% CI). aModel 1 is unadjusted model. bModel 2 was adjusted for age, gender, race/ethnicity, census division, residence and income, body mass index, smoking
status, diabetes, hypertension, number of chronic diseases, and dependency status. cFrailty was assessed by the physical frailty phenotype paradigm that is grounded in 5
criteria: exhaustion, low physical activity, weakness, slowness, and shrinking (www.nhats.org).
AMI ¼ acute myocardial infarction; CAD ¼ coronary artery disease; CI ¼ cognitive impairment; MACE ¼ major adverse cardiovascular event; NHATS ¼ National Health and
Aging Trends Study; PVD ¼ peripheral vascular disease.
8 Damluji et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Frailty and Cognitive Impairment in Older Adults With CVD MAY 2023:100318

T A B L E 4 Fine and Gray Competing Risk Analysis for Death on the Association Between Frailty First, CI First, and Both Frailty and CI on
Cardiovascular Outcomes in the NHATS

MACE AMI Stroke PVD CAD

Model 1a
Frailty firstc 1.17 (0.90–1.54) 2.28 (1.22–4.26) 1.29 (0.87–1.92) 1.58 (1.09–2.28) 1.21 (0.85–1.73)
CI first 0.86 (0.73–1.01) 1.06 (0.67–1.70) 0.73 (0.56–0.94) 1.17 (0.92–1.47) 0.93 (0.75–1.14)
Frail-CI 0.71 (0.53–0.95) 1.24 (0.60–2.56) 0.57 (0.34–0.95) 1.03 (0.69–1.54) 0.86 (0.60–1.24)
Model 2b
Frail firstc 0.99 (0.75–1.31) 2.24 (1.18–4.24) 1.12 (0.74–1.71) 1.29 (0.88–1.89) 1.02 (0.71–1.48)
CI first 0.83 (0.70–1.00) 1.14 (0.66–1.96) 0.69 (0.51–0.93) 1.06 (0.82–1.38) 0.96 (0.76–1.22)
Frail-CI 0.57 (0.40–0.80) 1.40 (0.57–3.44) 0.42 (0.23–0.75) 0.74 (0.46–1.19) 0.78 (0.50–1.20)

Values are HR (95% CI). aModel 1 is unadjusted model. bModel 2 was adjusted for age, gender, race/ethnicity, census division, residence and income, body mass index, smoking
status, diabetes, hypertension, number of chronic diseases, and dependency status. cFrailty was assessed by the physical frailty phenotype paradigm that is grounded in 5
criteria: exhaustion, low physical activity, weakness, slowness, and shrinking (www.nhats.org).
AMI ¼ acute myocardial infarction; CAD ¼ coronary artery disease; CI ¼ cognitive impairment; MACE ¼ major adverse cardiovascular event; NHATS ¼ National Health and
Aging Trends Study; PVD ¼ peripheral vascular disease.

CI is associated with worsened nutritional status, decline in cognitive functioning, memory, or execu-
functional status, and mortality. 17 Additionally, dur- tive function. When combined with physical frailty,
ing hospitalizations, CI is associated with increased signs like muscle weakness, fatigue, slowness, and
risk of falls, longer length of stay, frequent discharge gait impairment were associated with greater risk.
to nursing facilities, and higher inhospital mortality, Physical frailty is a well-established risk factor for
with worse outcomes in those with severe CI.18 CVD, and research suggests that cognitive frailty is
Cognitive decline increases the risk of nonadherence also associated with an increased risk of cardiovas-
to recommended lifestyle changes or medications to cular events.8,12 This is thought to be due to the
control cardiovascular risk factors, which increases complex interplay between cognitive and physical
cardiovascular risk. Cognitive decline can therefore functioning with cardiovascular health.
lead to poor physical health because of difficulty with This study has some limitations. First, MACE was
self-care, reduced physical activity due to muscle diagnosed using Medicare claims data from hospital
weakness fatigue or frailty, difficulty with mobility, and outpatient encounters. Although this method is
and reduced ability to communicate with health care widely used when studying cardiovascular outcomes,
workers and caregivers. All these factors can the severity of each outcome cannot be ascertained
contribute to poor overall physical health outcomes using this method. Second, frailty and CI are dynamic
which can be manifested as falls, infections, and processes which may reverse or change over time. In
other health problems observed in older adults. this study, we assessed both geriatric risks during the
This study represents the most comprehensive first and second clinic visits in NHATS. While this is
study to date documenting the association of hierar- important to acknowledge, the study results remain
chical development of CI and frailty with cardiovas- novel because the hierarchical association between
cular outcomes including AMI, CAD, PVD, and stroke, frailty and CI in the context of CVD has not been
in those without CAD at baseline. Integration of studied to this extent. Third, when examining the CI-
screening tools for geriatrics risks in the cardiovas- first vs frailty-first categories, the results might be
cular practice will result in the identification of pa- influenced by different thresholds of dichotomous
tients who may benefit from interventions to address categorizations of continuous variables.20 This can
geriatric risks and modify the approach for cardio- result in misclassification bias, which is sensitive to
vascular therapies. Multifaceted interventions to different thresholds of dichotomization. 20 While we
address physical function, nutritional status, and prefer to evaluate variables as continuous, the cate-
cognitive health are under investigation and may ul- gorization is often necessary for clinical decision
timately impact cardiovascular health. 19 making in practice when trying to decide whether an
Cognitive frailty, physical frailty, and CVD are older adult is suffering from CI or frailty. Additionally,
closely related. Cognitive frailty refers to the state of the instrument used to measure CI, the AD8 dementia
vulnerability in older adults characterized by the screening interview, is subject to variability in
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Damluji et al 9
MAY 2023:100318 Frailty and Cognitive Impairment in Older Adults With CVD

C ENTR AL I LL UST RA TI O N This Figure Illustrates the Findings of This Study in Which 46% of the Participants
Remained Intact Without Developing Either Cognitive Impairment or Frailty

Damluji AA, et al. JACC Adv. 2023;2(3):100318.

These participants had the lowest rate of major adverse cardiovascular events as illustrated in the bar graph to the right. A small percentage of participants (7%)
developed frailty first, and this group had a significantly higher incidence of major adverse cardiovascular events than those who were intact. A large percentage of
participants (38%) developed cognitive impairment first and the incidence of major adverse cardiovascular events as depicted on the bar graph to the right was greater
than it was for those who remained intact, although less than it was for those who developed frailty first. Lastly, the group that developed cognitive impairment and
frailty simultaneously comprised 9% of the participants and had the highest incidence of major adverse cardiovascular events. AMI ¼ acute myocardial infarction;
CAD ¼ coronary artery disease; CI ¼ cognitive impairment; MACE ¼ major adverse cardiovascular event; PVD ¼ peripheral vascular disease.

reporting and may have resulted in misclassification characteristics that are associated with patients who
bias. Despite this limitation, our estimates approxi- presented with both frailty and CI, as compared to
mate other cohorts of older adults in the community, those presenting with frailty alone or CI alone. To
and this study represents the best approximation of CI account for selection bias we have constructed a
and frailty in the study population. multivariable proportional hazard regression model
Internal validity refers to the degree to which the to account for baseline differences between groups.
results represent the true association between an After accounting for adjustments, the increase in
exposure and an outcome in a particular study MACE in patients with frailty and CI remains posi-
population. The main threats to internal validity of tive. While this is not a randomized study, our
this study include selection bias, testing, and his- sample size is large enough to increase the internal
tory. There could be preexisting clinical validity by reducing the chance of random error
10 Damluji et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Frailty and Cognitive Impairment in Older Adults With CVD MAY 2023:100318

associated with testing. Furthermore, the use of reverse these geriatric risks may improve cardiovas-
validated instruments and objective measures of cular outcomes in older population.
frailty and CI minimize measurement error. ACKNOWLEDGMENT The authors would like to
While there are specific criteria to diagnose de- acknowledge Devon Stuart, MA, CMI, for her valuable
mentia, outlined in the DSM (The Diagnostic and assistance with medical illustration.
Statistical Manual of Mental Disorders)-5, the AD8 is
one of many validated tools used to screen for CI in FUNDING SUPPORT AND AUTHOR DISCLOSURES
21
older adults. While there are several advantages to
Drs Damluji, Xue, Walston, and Gerstenblith have received research
using AD8 as a screening instrument, such as brevity, funding from the Pepper Scholars Program of the Johns Hopkins
ease of use, high sensitivity, and specificity to capture University Claude D. Pepper Older Americans Independence Center
dementia, it has important limitations. These include funded by the National Institute on Aging P30-AG021334. Dr Damluji
has received a Mentored Patient-Oriented Research Career Develop-
limited diagnostic accuracy, limited validation, de-
ment Award from the National Heart, Lung, and Blood Institute K23-
pendency on the interviewer, and inability to distin- HL153771-01. Dr Nanna has received funding from the American Col-
guish between the types and forms of dementia. lege of Cardiology Foundation supported by the George F. and Ann
While AD8 may have underestimated the true prev- Harris Bellows Foundation and from the National Institute on Aging/
National Institutes of Health from R03AG074067 (GEMSSTAR award).
alence of CI in practice, there were still not an insig-
All other authors have reported that they have no relationships
nificant number of individuals in this cohort with CI, relevant to the contents of this paper to disclose.
highlighting the importance of studying CI in this
older population at risk for CAD. After a patient ADDRESS FOR CORRESPONDENCE: Prof Abdulla A.
screens positive on any Alzheimer dementia Damluji, Johns Hopkins University School of Medi-
screening tool, a comprehensive evaluation by a cine, Inova Center of Outcomes Research, 3300
trained health care professional with experience in Gallows Road, Falls Church, Virginia 22042, USA.
diagnosing and treating CI and dementia should be E-mail: [email protected].
conducted. It is important to note that this study was
not designed to differentiate between specific sub-
PERSPECTIVES
types of dementia (eg, Alzheimer’s disease, dascular
dementia, etc) or the medications used to treat each
phenotype of dementia. COMPETENCY IN MEDICAL KNOWLEDGE: Older
adults with CI and/or frailty are at greater risk for
adverse cardiovascular outcomes than those who
CONCLUSIONS
develop neither. Although most older adults develop
CI first, followed by frailty, those who develop both
In the NHATS study population, we found that the
simultaneously are at the highest risk for adverse
simultaneous development of frailty and CI is asso-
cardiovascular outcomes.
ciated with significantly higher risk of MACEs and
mortality than development of frailty first or CI first,
TRANSLATIONAL OUTLOOK: Screening for CI and
even after controlling for cardiovascular risk factors.
frailty can identify patients at increased risk for poor
Assessment of frailty and CI in older adults should be
cardiovascular outcomes. Biologic underpinnings
part of their routine care, which may ultimately
responsible for these relationships should be studied.
affect therapeutic choices and modify cardiovascular
outcomes. Effective interventions to prevent and

REFERENCES

1. U.S. Census Bureau. Current population survey, disease: JACC state-of-the-art review. J Am Coll cognitive impairment alone, physical frailty
annual social and economic; table 1: population by Cardiol. 2022;79:482–503. alone, and both. J Am Geriatr Soc. 2020;68:
age and sex: 2019. 2020. Accessed April 15, 2023. 2822–2830.
4. O’Neill DE, Forman DE. Cardiovascular care of
http://census.gov/data/tables/2019/demo/age-
older adults. BMJ. 2021;374:n1593. 7. Kasper JD, Freedman VA. National Health and
and-sex/2019-older-population.html
5. Avila-Funes JA, Amieva H, Barberger-Gateau P, Aging Trends Study User Guide: Rounds 1-8 Final
2. Ortman JA, Velkoff VA, Hogan H. An Aging et al. Cognitive impairment improves the predic- Release. Johns Hopkins University School of Public
Nation: The Older Population in the United States. tive validity of the phenotype of frailty for adverse Health; 2019.
Current Population Reports. U.S. Census Bureau; health outcomes: the three-city study. J Am Ger-
8. Damluji AA, Chung S-E, Xue Q-L, et al. Frailty
2014. iatr Soc. 2009;57:453–461.
and cardiovascular outcomes in the National
3. Ijaz N, Buta B, Xue QL, et al. Interventions for 6. Ge ML, Carlson MC, Bandeen-Roche K, et al. Health and Aging Trends study. Eur Heart J.
frailty among older adults with cardiovascular U.S. national profile of older adults with 2021;42:3856–3865.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Damluji et al 11
MAY 2023:100318 Frailty and Cognitive Impairment in Older Adults With CVD

9. Chu NM, Bandeen-Roche K, Tian J, et al. Hier- 14. Fine JP, Gray RJ. A proportional hazards model impairment: an integrative review. Int J Geriatr
archical development of frailty and cognitive for the subdistribution of a competing risk. J Am Psychiatry. 2018;33(9):1177–1197.
impairment: clues into etiological pathways. Stat Assoc. 1999;94:496–509.
19. Fountotos R, Ahmad F, Bharaj N, et al. Multi-
J Gerontol A Biol Sci Med Sci. 2019;74:1761–1770.
15. Brigola AG, Rossetti ES, Dos Santos BR, et al. component intervention for frail and pre-frail
10. Fried LP, Tangen CM, Walston J, et al. Frailty Relationship between cognition and frailty in older adults with acute cardiovascular conditions:
in older adults: evidence for a phenotype. elderly: a systematic review. Dement Neuro- the TARGET-EFT randomized clinical trial. J Am
J Gerontol A Biol Sci Med Sci. 2001;56:M146– psychol. 2015;9:110–119. Geriatr Soc. Published online January 16, 2023.
M156. https://doi.org/10.1111/jgs.18228
16. Faulkner KM, Uchmanowicz I, Lisiak M,
11. Bandeen-Roche K, Seplaki CL, Huang J, et al. Cichon E, Cyrkot T, Szczepanowski R. Cognition 20. Altman DG, Royston P. The cost of dichoto-
Frailty in older adults: a nationally representative and frailty in patients with heart failure: a sys- mising continuous variables. BMJ. 2006;332:
profile in the United States. J Gerontol A Biol Sci tematic review of the association between frailty 1080.
Med Sci. 2015;70:1427–1434. and cognitive impairment. Front Psychiatry.
21. Yang L, Yan J, Jin X, et al. Screening for de-
2021;12:713386.
12. Damluji AA, Chung S-E, Xue Q-L, et al. Physical mentia in older adults: comparison of mini-mental
frailty phenotype and the development of geriatric 17. Malara A, Sgro G, Caruso C, et al. Relationship state examination, mini-cog, clock drawing test
syndromes in older adults with coronary heart between cognitive impairment and nutritional and AD8. PLoS One. 2016;11:e0168949.
disease. Am J Med. 2021;134:662–671.e1. assessment on functional status in Calabrian long-
term-care. Clin Interv Aging. 2014;9:105–110.
13. Galvin JE, Roe CM, Powlishta KK, et al. The
AD8: a brief informant interview to detect de- 18. Fogg C, Griffiths P, Meredith P, Bridges J. KEY WORDS cognitive impairment,
mentia. Neurology. 2005;65:559–564. Hospital outcomes of older people with cognitive coronary disease, frailty, older adults
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

EDITORIAL COMMENT

One Thing Leads to Another

Embracing Complexity in Geriatric Cardiology*

Ashok Krishnaswami, MD, MAS,a Min Ji Kwak, MD, MS, DRPHb

B orrowing from a 1983 popular song by the En-

glish band the Fixx, one thing can lead to
another.1 Although teenage romance is not
the current focus, the similarities with geriatric cardi-
(Figure 1). A mindset of embracing the complexity, by
increasing awareness and comfort of geriatric condi-
tions/syndromes, will bear fruit in multiple ways. At
the bedside, this is the comprehensive geriatric
ology is the necessity to embrace and navigate examination.
complexity. The teenage years are often linked to so- With this background, in this issue of JACC:
cial complexities. During the Elderhood 2 years, part Advances, Damluji et al3 report on a study evaluating
of the difficulty is the complexity connected with the relationship between physical function, cognitive
multiple chronic medical conditions. The interactions function, and the development of major adverse
between the passage of time and medical illnesses cardiovascular events (MACEs). Their source popula-
often result in a loss of skeletal muscle strength, tion was from the National Health and Aging Trends
mass, or function, and/or cognitive decline that can Study (NHATS) 4 data set, which began in 2011. The
range from minimally disruptive to sometimes devas- focus of NHATS was to foster research to decrease
tating consequences. In fact, these 2 disorders create disability, maximize health, and maintain indepen-
sufficiently high complexity in the lives of older dent functioning. 2 NHATS investigators conduct
adults and are closely associated with conditions yearly interviews of a nationally representative sam-
such as falls, loss of automobile driving privileges, ple of Medicare beneficiaries $65 years enriched with
diminished hearing, urinary and fecal incontinence, African Americans to gather a plethora of geriatric
depression, delirium, elder abuse, polypharmacy, and disease-specific conditions at baseline and
and a higher probability of residing at a skilled follow-up.
nursing home or other models of care aside from The current study population (n ¼ 2,189) were
one’s own home. those who did not have any coronary artery disease
Focusing only on the iceberg tip will certainly miss prior to their baseline visit in 2011. The exposure
the wealth of data beneath the surface. Addressing variable was a 4-level category based on the presence
social difficulties in the teenage years as a behavior or absence of cognitive impairment (CI) (AD8
problem can miss important details. Similarly, the score $2), impairment in at least 1 cognitive domain
focus on the dominant cardiovascular morbidity will (executive function, memory), self-reported diag-
miss the myriad of other geriatric conditions in older nosis of minor neurocognitive impairment or major
adults that are pertinent toward cardiovascular care neurocognitive impairment (dementia), and physical
frailty (Fried score) during the 2 years of follow-up. It
was divided as Intact (no frailty or CI reference);
*Editorials published in JACC: Advances reflect the views of the authors Frail-first (physical frailty, no CI); CI-first (CI, no
and do not necessarily represent the views of JACC: Advances or the
physical frailty); Frail-CI (both). The primary com-
American College of Cardiology.
posite (all-cause mortality, acute myocardial infarc-
From the aDivision of Cardiology, Kaiser Permanente San Jose Medical
b tion, any coronary heart disease, stroke, or peripheral
Center, San Jose, California, USA; and the Division of Geriatric and
Palliative Medicine, McGovern Medical School, University of Texas, vascular disease) MACE outcome was the time to first
Houston, Texas, USA. event.
The authors attest they are in compliance with human studies commit-
For the cardiovascular team, some of the utilized
tees and animal welfare regulations of the authors’ institutions and Food
and Drug Administration guidelines, including patient consent where
gerocentric variables may not be recognized. The
appropriate. For more information, visit the Author Center. average age was highest in those with both physical

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100358

2 Krishnaswami and Kwak JACC: ADVANCES, VOL. 2, NO. 3, 2023

Embracing Complexity in Geriatric Cardiology MAY 2023:100358

F I G U R E 1 Looking Under the Iceberg Beyond the Dominant Cardiac Morbidity

and cognitive dysfunction and lowest in those with evaluation using Montreal Cognitive Assessment,
no impairments (83.9 years vs 74.6 years). Interest- Mini Mental Status Examination, or Saint Louis Uni-
ingly, there was a female:male (60:40) preponderance versity Mental Status will need additional expertise.
in the study cohort. Self-reported, noncardiac dis- However, the cardiovascular team should have a
eases were displayed and those with both frailty-CI cursory comfort of these tools.
had often the highest prevalence in the individual The authors describe the temporal occurrences of
disease states and in the cumulative number of dis- physical and CIs and the association with MACE.
eases. The investigators did not have a cumulative Among a total of 2,189 older adults, they found that
burden index, such as the Elixhauser or Charlson 7% developed Frail-first, 38% developed CI-first, 9%
scores, which could have improved the characteriza- developed both Frail-CI with 46% classified as Intact.
tion of global multimorbidity. Understandably, those The risk-adjusted Frail-CI group showed the highest
with both CI and frailty had the greatest number of hazard ratio (HR) of MACE (1.81, 95% CI: 1.47-2.23),
Katz activities of daily living (ADL) or Lawton followed by the Frail-first (1.46, 95% CI: 1.17-1.81) and
instrumental ADL (IADL) disabilities along with the CI-first (1.31, 95% CI: 1.15-1.50). A competing risks
highest health care utilizations as per hospital length analysis found the risk-adjusted subhazard ratio for
of stay. MACE in the Frail-CI group significantly reduced at
A standardized geriatric evaluation across cardio- 0.57 (95% CI: 0.40-0.80) possibly related to the high
vascular diseases is needed. Although the ascertain- risk of death, confounding or undetected effect
ment of physical and cognitive function at the time of modification.
surgery or transcatheter approaches to valvular dis- Although the exposure category suggests a tem-
eases have caught the cardiovascular teams’ eye, poral trend of one occurring before the other, the
knowledge and understanding of independence in potential mechanisms are not clear and could be as-
Katz ADLs (bathing, toileting, walking/transferring, sociations (with shared risk factors), causal (through
eating, dressing), Lawton IADLs (telephone use, different pathways), or epiphenomenon (treatment of
driving, chores [preparing meals, light housework, hypertension such as with beta-blockers causing
shopping], managing money, managing medications) frailty or CI). Historically, major cardiovascular ran-
and cognition (age related, minor CI, or dementia) is domized controlled trials have excluded older adults
essential in routine cardiovascular care. There are with multiple geriatric conditions, such as poly-
well-validated tools to assess physical functional pharmacy, frailty, or CIs, making the understanding
changes such as Fried Frailty Score, and screen for of the interplay on cardiovascular outcomes even
cognitive dysfunction by Alzheimer’s dementia harder. In that context, the current article holds the
screening (AD8), and Mini-Cog. More formal critical value of attempting to untangle the complex
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Krishnaswami and Kwak 3
MAY 2023:100358 Embracing Complexity in Geriatric Cardiology

interplay of 2 dominant geriatric conditions on MACE To establish a diagnosis of dementia, the Diag-
among older adults with incident cardiovascular nostic and Statistical Manual of Mental Disorders-5
diseases. (DSM-5) requires 3 items. A cognitive domain deficit.
The investigators should be commended on their That the deficit be associated with diminished IADL
desire to look beyond the surface. They attempted to (functional loss) and that the decline is new and not
untangle the complex relationships between geriatric associated with delirium or other medical/psychiatric
conditions and cardiovascular diseases to identify conditions. Well-validated tools to screen for de-
which one thing, among the 3 (frailty, CI, or MACE), mentia, such as Montreal Cognitive Assessment or
leads to another. In geriatric cardiology, the associa- Saint Louis University Mental Status require not only
tions between a geriatric condition and cardiovascu- extra time to perform but also training. A Mini-Cog is
lar diseases outcomes or between 2 geriatric a quick screening option. Of course, older adults with
conditions are fairly well studied. 5 Previous work cardiovascular diseases may have other geriatric
from this group showed that frailty, without prior conditions including falls, social isolation, or malnu-
coronary artery disease, was associated with a higher trition that deserves attention from the cardiovascu-
relative risk of MACE and all-cause mortality (HR: lar clinicians. Therefore, practical and feasible tools
1.77, 95% CI: 1.53-2.06).6 Other studies have shown that can assess geriatric conditions in a comprehen-
that frailty is a risk factor of accelerated development sive fashion but also fit the cardiovascular clinic’s
of other geriatric adverse outcomes such as dementia, resources should be selected to provide optimal
loss of independence, ADL disability, IADL disability, geriatric cardiology care. 6
and mobility disability. In conclusion, one thing can lead to another but
Additionally, the authors point out the important the whammy of 2 things (frailty and CI) is often
clinical implications of cardiovascular clinicians fraught with badness. The cardiovascular team
routinely incorporating geriatric condition assess- should recognize the role of physical frailty, CI, and
ments, such as frailty or CI, before planning a car- its associations with MACE. The assessment and
diovascular care strategy. Frailty assessments using care of older adults with cardiovascular diseases
Fried’s frailty phenotype and CI using established and geriatric conditions is difficult. It will require a
diagnosis or AD8 score were the tools available in the multidisciplinary team approach. This is one fron-
data set. However, in real practice, cardiovascular tier among the challenges facing geriatric
clinicians may not have sufficient training and re- cardiology.
sources to assess geriatric conditions using such
tools. For example, to use the Fried’s frailty pheno-
FUNDING SUPPORT AND AUTHOR DISCLOSURES
type, grip strength and walking speed need to be
The authors have reported that they have no relationships relevant to
assessed. However, most cardiovascular clinics may the contents of this paper to disclose.
not have the dynamometer to assess the grip strength
or ruler and stop-watch to check the walking speed.
Furthermore, their fast-paced clinics may not have ADDRESS FOR CORRESPONDENCE: Dr Ashok Krish-
embedded workflows to assess frailty or CI. In such naswami, Division of Cardiology, Kaiser Permanente
cases, cardiovascular clinics may choose to use other San Jose Medical Center, 270 International Circle,
tools to assess frailty using Clinical Frailty Scale or Building 3, 2nd Floor, San Jose, California 95119, USA.
FRAIL scale questionnaire, which do not require E-mail: [email protected]. Twitter:
additional equipment. @cardskrish, @minjikwak2.

REFERENCES

1. The Fixx, Woods A, Curnin C, West-Oram J, impairment and their association with incident 6. Damluji AA, Chung SE, Xue QL, et al. Frailty and
Greenall P. One Thing Leads to Another. Album- cardiovascular disease. JACC: Adv. 2023;2:100318. cardiovascular outcomes in the National Health
Reach the Beach. 1983. Accessed April 20, 2023. and Aging Trends study. Eur Heart J. 2021;42:
4. Freedman VA, Schrack JA, Skehan ME,
https://www.thefixx.com/music 3856–3865.
Kasper JD. National Health and Aging Trends
2. Aronson L. Elderhood: Redefining Aging, Trans- Study User Guide: Rounds 1-11 Final Release.
forming Medicine, Reimagining Life. Bloomsbury Johns Hopkins University School of Public Health;
Publishing; 2019. 2022.
KEY WORDS cognitive function,
3. Damluji AA, Ijaz N, Chung S-E, et al. Hierarchical 5. Goyal P, Kwak MJ, Al Malouf C, et al. Geriatric car- comprehensive geriatric assessments, dementia,
development of physical frailty and cognitive diology: coming of age. JACC: Adv. 2023;2:100318. frailty, geriatric conditions, physical function
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ORIGINAL RESEARCH

EMERGING TECHNOLOGIES AND INNOVATIONS

A Computable Algorithm for Medication

Optimization in Heart Failure With
Reduced Ejection Fraction
Michael P. Dorsch, PHARMD, MS,a,b Aaron Sifuentes, MD,c David J. Cordwin, PHARMD,a Rachel Kuo, PHARMD,a
Brigid E. Rowell, MA,a Juan J. Arzac,a Ken DeBacker,a Jessica L. Guidi, MD,b,c Scott L. Hummel, MD, MS,b,c,d
Todd M. Koelling, MDb,c

ABSTRACT

BACKGROUND Guideline-directed medical therapy (GDMT) optimization can improve outcomes in heart failure with
reduced ejection fraction.

OBJECTIVES The objective of this study was to determine if a novel computable algorithm appropriately recommended
GDMT.

METHODS Clinical trial data from the GUIDE-IT (Guiding Evidence-Based Therapy Using Biomarker Intensified Treat-
ment in Heart Failure) and HF-ACTION (Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training)
trials were evaluated with a computable medication optimization algorithm that outputs GDMT recommendations and a
medication optimization score (MOS). Algorithm-based recommendations were compared to medication changes. A Cox
proportional-hazards model was used to estimate the associations between MOS and the composite primary end point for
both trials.

RESULTS The algorithm recommended initiation of angiotensin-converting enzyme inhibitor/angiotensin receptor

blocker, beta-blockers, and mineralocorticoid receptor antagonists in 52.8%, 34.9%, and 68.1% of GUIDE-IT visits,
respectively, when not prescribed the drug. Initiation only occurred in 20.8%, 56.9%, and 15.8% of subsequent visits.
The algorithm also identified dose titration in 48.8% of visits for angiotensin-converting enzyme inhibitor/angiotensin
receptor blockers and 39.4% of visits for beta-blockers. Those increases only occurred in 24.3% and 36.8% of
subsequent visits. A higher baseline MOS was associated with a lower risk of cardiovascular death or heart failure
hospitalization (HR: 0.41; 95% CI: 0.21-0.80; P ¼ 0.009) in GUIDE-IT and all-cause death and hospitalization (HR: 0.61;
95% CI: 0.44-0.84; P ¼ 0.003) in HF-ACTION.

CONCLUSIONS The algorithm accurately identified patients for GDMT optimization. Even in a clinical trial with robust
protocols, GDMT could have been further optimized in a meaningful number of visits. The algorithm-generated MOS was
associated with a lower risk of clinical outcomes. Implementation into clinical care may identify and address suboptimal
GDMT in patients with heart failure with reduced ejection fraction. (JACC Adv 2023;2:100289) © 2023 The Authors.
Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an open access article under the
CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

From the aDepartment of Clinical Pharmacy, College of Pharmacy, University of Michigan, Ann Arbor, Michigan, USA; bFrankel
Cardiovascular Center, University of Michigan, Ann Arbor, Michigan, USA; cDepartment of Internal Medicine, Medical School,
University of Michigan, Ann Arbor, Michigan, USA; and the dAnn Arbor Veterans Affairs Health System, Ann Arbor, Michigan, USA.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received July 20, 2022; revised manuscript received December 5, 2022, accepted January 30, 2023.

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100289

2 Dorsch et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

HFrEF Medication Optimization Algorithm MAY 2023:100289

I
ABBREVIATIONS n heart failure with a reduced ejection PATIENT POPULATIONS. The GUIDE-IT trial was
AND ACRONYMS fraction (HFrEF), clinical trials have an 894-patient randomized controlled trial that
established the benefit of medications 1-9 studied the effects of a natriuretic peptide-guided
ACEI = angiotensin-converting
enzyme inhibitor
and as national guidelines10-12 that recom- management strategy compared to the standard of
mend those therapies, but it has been well care in patients with HFrEF with an emphasis on
API = Application Programming
Interface documented that this knowledge can take titrating medical therapy, including GDMT. 17 The
ARB = angiotensin receptor years to be broadly implemented into prac- GUIDE-IT trial was an ideal data set to validate this
blocker tice.13 The registry to Improve the Use of algorithm because of the extensive medication data
CDSS = clinical decision Evidence-Based Heart Failure Therapies in collected and the purpose of the study was to opti-
support systems the Outpatient Setting prospective cohort mize GDMT over time. Blood pressure, heart rate,
CV = cardiovascular study recently demonstrated low use of potassium, serum creatinine, and medications (drug
EHR = electronic health record guideline-directed medical therapy (GDMT) and daily doses) were documented at baseline,
FDR = false discovery rate in U.S. patients with HFrEF. 14 Angiotensin- 2 weeks, 6 weeks, 3 months, and every 3 months
GDMT = guideline-directed converting enzyme inhibitors or angiotensin thereafter until month 24 of the study, a potential 11
medical therapy receptor blockers (ACEIs/ARBs) were only visits per patient. The primary outcome for the
HF = heart failure used in 79.6% of patients, beta-blockers GUIDE-IT trial was the composite of cardiovascular
HFrEF = heart failure with a were only used in 86% of patients, and miner- (CV) death or HF hospitalization.
reduced ejection fraction alocorticoid receptor antagonists (MRAs) HF-ACTION was a 2,331-patient randomized
MOS = medication optimization were only used in 36.1% of patients. More controlled trial that studied the effects of aerobic
score
recently, the Change the Management of Pa- exercise training in ambulatory patients with HF and
MRA = mineralocorticoid
tients with Heart Failure registry has also an ejection fraction of <35%.18 Values for the algo-
receptor antagonist
demonstrated the need for further GDMT rithm variables were available at baseline. Subse-
SBP = systolic blood pressure
optimization. 15 quent time points were not included because the
To address these issues, the American College input data needed for the algorithm were not
of Cardiology published an expert consensus de- collected. The primary outcome for the HF-ACTION
cision pathway for optimizing heart failure (HF) trial was the composite all-cause death or
treatment that recommended the use of electronic hospitalization.
health records (EHRs) to reduce errors, improve The 2 data sets, GUIDE-IT and HF-ACTION, com-
decision support, and facilitate guideline adher- plement each other well to evaluate the computable
ence.16 This led our research group to create an medication optimization algorithm. GUIDE-IT pro-
Application Programming Interface (API) using the vides rich, longitudinal data for patients enrolled at
American College of Cardiology/American Heart hospital discharge that includes all variables needed
Association HFrEF guidelines. The objective of this to assess medication optimization at each time point.
study was to validate that the API appropriately HF-ACTION provides a large data set of patients with
recommended GDMT from clinical trial data and to HFrEF enrolled in the ambulatory setting with all
determine if the API-generated information about variables needed to assess medication optimization at
medication optimization was associated with clin- baseline.
ical outcomes. COMPUTABLE MEDICATION OPTIMIZATION ALGORITHM.
The HFrEF medication optimization algorithm crea-
METHODS tion began with the narrative guideline, 11 then led to
structured decision trees developed for each drug
This study retrospectively validated a computable class (ACEI/ARB/angiotensin receptor/neprilysin in-
medication optimization algorithm created for hibitor (ARNI), beta-blocker, MRA, and hydralazine/
HFrEF using the GUIDE-IT (Guiding Evidence-Based nitrate). The algorithm was validated by HF physi-
Therapy Using Biomarker Intensified Treatment in cians and cardiology pharmacists and was coded in an
Heart Failure) 17 and HF-ACTION (Heart Failure: A executable computer format. The executable format
Controlled Trial Investigating Outcomes of Exercise allowed for more large-scale validation of the algo-
Training) clinical trial data acquired from the NHLBI rithm with synthetically generated patient data. The
BioLINCC (National Heart Lung and Blood Institute, next step in testing the algorithm was this study.
Biological Specimen and Data Repository Informa- The HFrEF medication optimization algorithm was
tion Coordinating Center) repository. The study was created as a REpresentational State Transfer API, so
approved and determined to be not regulated by inputs, generated from an outside source, could be
the University of Michigan Institutional Review computed by the algorithm to provide recommenda-
Board. tions for the medications to optimize on an individual
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Dorsch et al 3
MAY 2023:100289 HFrEF Medication Optimization Algorithm

C ENTR AL I LL UST RA TI O N Overview of the Medication Optimization Algorithm

Dorsch MP, et al. JACC Adv. 2023;2(3):100289.

The left top side of the figure shows the inputs of the algorithm. The right top side shows the outputs of the algorithm and the potential
future applications for each output. The bottom table shows opportunities for GDMT medication optimization in GUIDE-IT visits.
ACEI ¼ angiotensin-converting enzyme inhibitor; API ¼ Application Programming Interface; ARB ¼ angiotensin receptor blocker;
GDMT ¼ guideline directed medical therapy; GUIDE-IT ¼ Guiding Evidence-Based Therapy Using Biomarker Intensified Treatment in Heart
Failure; MRA ¼ mineralocorticoid receptor antagonist.

patient. The current version of the algorithm (version

T A B L E 1 Examples of Medications, the Algorithm Recommendations, and the Medication
0.487) is designed for the inputs of medications Optimization Score Value
(name and daily dose), New York Heart Association
Example Medications Recommendations MOS
classification, systolic blood pressure (SBP), heart rate
Patient 1 Metoprolol succinate 25 mg daily,  Increase the patient’s current 22%
(HR), serum creatinine, potassium, allergies/intoler- lisinopril 2.5 mg daily beta blocker.
ance, and race from one patient at a time. The  Increase the patient’s current
ACE inhibitor.
Central Illustration demonstrates a visual for the in-  Start a mineralocorticoid
puts and outputs of the algorithm. receptor antagonist.
Patient 2 Carvedilol 25 mg twice daily,  Switch to sacubitril/valsartan. 56%
The algorithm identifies specific HFrEF GDMT
lisinopril 20 mg daily  Start a mineralocorticoid
medications approved by the Food and Drug Admin- receptor antagonist.
istration. The algorithm outputs medication optimi- Patient 3 Metoprolol succinate 200 mg daily,  Switch to sacubitril/valsartan. 89%
enalapril 40 mg daily,
zation recommendations for each GDMT drug class. spironolactone 25 mg daily
The algorithm also provides a medication optimiza-
tion score (MOS). The MOS is a percent between The table provides examples of the algorithm recommendations and MOS when a patient is NYHA functional
class II and has normal potassium, serum creatine, systolic blood pressure, and heart rate.
0 (least optimized) and 100 (most optimized) which ACE ¼ angiotensin-converting enzyme.
represents the extent of medication optimization that
4 Dorsch et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

HFrEF Medication Optimization Algorithm MAY 2023:100289

needs to be performed. As the patient is prescribed predicts the primary outcome. The findcut Statistical
more GDMT and at higher doses, the score is closer to Analysis System macro created by the Mayo Clinic
100%. Table 1 provides examples of the algorithm was used to perform this analysis. 20-22 The MOS was
recommendations and the MOS value. To test the the continuous measure and the first all-cause death
inputs and results for the algorithm, visit https:// or hospitalization was the time-to-event outcome in
decisionalgorithm.shinyapps.io/heartfailure/. the macro. A combination of the Cox model Wald P-
values and false discovery rate (FDR) P values from
STATISTICAL ANALYSIS. Patients were excluded the macro were used to identify cut points for the
from the analysis if the data needed for the algorithm MOS. The FDR P values, also known as q values, are
were unavailable at the time point for the analysis. corrected to adjust for multiple comparisons. For all
For baseline demographics and outcomes, categorical statistical tests, differences were considered statisti-
variables were described with frequencies and per- cally significant at a P value of <0.05. All statistical
centages, and continuous variables were described analyses were performed with SAS version 9.4
with mean  SD or median (IQR) where appropriate. (SAS Institute, Inc).
Continuous variables were compared using a t-test
when normally distributed and a Wilcoxon rank RESULTS
test when the variable is not normally distributed.
Categorical variables were analyzed using a chi- In the GUIDE-IT data, 841 patients had appropriate
square test or Fisher exact test where appropriate. data at baseline visit to be included. At baseline,
Using the GUIDE-IT trial data, the recommenda- 80.9% (n ¼ 681) of patients were taking an ACEI/ARB,
tions generated in the algorithm were compared to 96.4% (n ¼ 811) of patients were taking a beta-
the medications documented at each visit. We blocker, and 50.7% (n ¼ 426) of patients were taking
compared the important clinical parameters be- an MRA. The algorithm identified the need for GDMT
tween those visits that a patient would receive and optimization in 69.4% (n ¼ 584) of patients.
not receive a recommendation for that drug class In the HF-ACTION data, 2,130 patients had appro-
from the medication optimization algorithm. The priate data at the HF-ACTION baseline visit to be
important clinical parameters for the ACEI/ARB included. At baseline, 94.4% (n ¼ 2,011) of patients
class were SBP, serum creatinine, and potassium. were taking an ACEI/ARB, 94.5% (n ¼ 2,012) of pa-
The important clinical parameters for the beta- tients were taking a beta-blocker, and 45.2% (n ¼ 962)
blocker class were SBP and HR. The important of patients were taking an MRA. The algorithm iden-
clinical parameters for the MRA class were serum tified the need for GDMT optimization in 72.8%
creatinine and potassium. The change in medica- (n ¼ 1,552) of patients. Table 2 demonstrates the de-
tions from a visit to visit was also compared to mographics of the patients from the baseline visit.
those recommended by the algorithm. In the GUIDE-IT data, 883 patients had enough data
A Cox proportional-hazards model was used to at any follow-up time to be included in the analysis,
estimate the associations between MOS (independent with a median of 6 follow-up time points and a total of
variable) with clinical outcomes. The composite pri- 5,733 visits. Over the 5,733 visits, 79.7% (n ¼ 4,567),
mary outcome for the GUIDE-IT trial was the first CV 96.6% (n ¼ 5,535), and 56.5% (n ¼ 3,218) visits
death or HF hospitalization (dependent variable). The demonstrated patients were taking an ACEI/ARB,
MOS was also treated as a time-dependent covariate beta-blocker, and MRA, respectively. For all visits,
where the MOS immediately preceding an event or the MOS was a median of 76% (IQR: 50%-100%). At
censoring timepoint was used as the independent baseline, the median MOS was 61% (IQR: 40%-100%).
variable. The composite primary outcome for the HF- In those patients that the algorithm identified needing
ACTION trial was first all-cause death or hospitaliza- medication optimization (MOS <100%), the median
tion (dependent variable). Additional analysis was MOS was 50% (IQR: 34%-64%). Figure 1 shows a
performed to control for the HF-ACTION risk score19 Sankey diagram for the flow of patients between
as a covariate because none of these variables are different MOS categories throughout the study.
used as inputs for the medication optimization algo- At baseline in HF-ACTION, the median MOS was
rithm. A sensitivity analysis was also performed using 57% (IQR: 44%-100%). In HF-ACTION patients that
the lowest dose for all ARBs, which is the base anal- the algorithm identified needing medication optimi-
ysis, and the highest dose for all ARBs because ARB zation (MOS <100%), the median MOS was 50%
dosing was not available in HF-ACTION. A cut point (IQR: 40%-61%). See the supplemental figures,
analysis was performed using the HF-ACTION data to Supplemental Figures 1 and 2, representing the his-
find the ideal cut point where the MOS percentage togram of the MOS in each data set.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Dorsch et al 5
MAY 2023:100289 HFrEF Medication Optimization Algorithm

T A B L E 2 Baseline Demographics

GUIDE-IT HF-ACTION

Total MOS <100% MOS 100% Total MOS <100% MOS 100%
(n ¼ 841) (n ¼ 584) (n ¼ 257) (n ¼ 2,130) (n ¼ 1,552) (n ¼ 578)

Age, y 61.3  14 61.3  14 61.1  15 58.6  13 58.9  12 57.7  12

Women 262 (31.2) 188 (32.2) 74 (28.8) 599 (28.1) 437 (28.2) 162 (28)
Race
White 471 (56) 302 (51.7) 169 (65.7) 1,322 (62.1) 953 (61.4) 369 (63.8)
Black 299 (35.6) 235 (40.2) 64 (24.9) 669 (31.4) 501 (32.3) 168 (29.1)
Other 71 (8.4) 47 (8) 24 (9.3) 139 (6.5) 98 (6.3) 41 (7.1)
Ischemic HF etiology 421 (50) 280 (48) 141 (54.9) 1,096 (51.5) 796 (51.29) 300 (51.9)
Diabetes mellitus 388 (46.1) 252 (48.3) 106 (41.3) 678 (31.8) 499 (32.2) 179 (31)
Chronic kidney diseasea 309 (36.7) 225 (38.5) 84 (32.7) 742 (34.8) 508 (32.7) 234 (40.5)
Systolic BP, mm Hg 115.8  20 122  18 101.7  15 113.8  18 119.1  17 99.5  13
Heart Rate, beats/min 77  15 75  14 78  15 70  11 71  12 70  11
Potassium, mmol/L 4.4  0.6 4.3  0.5 4.5  0.7 4.3  0.8 4.3  0.5 4.5  0.6
Creatinine, mg/dL 1.45  0.6 1.4  0.5 1.56  0.8 1.33  0.8 1.26  0.5 1.51  1.3
HF-ACTION risk score - - - 53  12 52.7  12 53.8  12
ACEI/ARB 681 (81) 465 (79.6) 216 (84.1) 2011 (94.4) 1,456 (93.8) 555 (96)
Beta-blocker 811 (96) 565 (96.8) 246 (95.7) 2012 (94.5) 1,466 (94.5) 546 (94.5)
MRA 426 (50.7) 221 (37.8) 205 (79.8) 962 (45.2) 459 (29.6) 503 (87)

Values are mean  SD or n (%). aIn GUIDE-IT, a chronic kidney disease diagnosis was collected at baseline and HF-ACTION chronic kidney disease was defined as a baseline eGFR category of 3
or greater.
ACEI/ARB ¼ angiotensin-converting enzyme inhibitor/angiotensin receptor blocker; BP ¼ blood pressure; GUIDE-IT ¼ Guiding Evidence-Based Therapy Using Biomarker Intensified
Treatment in Heart Failure; HF ¼ heart failure; HF-ACTION ¼ Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training.

GUIDE-IT MEDICATION ADJUSTMENTS AND THE (50.5%) patient visits, leaving 577 (49.5%) visits
ALGORITHM. Over the course of the GUIDE-IT trial, where the algorithm would not recommend initiation
there were 1,166 (20.3%) visits where patients were of ACEI/ARB. Patient visits where ACEI/ARBs were
not on an ACEI/ARB. Of those not on ACEI/ARB, the not recommended by the algorithm had significantly
algorithm recommended initiating ACEI/ARB in 589 lower SBP (109  20 mm Hg vs 127.9  14 mm Hg;

F I G U R E 1 Sankey Plot of Patients Transferring Between Different Medication Optimization Score Categories Over Time

Node height represents the proportion of patients in each category at the given times of baseline, 6 months, 12 months, and 24 months. Links
represent the proportion of patients moving from their node of origin to their respective category at the next timepoint.
6 Dorsch et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

HFrEF Medication Optimization Algorithm MAY 2023:100289

T A B L E 3 Opportunities for Guideline-Directed Medication Therapy Optimization in GUIDE-IT Visits

ACEI/ARB Beta-Blocker MRA

Initiation Titration Initiation Titration Initiation

Algorithm identified visits for GDMT optimization 53% (480/908) 49% (1,475/3,025) 35% (51/146) 39% (1,158/2,942) 68% (1,421/2088)
Medication prescribed at the next visit 21% (100/480) 24% (358/1,475) 57% (29/51) 37% (426/1,158) 16% (224/1,421)
Opportunities for GDMT optimization 79% (380/480) 76% (1,117/1,475) 43% (22/51) 63% (732/1,158) 84% (1,197/1,421)

Values are % (n/N). The table summarizes the opportunities for GDMT optimization identified by the algorithm.
ACEI/ARB ¼ angiotensin-converting enzyme inhibitor/angiotensin receptor blocker; GDMT ¼ guideline-directed medical therapy; MRA ¼ mineralocorticoid receptor
antagonist.

P < 0.0001), higher serum creatinine (2.1  1.0 mg/dL visits where a beta-blocker was not prescribed. The
vs 1.6  0.6 mg/dL; P < 0.0001), and higher potassium algorithm identified 64 (32.5%) visits where a beta-
levels (4.3  0.7 vs 4.2  0.5; P ¼ 0.003). The algo- blocker was recommended to be initiated, leaving
rithm did not recommend initiating an ACEI/ARB on 133 (67.5%) visits where the algorithm did not
4,472 (97.9%) visits where it correctly identified the recommend initiating a beta-blocker. The visits
patient as already receiving an ACEI/ARB. The 95 where the algorithm recommended beta-blocker
(2.1%) instances where the patient was on an ACEI/ initiation had significantly higher SBP (124.3  15 vs
ARB and the algorithm recommended it were due to 107.8  16 mm Hg; P < 0.0001) and higher HR (86  13
the patient being on an “other” ACEI/ARB in the vs 77  20 beats/min; P < 0.0001) compared to the
GUIDE-IT data. Of the total visits in which an ACEI/ visits where a beta-blocker was not recommended.
ARB was not prescribed, there were 908 visits where The algorithm recommended the initiation of a beta-
initiation of an ACEI/ARB was able to be assessed at blocker in 40 (0.7%) patient visits due to the patient
the next visit. The algorithm recommended the being on a non-HF beta-blocker (9 atenolol and 31
initiation of an ACEI/ARB for 480 of the 908 visits “other” beta-blockers). Of the total visits in which a
(52.9%). An ACEI/ARB was only initiated on 100 out of beta-blocker was not prescribed, there were 146 visits
the 480 visits (20.8%). where initiation of a beta-blocker could be assessed at
For the GUIDE-IT visits where the patient was the next visit. The algorithm recommended initiating
already prescribed an ACEI/ARB, there were 3,494 a beta-blocker at 51 visits (34.9%), but a beta-blocker
(76.5%) visits where the patient was receiving less was only initiated at 29 (56.9%) of those visits.
than the target ACEI/ARB dose. The algorithm iden- Of the patients that were prescribed a beta-blocker,
tified 1,689 (48.3%) visits where the ACEI/ARB could there were 3,437 (62.1%) visits where the beta-blocker
be up-titrated. The titration recommendations con- dose was less than the target dose. The algorithm
sisted of increasing the ACEI/ARB dose (79%), identified 1,345 (39.1%) visits where the beta-blocker
switching to ARNI (16%), or initiating an ACEI/ARB could be up-titrated. The patients in these visits
(the trial documentation collected “other” as the had a significantly higher SBP (127.8  15 vs
ACEI/ARB) (5%). The patients where the algorithm 109.1  19 mm Hg; P < 0.0001) and HR (84  11 vs
recommended titration had significantly higher SBP 70  13 beats/min; P < 0.0001) compared to the pa-
(127.1  15 vs 103.3  16 mm Hg; P < 0.0001), lower tients where the algorithm did not recommend titra-
serum creatinine (1.3  0.4 vs 1.5  0.7 mg/dL; tion. There were 2,942 visits where the up-titration of
P < 0.0001), and lower potassium levels (4.2  0.4 vs a beta-blocker could be assessed at the next patient
4.5  0.6 mEq/L; P < 0.0001) compared to the visits visit. The algorithm recommended an increase at
where the algorithm did not recommend a dose 1,158 (39.4%) visits, but the dose was only increased
titration. There were 1,073 (23.5%) visits where the at 426 (36.8%) of those visits.
patient was at or above the target dose for the ACEI/ Patients were not prescribed an MRA in 2,492
ARB. The algorithm identified 555 visits where an (43.5%) visits. The algorithm identified 1,668 (66.9%)
ARNI could be initiated. There were 3,025 visits visits where an MRA was recommended to be initi-
where the up-titration of an ACEI/ARB was able to be ated, leaving 824 (33.1%) visits where the algorithm
evaluated at the next visit. The algorithm recom- did not recommend initiating an MRA. The patients
mended up-titration in 1,475 visits (48.7%), while the that were recommended to initiate an MRA had
medications were only up-titrated at 358 (24.2%) of significantly lower serum creatinine (1.4  0.5 vs
those visits. 2.0  1.2 mg/dL; P < 0.0001) and potassium levels
Beta-blocker prescribing in the GUIDE-IT trial was (4.2  0.5 vs 4.6  0.7 mEq/L; P < 0.0001) compared
higher compared to ACEI/ARB with only 197 (3.4%) to the patients that the algorithm did not recommend
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Dorsch et al 7
MAY 2023:100289 HFrEF Medication Optimization Algorithm

F I G U R E 2 The Cumulative Probability for the Primary End Point for a Medication Optimization Score of 25%, 50%, and 75%

(A) The cumulative probability estimated by the Cox model is for the composite end point of cardiovascular death or heart failure hospitalization, the GUIDE-IT (Guiding
Evidence-Based Therapy Using Biomarker Intensified Treatment in Heart Failure) trial’s primary outcome. (B) The cumulative probability estimated by the Cox model is
for the composite end point of all-cause death or hospitalizations, the HF-ACTION (Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training) trial’s
primary outcome, adjusted by the HF-ACTION risk score. MOS ¼ medication optimization score.

initiating an MRA. The algorithm correctly identified associated with a reduced risk of all-cause death or
3,218 (99.3%) visits where the patient was already on hospitalizations (HR: 0.60; 95% CI: 0.44-0.84;
an MRA. There were 22 (0.7%) visits where the algo- P ¼ 0.002). When controlling for the HF-ACTION risk
rithm did not recognize the patient was on an MRA score, the adjusted HR was similar to the unadjusted
since the patient was documented as being on an HR (adjusted HR: 0.60; 95% CI: 0.43-0.83; P ¼ 0.002).
“other” MRA in the GUIDE-IT data. Of the total visits Figure 2 demonstrates the cumulative probability of
in which an MRA was not prescribed, there were the composite endpoint for a MOS value of 25%, 50%,
2,088 patient visits where MRA status could be and 75% in patients needing optimization for both
assessed at the next visit. The algorithm recom- trials. In a sensitivity analysis, assigning all patients
mended the initiation of an MRA in 1,421 (68.1%) taking ARBs to the highest dose did not significantly
visits. Initiation of an MRA only occurred in 224 change the association of the MOS with the primary
(15.8%) of those patient visits. Table 3 summarizes the outcome in HF-ACTION (HR: 0.60; 95% CI: 0.43-0.83;
opportunities for GDMT optimization identified by P ¼ 0.002).
the algorithm. A cut point analysis was performed to classify MOS
CLINICAL OUTCOMES AND THE MOS IN GUIDE-IT percent ranges that could be used to identify patients
AND HF-ACTION. In GUIDE-IT, 69.4% (n ¼ 584) of who benefit from medication optimization using the
patients were identified by the algorithm as needing algorithm. Using the Wald and FDR P values for
medication optimization at baseline. A higher MOS interpretation, MOS of 47% and 75% were identified
was associated with a reduced risk of CV death or as cut points to use to identify patients that would
HF hospitalizations (HR: 0.41; 95% CI: 0.21-0.80; benefit from medication optimization. Figure 3 dem-
P ¼ 0.009). Due to the protocol-defined changes in onstrates the Wald and FDR P values for each MOS
medications over time in the GUIDE-IT trial and value to determine the cut point. See the supplement
clinical variables changing, the MOS was analyzed as for a table, Supplemental Table 1, that includes the
a time-dependent covariate. As a time-dependent P values for each cut level.
covariate, a higher MOS score was associated with a
reduced risk of CV death or HF hospitalizations (HR: DISCUSSION
0.38; 95% CI: 0.21-0.66; P < 0.001).
In HF-ACTION, 72.9% (n ¼ 1,552) of patients were In this study, our guideline-based computable algo-
identified by the algorithm as needing medication rithm accurately selected patients with HFrEF who
optimization at baseline. A higher MOS was were eligible for GDMT initiation or up-titration. Even
8 Dorsch et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

HFrEF Medication Optimization Algorithm MAY 2023:100289

F I G U R E 3 Wald and False Discovery Rate P Values for Each Medication Optimization Score Value

The P values were generated from the cut point analysis in HF-ACTION. The horizontal dotted line represents a P value of 0.05. The vertical
dotted line represents the statistical cut points. Solid dots represent the Cox model’s Wald P values, and open triangles represent the false
discovery rate P values. FDR ¼ false discovery rate; MOS ¼ medication optimization score.

in the GUIDE-IT trial, which specifically focused on 34%, and 30% for ACEI/ARBs, beta-blockers, and
GDMT optimization, the algorithm identified MRAs, respectively. 23 Furthermore, outcomes have
numerous opportunities to improve medical therapy; also been shown to correlate with increasing doses of
these findings were replicated in HF-ACTION. In both ACE inhibitors and beta-blockers. The ATLAS
data sets, the MOS generated by the algorithm was (Assessment of Treatment with Lisinopril and Sur-
prognostically relevant. vival) trial compared the effect of low-dose lisinopril
Using the GUIDE-IT data, our guideline-based (2.5-5 mg daily) to high-dose lisinopril (32.5-35 mg
computable algorithm accurately identified patients daily) on all-cause mortality and CV hospitalizations.
with HFrEF who were eligible for GDMT initiation or While the high dose lisinopril did not improve all-
up-titration based on the available inputs. When cause mortality compared to low dose lisinopril, the
medication optimization was able to be assessed at high dose lisinopril did significantly reduce all-cause
the next visit, there was significant room for GDMT hospitalizations by 13% (P ¼ 0.021) and heart failure
improvement. Our analysis identified missed oppor- hospitalizations by 24% (P ¼ 0.002).9 If all other
tunities for optimal GDMT in 79.2%, 43.1%, and 84.2% medications were optimized, the MOS from this al-
of patient visits when the algorithm recommended gorithm is 78% for a patient on lisinopril 2.5 mg daily
initiation of ACEI/ARB, beta-blockers, or MRAs, and 89% for a patient on lisinopril 40 mg daily. Based
respectively, when the medications were not pre- on our analysis, this 11% change in MOS would lead to
scribed. Additionally, when the algorithm identified about a 9% lower risk of CV death or HF
appropriate up-titration, increasing the dose of an hospitalization.
ACEI/ARB only occurred at 24.3% of patient visits and Beta-blockers have also been shown to have a
36.8% of patient visits for beta-blockers. Finally, the dose-related effect on heart failure outcomes. The
MOS generated by the algorithm also demonstrated HF-ACTION study revealed a 13% relative risk lower
an association with composite primary end points risk of all-cause death or all-cause rehospitalization
adjudicated in 2 HFrEF clinical trials, GUIDE-IT, and when comparing high dose beta-blockers ($25 mg
HF-ACTION. daily carvedilol equivalents) vs low dose beta-
Optimizing heart failure medication confers sig- blockers when adjusting for various factors. 24 If all
nificant morbidity and mortality benefit. Specifically, other medications were optimized, the MOS from this
GDMT has been shown to reduce mortality by 17%, algorithm is 78% for a patient on carvedilol 12.5 mg
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Dorsch et al 9
MAY 2023:100289 HFrEF Medication Optimization Algorithm

twice daily and 89% for a patient on carvedilol 25 mg these medication classes, and will be validated in
twice daily. Based on our analysis, this 11% change in health system data and, when available, clinical trial
MOS would lead to about a 9% lower risk of CV death data sets.
or HF hospitalization.
CONCLUSIONS
With the absence of optimal GDMT in a significant
portion of the HF population and the increase in
The algorithm accurately identified patients with
complexity of HFrEF GDMT with the addition of
HFrEF that have a potential opportunity for GDMT
newer drug classes, clinical decision support systems
initiation and titration. Even in a clinical trial with
(CDSS) have the potential to improve therapeutic
robust protocols, GDMT could have been further
regimens for a larger portion of patients with HFrEF.
optimized in a meaningful number of visits. The
Prior attempts at CDSS in the HF population have
algorithm-generated MOS was associated with a
shown poor results, 25,26 but failed to include all 4
lower risk of clinical outcomes. Implementation into
features shown to predict improved clinical practice
clinical care may identify and address suboptimal
when using CDSS.27 Embedding this HF medication
GDMT in patients with HFrEF.
optimization algorithm into CDSS within or outside
ACKNOWLEDGMENTS The authors would like to
the EHR may allow for clinicians to improve
acknowledge the participants and investigators of the
morbidity and mortality outcomes in patients by fully
GUIDE-IT trial, the participants and investigators of
optimizing HF regimens. The MOS score could be
the HF-ACTION trial, and the NHLBI BioLINCC for
presented in an EHR clinical dashboard for providers
access to the data in this analysis.
to prioritize the patients that need optimization. The
recommendations generated from the algorithm FUNDING SUPPORT AND AUTHOR DISCLOSURES
could be shown to patients before a clinic visit to
facilitate patient-provider communication about Dr Dorsch is supported by R18 HS026874 from the Agency for Health
Research and Quality, R01 AG062582 and R61 HL155498 from the
GDMT optimization, like in the EPIC-HF (Electroni-
National Institutes of Health, and the American Health Association
cally delivered, Patient-activation tool for Intensifi- Health IT Strategically Focused Research Network. Dr Hummel is
cation of Chronic medications for Heart Failure with supported by a grant from Veterans Affairs CARA-009-16F9050 and
reduced ejection fraction) trial, 28 or in an EHR alert, R01 AG062582 and R61 HL155498 from the National Institutes of
Health. Dr Koelling is supported by R01 AG062582 from the National
like in the PROMPT-HF (Pragmatic Trial Of Messaging
Institutes of Health. All other authors have reported that they have no
to Providers Trials Heart Failure) trial.29 This algo- relationships relevant to the contents of this paper to disclose.
rithm could be the basis for many tools used for
GDMT optimization. ADDRESS FOR CORRESPONDENCE: Dr Michael P.
Dorsch, University of Michigan, 428 Church Street,
STUDY LIMITATIONS. There are some limitations to
Ann Arbor, Michigan 48130, USA. E-mail: mdorsch@
our analysis. First, the data for this analysis was from
med.umich.edu.
large clinical trials and cannot be extrapolated to
electronic medical record data at this time. In addi-
tion, ARNI evaluation is included in our algorithm, PERSPECTIVES
but ARNIs were not available to use during the
GUIDE-IT or HF-ACTION trials. Additionally, not all
COMPETENCYIN PATIENT CARE AND PROCEDURAL
patients had follow-up visits to assess the change in
SKILLS: Computable algorithms can accurately identify patients
GDMT in GUIDE-IT. We were only able to analyze the
that have a potential opportunity for GDMT initiation and titra-
change in GDMT in patients that had appropriate
tion. This was shown even in a setting where protocols were in
follow-up visits.
place to optimize GDMT.
The version of the algorithm used in this analysis
also does not include an evaluation of certain drugs
TRANSLATIONAL OUTLOOK: Further implementation of this
(sodium-glucose cotransporter 2 inhibitors and ivab-
algorithm into clinical care should be studied to determine if the
radine) that are more prominent in the recent version
algorithm can assist with GDMT optimization in patients with
of the guidelines or other clinical parameters that
HFrEF.
may influence a provider’s decision to optimize
GDMT. 12 The next version of the algorithm includes
10 Dorsch et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

HFrEF Medication Optimization Algorithm MAY 2023:100289

REFERENCES

1. The cardiac insufficiency bisoprolol study II Society of America. Circulation. 2017;136(6):e137– 22. Lausen B, Schumacher M. Evaluating the effect
(CIBIS-II): a randomised trial. Lancet. e161. of optimized cutoff values in the assessment of
1999;353(9146):9–13. prognostic factors. Comput Stat Data Anal.
12. Heidenreich PA, Bozkurt B, Aguilar D, et al.
1996;21(3):307–326.
2. Packer M, Coats AJ, Fowler MB, et al. Effect of 2022 AHA/ACC/HFSA guideline for the manage-
carvedilol on survival in severe chronic heart fail- ment of heart failure: a report of the American 23. Fonarow GC, Yancy CW, Hernandez AF,
ure. N Engl J Med. 2001;344(22):1651–1658. College of Cardiology/American Heart Association Peterson ED, Spertus JA, Heidenreich PA. Poten-
Joint Committee on Clinical Practice Guidelines. tial impact of optimal implementation of
3. Cohn JN, Johnson G, Ziesche S, et al.
J Am Coll Cardiol. 2022;79(17):e263–e421. evidence-based heart failure therapies on mortal-
A comparison of enalapril with hydralazine-
ity. Am Heart J. 2011;161(6):1024–1030.e3.
isosorbide dinitrate in the treatment of chronic 13. Balas EA, Boren SA. Managing clinical knowl-
congestive heart failure. N Engl J Med. edge for health care improvement. Yearb Med 24. Fiuzat M, Wojdyla D, Pina I, Adams K,
1991;325(5):303–310. Inform. 2000;1:65–70. Whellan D, O’Connor CM. Heart rate or beta-
blocker dose? Association with outcomes in
4. SOLVD Investigators, Yusuf S, Pitt B, Davis CE, 14. Yancy CW, Fonarow GC, Albert NM, et al.
ambulatory heart failure patients with systolic
Hood WB, Cohn JN. Effect of enalapril on survival Adherence to guideline-recommended adjunctive
dysfunction: results from the HF-ACTION trial.
in patients with reduced left ventricular ejection heart failure therapies among outpatient cardiol-
J Am Coll Cardiol HF. 2016;4(2):109–115.
fractions and congestive heart failure. N Engl J ogy practices (findings from IMPROVE HF). Am J
Med. 1991;325(5):293–302. Cardiol. 2010;105(2):255–260. 25. Ansari M, Shlipak MG, Heidenreich PA, et al.
Improving guideline adherence: a randomized trial
5. Effect of metoprolol CR/XL in chronic heart 15. Greene SJ, Butler J, Hellkamp AS, et al.
evaluating strategies to increase beta-blocker use
failure: metoprolol CR/XL randomised intervention Comparative effectiveness of dosing of medical
in heart failure. Circulation. 2003;107(22):2799–
trial in congestive heart failure (MERIT-HF). Lan- therapy for heart failure: from the CHAMP-HF
2804.
cet. 1999;353(9169):2001–2007. registry. J Card Fail. 2022;28(3):370–384.
26. Subramanian U, Fihn SD, Weinberger M, et al.
6. Cohn JN, Tognoni G, Valsartan Heart Failure 16. Yancy CW, Januzzi JL Jr, Allen LA, et al. 2017
A controlled trial of including symptom data in
Trial Investigators. A randomized trial of the ACC expert consensus decision pathway for opti-
computer-based care suggestions for managing
angiotensin-receptor blocker valsartan in chronic mization of heart failure treatment: answers to 10
patients with chronic heart failure. Am J Med.
heart failure. N Engl J Med. 2001;345(23):1667– pivotal issues about heart failure with reduced
2004;116(6):375–384.
1675. ejection fraction: a report of the American College
of Cardiology Task Force on Expert Consensus 27. Kawamoto K, Houlihan CA, Balas EA,
7. Pitt B, Zannad F, Remme WJ, et al. The effect of
Decision Pathways. J Am Coll Cardiol. 2018;71(2): Lobach DF. Improving clinical practice using clin-
spironolactone on morbidity and mortality in pa-
201–230. ical decision support systems: a systematic review
tients with severe heart failure. Randomized
of trials to identify features critical to success.
Aldactone Evaluation Study Investigators. N Engl J 17. Felker GM, Anstrom KJ, Adams KF, et al. Effect
BMJ. 2005;330(7494):765.
Med. 1999;341(10):709–717. of natriuretic peptide–guided therapy on hospi-
talization or cardiovascular mortality in high-risk 28. Allen LA, Venechuk G, McIlvennan CK, et al.
8. Zannad F, McMurray JJV, Krum H, et al. Epler-
patients with heart failure and reduced ejection An electronically delivered patient-activation
enone in patients with systolic heart failure and
fraction. JAMA. 2017;318(8):713. tool for intensification of medications for
mild symptoms. N Engl J Med. 2011;364(1):11–21.
chronic heart failure with reduced ejection frac-
18. O’Connor CM, Whellan DJ, Lee KL, et al. Effi-
9. Packer M, Poole-Wilson PA, Armstrong PW, tion: the EPIC-HF trial. Circulation. 2021;143(5):
cacy and safety of exercise training in patients
et al. Comparative effects of low and high doses of 427–437.
with chronic heart failure: HF-ACTION randomized
the angiotensin-converting enzyme inhibitor, lisi-
controlled trial. JAMA. 2009;301(14):1439–1450. 29. Ghazi L, Yamamoto Y, Riello RJ, et al.
nopril, on morbidity and mortality in chronic heart
Electronic alerts to improve heart failure
failure. ATLAS Study Group. Circulation. 19. O’Connor CM, Whellan DJ, Wojdyla D, et al.
therapy in outpatient practice: a cluster ran-
1999;100(23):2312–2318. Factors related to morbidity and mortality in pa-
domized trial. J Am Coll Cardiol. 2022;79(22):
tients with chronic heart failure with systolic
10. Yancy CW, Jessup M, Bozkurt B, et al. 2013 2203–2213.
dysfunction: the HF-ACTION predictive risk score
ACCF/AHA guideline for the management of heart
model. Circ Heart Fail. 2012;5(1):63–71.
failure: a report of the American College of Car-
diology Foundation/American Heart Association 20. Meyers J, Mandrekar J. Cutpoint determina- KEY WORDS computable knowledge,
Task Force on Practice Guidelines. J Am Coll Car- tion methods in survival analysis using SAS: digital health, health technology, heart
diol. 2013;62(16):e147–239. updated %FINDCUT macro. 2015. Accessed July 8, failure, quality and outcomes, statements
2022. https://www.semanticscholar.org/paper/ and guidelines
11. Yancy CW, Jessup M, Bozkurt B, et al. 2017
d6337150334421959dfe2a2b5a36216aecab213e
ACC/AHA/HFSA focused update of the 2013 ACCF/
AHA guideline for the management of heart fail- 21. Contal C, O’Quigley J. An application of
ure: a report of the American College of Cardiol- changepoint methods in studying the effect of age A PPE NDI X For supplemental tables and
ogy/American Heart Association Task Force on on survival in breast cancer. Comput Stat Data figures, please see the online version of
Clinical Practice Guidelines and the Heart Failure Anal. 1999;30(3):253–270. this paper.
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHOR. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

EDITORIAL COMMENT

Clinical Decision Support Tools for

Optimizing Guideline-Directed Medical
Therapy for Heart Failure
Computing the Possibilities*

Tien M.H. Ng, PHARMD

G uideline-directed medical therapy (GDMT)

for heart failure (HF) with a reduced ejec-
tion fraction confers a substantial benefit
by reducing the risk for hospitalizations and mortality
One relatively naïve area of investigation is the
development of clinical decision support (CDS) tools
specifically for GDMT optimization. Literature of CDS
tools in HF is limited with only 2 recent studies tar-
by more than 75%, 1,2 in addition to positive effects on geting greater use of GDMT, 8-10 while others focused
ventricular function and quality of life. Importantly, on selecting patients for advanced HF therapies 11 and
the impressive benefits are based on several key diagnosis in primary care.12 Application of CDS for
evidence-based tenets. First, GDMT need to be used cardiovascular disease in general has been limited
in combination with the benefit proportional to the and with variable success. 13 Barriers to CDS design
number of therapies and the greatest benefit seen and implementation have been reviewed extensively
when the 4 pillars of renin-angiotensin-aldosterone elsewhere.13-16
system (RAAS) inhibitor, beta-blocker, mineralocorti- In this issue of JACC: Advances, Dorsch et al 17
coid receptor antagonist, and sodium-glucose describe the validation of an application program-
cotransporter-2 inhibitor comprise the regimen.3,4 ming interface computational algorithm for the
Second, the magnitude of benefit exhibits some identification and provision of recommendations for
dose-dependency. Therefore, the guidelines clearly GDMT optimization for HF with reduced ejection
indicate the need for continuous assessment of op- fraction. The algorithm also calculates a medication
portunities to optimize GDMT regimens. 2 optimization score (MOS), providing a percentage of
In reality, prescribing of GDMT remains subopti- how closely a regimen includes all evidence-based,
mal. Real-world data consistently demonstrate an first-line GDMT and proximity to target doses. The
underutilization of GDMT, and when prescribed, the decision-tree utilizes medication name and dose,
minority are at target doses shown to confer maximal New York Heart Association functional classification,
benefits. 5,6 There are many reasons, however, clinical race, allergies, and select readily available clinical
inertia is a major contributor. An analysis of outpa- data (systolic blood pressure, heart rate, serum
tient HF registry data found the majority of patients creatinine, and potassium levels). The algorithm was
fail to have GDMT added or doses uptitrated despite a tested by applying data from 2 clinical trials (GUIDE-
lack of obvious contraindications.7 IT [Guiding Evidence-Based Therapy Using Biomarker
Efforts to improve GDMT optimization have Intensified Treatment in Heart Failure] and ACTION-
included a variety of modalities and interventions. HF [Heart Failure: A Controlled Trial Investigating
Outcomes of Exercise Training]). Major findings were:
1) the algorithm correctly identified non-optimized
*Editorials published in JACC: Advances reflect the views of the authors medication regimens at any specific visit with high
and do not necessarily represent the views of JACC: Advances or the specificity (very few cases of failure to identify a
American College of Cardiology.
particular drug class); 2) the algorithm identified a
From the University of Southern California, Los Angeles, California, USA. significant number of potential opportunities to
The author attests they are in compliance with human studies commit-
optimize GDMT (initiating new therapies or uptitrat-
tees and animal welfare regulations of the author’s institutions and Food
and Drug Administration guidelines, including patient consent where
ing doses); and 3) the MOS was prognostic of the risk
appropriate. For more information, visit the Author Center. of HF hospitalizations and cardiovascular death at

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100354

2 Ng JACC: ADVANCES, VOL. 2, NO. 3, 2023

Clinical Decision Support for HF GDMT MAY 2023:100354

baseline and as a time-dependent covariate. Analyses the relationship between MOS and the relative
of the MOS also confirmed incremental benefit of magnitude of outcome improvement was not exam-
GDMT by number of medications and proximity of ined by specific drug class. Thus, it may be less
dosing to evidence-based target doses. Of note, the helpful in differentiating between multiple GDMT
MOS suggested substantial room for improvement in recommendations when a step-wise approach to
GDMT prescribing despite high rates of baseline use optimization is desired. Refining the MOS, allowing
in prospective studies specifically designed to each therapy decision to be quantified, may aid in
improve HF care. shared decision-making discussions with patients.
The algorithm developed by these authors is The authors should be commended in developing
noteworthy as it satisfies many of the desired an application programming interface algorithm with
characteristics of effective CDS tools. Inputs utilize much potential value. In the time of digital health,
individual patient-level data easily gathered auto- the development of effective tools that utilize EHR
matically through electronic health records (EHRs), it and can be easily integrated into CDS systems re-
can be embedded into an EHR system, it not only mains a high priority. To encourage adoption of these
provides assessment-based prompts but also specific tools, their functionality must be efficient, precise,
recommendations, and recommendations are tied to and recommendations accurate. For optimization of
well-accepted evidence-based guidelines. The out- HF GDMT, incorporation of other clinical, patient,
puts appear modifiable for different stakeholders, and health system factors would be important to
including providers, patients, or health systems for improve precision and accuracy. Utilizing artificial
benchmarking and continuous quality improvement. intelligence and machine learning has advantages
Since the MOS is associated with clinical outcomes, over regression-based computational approaches as
this tool may also have utility for population health if many more factors can be incorporated. 18 Addition of
adaptable to other populations. predictive analytics, such as using EHR data to select
As currently constructed, the algorithm was indeed patients that would benefit or be at risk from further
effective at identifying shortcomings in medication optimization steps, could also enhance accuracy.
regimens, however, there are limitations. As These computational techniques could allow for more
acknowledged by the investigators, validation with individualized recommendations. These technologies
more contemporary clinical trial data that includes could then be tailored to different populations and
angiotensin receptor neprilysin inhibitor and sodium- care settings more easily.
glucose cotransporter-2 inhibitor use and with real- As the care of patients with cardiovascular dis-
world data is needed. The algorithm itself has limi- eases becomes more complex, integration of digital
tations that may hinder implementation. It uses health technologies continues to be increasingly
select clinical factors with fixed thresholds and ap- relevant. Achieving optimal GDMT remains a major
pears to handle inputs as static values. Trends or challenge. The demonstration of a simple EHR
directional changes of lab values and vital statistics computational approach to systematically improving
that may affect clinical decision-making are not HF medication optimization is step in the right di-
accounted for. Taken further, clinical decision- rection. More research is needed to define best
making is much more complex than the factors practices for the design and implementation of
currently included in the tool. Other factors remain computational tools for HF, but the possibilities are
unaccounted for that may influence the decision not within reach.
to optimize certain therapies during certain visits
(logistical, patient, other clinical factors). As such, the
algorithm recommendations may not reflect real-
FUNDING SUPPORT AND AUTHOR DISCLOSURES
world clinical complexity which could contribute to
The author has reported that he has no relationships relevant to the
lack of provider trust, alert fatigue, and other iden- contents of this paper to disclose.
tified barriers to successful CDS system implementa-
tion. 13,14 Importantly, the algorithm currently does
not incorporate a determination of safety where de- ADDRESS FOR CORRESPONDENCE: Dr Tien M.H. Ng,
escalation of therapy may be required (eg, high po- Clinical Pharmacy and Medicine, Titus Department of
tassium does not trigger advice about lowering or Clinical Pharmacy, USC Alfred E. Mann School of
discontinuing RAASi therapy). Pharmacy and Pharmaceutical Sciences, Keck School
The MOS was associated with patient outcome, of Medicine, University of Southern California, 1985
potentially making this metric a useful tool for Zonal Avenue, Los Angeles, California 90089-9121,
benchmarking and improving outcomes. However, USA. E-mail: [email protected].
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Ng 3
MAY 2023:100354 Clinical Decision Support for HF GDMT

REFERENCES

1. Fonarow GC, Hernandez AF, Solomon SD, heart failure therapies in the outpatient setting 13. Lu Y, Melnick ER, Krumholz HM. Clinical deci-
Yancy CW. Potential mortality reduction with (IMPROVE HF). Circulation. 2010;122:585–596. sion support in cardiovascular medicine. BMJ.
optimal implementation of angiotensin receptor 2022;377:e059818.
7. Greene SJ, Fonarow GC, DeVore AD, et al.
neprilysin inhibitor therapy in heart failure. JAMA
Titration of medical therapy for heart failure with 14. Kawamoto K, Houlihan CA, Balas EA,
Cardiol. 2016;1:714–717.
reduced ejection fraction. J Am Coll Cardiol. Lobach DF. Improving clinical practice using clin-
2019;73:2365–2383. ical decision support systems: a systematic review
2. Heidenreich PA, Bozkurt B, Aguilar D, et al.
of trials to identify features critical to success.
2022 AHA/ACC/HFSA guideline for the man- 8. Silva Almodovar A, Nahata MC. Implement-
BMJ. 2005;330:765.
agement of heart failure: a report of the Amer- ing clinical decision support tools and pharma-
ican College of Cardiology/American Heart covigilance to reduce the use of potentially 15. Jaspers MW, Smeulers M, Vermeulen H,
Association Joint Committee on Clinical Practice harmful medications and health care costs in Peute LW. Effects of clinical decision-support
Guidelines. J Am Coll Cardiol. 2022;79(17):e263– adults with heart failure. Front Pharmacol. systems on practitioner performance and patient
e421. 2021;12:612941. outcomes: a synthesis of high-quality systematic
review findings. J Am Med Inform Assoc. 2011;18:
3. Tromp J, Ouwerkerk W, van Veldhuisen DJ, 9. Allen LA, Venechuk G, McIlvennan CK, et al. An
327–334.
et al. A systematic review and network meta- electronically delivered patient-activation tool for
analysis of pharmacological treatment of heart intensification of medications for chronic heart 16. Roshanov PS, Fernandes N, Wilczynski JM,
failure with reduced ejection fraction. J Am Coll failure with reduced ejection fraction: the EPIC-HF et al. Features of effective computerised clinical
Cardiol HF. 2022;10:73–84. trial. Circulation. 2021;143:427–437. decision support systems: meta-regression of 162
randomised trials. BMJ. 2013;346:f657.
10. Ghazi L, Yamamoto Y, Riello RJ, et al. Elec-
4. Greene SJ, Butler J, Hellkamp AS, et al.
tronic alerts to improve heart failure therapy in 17. Dorsch M, Sifuentes A, Cordwin DJ, et al.
Comparative effectiveness of dosing of medical
outpatient practice: a cluster randomized trial. A computable algorithm for medication optimiza-
therapy for heart failure: from the CHAMP-HF
J Am Coll Cardiol. 2022;79:2203–2213. tion in heart failure with reduced ejection fraction.
registry. J Card Fail. 2022;28:370–384.
JACC: Adv. 2023;2:100289.
11. Evans RS, Kfoury AG, Horne BD, et al. Clinical
5. Greene SJ, Butler J, Albert NM, et al. Medical decision support to efficiently identify patients 18. Averbuch T, Sullivan K, Sauer A, et al. Appli-
therapy for heart failure with reduced ejection eligible for advanced heart failure therapies. cations of artificial intelligence and machine
fraction: the CHAMP-HF registry. J Am Coll Car- J Card Fail. 2017;23:719–726. learning in heart failure. Eur Heart J Digit Health.
diol. 2018;72:351–366. 2022;3:311–322.
12. Toth-Pal E, Wardh I, Strender LE, Nilsson G.
6. Fonarow GC, Albert NM, Curtis AB, et al. A guideline-based computerised decision support
Improving evidence-based care for heart failure in system (CDSS) to influence general practitioners
outpatient cardiology practices: primary results of management of chronic heart failure. Inform Prim KEY WORDS clinical decision support, health
the registry to improve the use of evidence-based Care. 2008;16:29–39. technology, heart failure, medications
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ORIGINAL RESEARCH

EMERGING TECHNOLOGIES AND INNOVATIONS

Refined Balloon Pulmonary

Angioplasty in Chronic Thromboembolic
Pulmonary Hypertension
Initial Results of U.S. Regional Program
Riyaz Bashir, MD,a Ali Noory, MD,b Estefania Oliveros, MD,a Carlos Manuel Romero, MD,c Rohit Maruthi, MD,c
Arslan Mirza, MD,d Vladimir Lakhter, DO,a Huaqing Zhao, PHD,e Meredith Brisco-Bacik, MD, MSCE,a
Anjali Vaidya, MD,a William R. Auger, MD,f Paul Forfia, MDa

ABSTRACT

BACKGROUND Balloon pulmonary angioplasty (BPA) is rapidly evolving therapeutic option for patients with nonsur-
gical chronic thromboembolic pulmonary hypertension (CTEPH). There are few U.S. studies that have reported on the
outcomes of this novel therapeutic option.

OBJECTIVES The authors sought to evaluate the efficacy and safety of refined BPA in the treatment of patients with
CTEPH.

METHODS This is a retrospective study of CTEPH patients that underwent BPA. The primary efficacy endpoint was the
change in pulmonary vascular resistance after BPA as compared to baseline and the primary safety endpoint was the rate
of hemoptysis within 24 hours. Secondary endpoints included death, World Health Organization functional class, and 6-
minute walk distance. Logistic regression was used to evaluate factors associated with a hemodynamic and functional
response.

RESULTS A total of 211 BPA sessions were performed on 77 patients (average 2.7
1.7 sessions/patient). After BPA the
mean pulmonary vascular resistance improved by 26% (6.5
3.4 WU to 4.8
2.9 WU, P < 0.001). The mean 6-minute
walk distance improved by 71.7 m (P < 0.001), and WHO functional class improved by 1 functional class (P < 0.001).
There was one death related to reperfusion lung injury. Ten sessions (4.7%) were complicated by hemoptysis. Inde-
pendent factors associated with improved functional and hemodynamic response included preprocedural use of riociguat,
reduce baseline PA compliance, and >3 BPA sessions/patient.

CONCLUSIONS This single center study from the United States showed that BPA with refined techniques in
patients with CTEPH was relatively safe and was associated with significant improvements in pulmonary
hemodynamics and functional capacity. (JACC Adv 2023;2:100291) © 2023 The Authors. Published by Elsevier on
behalf of the American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

From the aDivision of Cardiovascular Disease, Lewis Katz School of Medicine at Temple University, Philadelphia, Pennsylvania,
USA; bDepartment of Medicine, Cardiovascular Institute at Allegheny Health Network, Pittsburg, Pennsylvania, USA; cDepartment
of Medicine, Lewis Katz School of Medicine Temple University, Philadelphia, Pennsylvania, USA; dDepartment of Medicine,
Prairie Heart Institute of Illinois, Springfield, Illinois, USA; eDepartment of Biomedical Education and Data Science, Temple
University, Philadelphia, Pennsylvania, USA; and the fDepartment of Medicine, University of California, San Diego, California,
USA.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received December 9, 2022; revised manuscript received January 26, 2023, accepted February 14, 2023.

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100291

2 Bashir et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Refined Balloon Pulmonary Angioplasty in CTEPH MAY 2023:100291

C
ABBREVIATIONS hronic thromboembolic pulmonary ventilation, which led to poor adoption of BPA in this
AND ACRONYMS hypertension (CTEPH) and chronic country.6 During the past decade in Japan and
thromboembolic pulmonary disease Europe, the technique of BPA has been evolving
BPA = balloon pulmonary
angioplasty
(CTEPD) are obstructive pulmonary vascular including use of pressure wire guidance and under
disorders that affect patients who have expe- sizing balloons in initial sessions, which has been
CTEPD = chronic
thromboembolic pulmonary rienced a pulmonary embolism (PE).1 referred to as “refined BPA”.7 These refinements have
disease Although most PE events resolve with mini- resulted in notable improvements in safety, cardio-
CTEPH = chronic mal residual pulmonary vascular abnormal- pulmonary hemodynamics, as well as functional ca-
thromboembolic pulmonary
ities, there is a subset of patients who go on pacity.8 These results have encouraged the adoption
hypertension
to develop symptomatic pulmonary vascular of this procedure in the United States. However there
PA = pulmonary artery
obstruction due to organized thrombus. If is very limited data from the United States, describing
PE = pulmonary embolism
left untreated, this may result in pulmonary the outcomes of BPA in these patients.9,10
PH = pulmonary hypertension
hypertension and right ventricular dysfunc- In this retrospective study, we evaluated the ef-
PTE = pulmonary tion associated with considerable morbidity fects of BPA on pulmonary vascular resistance (PVR)
thromboendarterectomy
as well as a mortality risk.2,3 and World Health Organization (WHO) functional
PVR = pulmonary vascular
Though the treatment of choice for CTEPH class among all patients who underwent BPA in our
resistance
is surgical pulmonary thromboendarter- tertiary care referral center. We further evaluated the
RV = right ventricle
ectomy (PTE),4 for patients deemed inoper- parameters that were associated with a positive
WHO = World Health
Organization
able or at very high risk for PTE, balloon functional and hemodynamic response to BPA.
pulmonary angioplasty (BPA) is rapidly
evolving as an effective therapeutic option. Recent METHODS
2022 European Society of Cardiology and European
Respiratory Society guidelines have recommended STUDY DESIGN AND PATIENT SELECTION. A retro-
BPA as Class I treatment for patients with inoperable spective chart review was conducted of all patients
disease or those with residual symptomatic CTEPH who underwent BPA after a comprehensive evalua-
after PTE.5 BPA has been studied as a modality of tion by the Temple University Hospital Pulmonary
treatment in such patients since early 2000.6,7 The Hypertension, Right Heart Failure, and CTEPH pro-
early results from the United States showed a high gram within the Temple Heart and Vascular Institute.
rate of major complications including mortality, he- Patients were evaluated by an expert multidisci-
moptysis, and the need for positive pressure plinary CTEPH team, which includes PH/CTEPH

F I G U R E 1 Selective Angiogram of Right Pulmonary Artery Segmental Branches

Selective angiogram of the right pulmonary artery segmental arteries shows distal disease (arrows) in these branches which is not well visualized in the nonselective
angiograms. CTEPH ¼ chronic thromboembolic pulmonary hypertension.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Bashir et al 3
MAY 2023:100291 Refined Balloon Pulmonary Angioplasty in CTEPH

C ENTR AL I LL UST RA TI O N Refined Balloon Pulmonary Angioplasty Technique and Outcomes

Bashir R, et al. JACC Adv. 2023;2(3):100291.

(A) This figure shows a typical pressure waveform (red arrows) as measured by a pressure wire distal to the lesion, within the lesion where the waveform is ven-
tricularized and distal to the lesion after balloon inflation with an undersized balloon and after dilatation with optimal balloon diameter. (B) Shows the effect of BPA
on pulmonary vascular resistance, 6 minute walk distance, B-type natriuretic peptide, and on pulmonary artery compliance. BPA ¼ balloon pulmonary angioplasty.

cardiology experts, interventional cardiology, subgroup analysis to assess the factors associated
cardiothoracic surgery, and diagnostic radiology. Pa- with a positive hemodynamic and/or func-
tients provided informed consent for the procedure, tional response.
and the study was approved by our Institutional Re- PREOPERATIVE EVALUATION. All patients with
view Board. Consecutive patients who underwent CTEPH and CTEPD were presented at a weekly
BPA for CTEPH and CTEPD between August 2015 and multidisciplinary CTEPH conference with extensive
May 2022 were included. The primary efficacy case reviews. History and physical examinations were
endpoint was reduction in PVR and the primary reviewed, including history of previous venous
safety endpoint was the rate of hemoptysis within thromboembolism and pulmonary hypertension (PH)
24 hours of the procedure. Secondary effectiveness medical therapy. Imaging studies reviewed included
endpoints included change in WHO functional class, echocardiography, contrast enhanced lung computed
6-minute walk distance (6MWD) (in meters), change tomography, ventilation/perfusion (V/Q) lung scin-
in cardiac output and index, mean pulmonary artery tigraphy, invasive hemodynamic parameters,
(PA) pressure, PA compliance, and plasma B-type catheter-directed pulmonary angiography, and when
natriuretic peptide (BNP) levels. The secondary safety available, invasive or noninvasive cardiopulmonary
endpoints included any procedure-related adverse exercise testing. The multidisciplinary team then
events including reperfusion lung injury, PA dissec- selected patients for BPA, consisting of patients with
tion without hemoptysis, hemoptysis after 24 hours inoperable CTEPH, those with residual PH post-PTE
of BPA to 30 days, and death. We also performed a or patients who were deemed too high risk for PTE
4 Bashir et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Refined Balloon Pulmonary Angioplasty in CTEPH MAY 2023:100291

catheter is exchanged over a 0.035-inch storq wire for

T A B L E 1 Baseline Characteristics of Total Cohort (N ¼ 77)
a 7-F 70-cm long sheath whose distal tip is positioned
Age (y) 61.9
14.1
in the lobar artery. Through this sheath a 6-F guiding
Female 40 (51.9)
catheter (usually multipurpose or JR-4 for the right
Race
Asian 1 (1.3) PA branches and AL1 for the left PA branches) is used
Black 19 (24.7) to selectively cannulate segmental vessels and a
Hispanic 7 (9.1) baseline angiography is performed (Figure 1). The
White 50 (64.9) location of these selective angiograms is decided
Body mass index (kg/m2) 30.3
8.2 based on the review of the baseline studies including
WHO functional class
computed tomography angiography of chest, V/Q
I 6 (7.9)
scan, and nonselective pulmonary angiograms. Once
II 17 (22.4)
III 35 (46.1)
the lesion is identified, a 300 cm long radi pressure
IV 18 (23.7) wire (Abbott Laboratories) is used to cross the lesion
6 min walk distance (m) 391
248 and assess the distal pressure and waveform. If the
Baseline PVR <4 WU 23 (29.8) exact location of the lesion is not clear angiograph-
Baseline PVR <3 WU 11 (14.2) ically, gradual pullback of the pressure wire is per-
Comorbidities
formed. When the pressure transducer reaches the
History of deep venous thrombosis 32 (41.6)
site of the lesion, the pressure tracing suddenly
History of pulmonary embolism 55 (71.4)
changes into a ventricularized pressure waveform;
May Thurner syndrome 3 (3.9)
Uterine fibroid 6 (7.8)
further pullback proximal to the lesion, will demon-
Pacemaker/venous port 7 (9.1) strate a normal PA pressure waveform similar to the
Splenectomy 6 (7.8) one transduced by the guiding catheter (Central
Hemoglobinopathy 6 (7.8) Illustration). Following this the wire is re-advanced
Coagulopathy 8 (10.5) again into the distal vessel. The BPA is then per-
Venous malformation 4 (5.2)
formed over the pressure wire usually with a small
History of cancer 16 (20.8)
over-the-wire balloon (eg, 2.0  20 mm noncompliant
Atrial fibrillation 6 (7.9)
balloon) and its effect on distal pressure is monitored
Diabetes mellitus 10 (13)
Thyroid disease 10 (13) continuously. We continue to dilate until the mean
Asthma 10 (13) distal pressure reaches a value around 35 mm Hg, 8
Home oxygen use 15 (19.5) there is restoration of pulsatile flow and adequate
Hypertension 34 (44.2) pulmonary venous return is seen. For lesions that
Chronic kidney disease 9 (11.7) cannot be crossed with a pressure wire, we switch
Chronic obstructive pulmonary disease 15 (19.5)
over to 0.014-inch workhorse wires like a Hi-Torque
Liver disease 1 (1.3)
balance middle weight or Runthrough wire. Our
Coronary artery disease 5 (6.5)
Stroke 3 (3.9)
rationale for routinely using the pressure wire in
Sleep disorders 20 (26) these cases included the following: 1) we can accu-
Tobacco use 29 (37.7) rately identify the location of the lesions that are not
Drug abuse 2 (2.6) angiographically well visualized; 2) in patients with
high PA pressures (>35 mm Hg) we keep the pressure
Values are mean
SD or n (%).
PVR ¼ pulmonary vascular resistance; WHO ¼ World Health Organization.
distal to the lesion <35 mm Hg mean, which has been
shown to reduce reperfusion pulmonary edema;8 and
3) we have found the exchange length pressure wire
over a balloon is able to cross two-third of these le-
surgery. All patients had 6MWD, BNP, baseline right
sions, which in turn avoids an added step of
heart catheterization, and selective pulmonary angi-
exchanging out the pressure or support catheter.
ography before BPA. Patients with normal resting
The number of segments that we revascularized in
hemodynamics underwent cardiopulmonary exercise
one session depended on the risks of reperfusion lung
testing prior to BPA.
injury. If patient had very high PVR (>12 WU) we
BPA TECHNIQUE. After obtaining informed consent would limit revascularization to one lobe and not
(which includes a discussion of the unconfirmed treat chronic total occlusions in the first session. On
benefit and risks of hemoptysis and death), a right the other hand, if a patient had low PVR (<6 WU) we
heart catheterization usually via a common femoral will revascularize more than one lobe in a
vein approach was performed using a 7-F Swan Ganz single session. We use Iodixanol 320 (Visipaque, GE
catheter with a 0.035-inch guidewire lumen. The Healthcare), which is diluted with 30% normal saline.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Bashir et al 5
MAY 2023:100291 Refined Balloon Pulmonary Angioplasty in CTEPH

T A B L E 2 Pulmonary Hypertension Medications (N ¼ 77) F I G U R E 2 Number of Sessions per Patient

None 30 (39.0)
Riociguat 30 (39.0)
Sildenafil 12 (15.6)
Macitentan 10 (13.0)
Ambrisentan 3 (3.9)
Tadalafil 5 (6.5)
Parenteral treprostinil 1 (1.3)
Oral treprostinil 1 (1.3)
Selexipag 1 (1.3)
Treprostinil 1 (1.3)
1 PH medication 33 (42.9)
2 PH medications 11 (14.3)
3 PH medications 3 (3.3)

Values are n (%).

PH ¼ pulmonary hypertension.

In patients with abnormal renal function, we would

limit the contrast volume to 3 times the glomerular
filtration rate. We prefer to revascularize lower
This figure shows how many sessions each patient had during the study period. The
lobes first, as the volume of alveolar perfusion is median number of sessions was 2.0 and the average number of sessions was 2.7. Seven
greatest in the lower lobes. All procedures were patients had more than 4 sessions.
done under conscious sedation and supplemental
oxygen to keep oxygen saturations >98%. Outpa-
tient anticoagulation regimen was held prior to BPA
and intravenous heparin is used during the pro-
cedure with a goal ACT of >200. Therapeutic anti- HEMODYNAMIC AND FUNCTIONAL RESPONDERS. In
coagulation was resumed 1 hour after sheath absence of an established definition, we defined
removal unless the patient experienced a bleeding hemodynamic responders as those patients whose
complication. PVR decreased by >10%. Functional responders
POSTPROCEDURAL MANAGEMENT. Following comple- were defined as those patients who improved their
tion of BPA, patients are given a dose of intravenous WHO functional class by $1. We performed multi-
diuretic before they leave the catheterization labora- variate logistic regression to identify factors asso-
tory. Patients were monitored in the cardiac intensive ciated with a hemodynamic and/or functional
care unit during our early experience (from August response.
2015 to December 2019) and after that most of the STATISTICAL AND SENSITIVITY ANALYSIS. Contin-
patients were monitored on the PH and heart failure uous variables were reported as mean
SD and cat-
telemetry floor service barring any intraprocedural egorical variables were reported as percentages.
complications. PH-specific medical therapy and di- Continuous variables were compared using a 2-
uretics were continued as deemed necessary by the sample student’s t-test and categorical variables
PH/CTEPH cardiologists. Chest radiography was ob- were compared using Pearson chi-square test. Uni-
tained the following morning and patients were variate analysis was used to identify factors with a
monitored overnight for hemoptysis, reperfusion hemodynamic and or functional response and if sig-
lung injury, dyspnea, and hypoxia. After discharge, nificant, these variables were then tested in a multi-
patients had follow-up outpatient evaluation with variate logistic regression analysis model. All
our PH/CTEPH cardiology experts. Follow-up outpa- statistical analyses were considered statistically sig-
tient studies included transthoracic echocardiogra- nificant if the P values were <0.05. Sensitivity anal-
phy, V/Q scan, right heart catheterization, and a ysis was performed using multiple imputation (MI) to
6MWD. However, during the peak of the COVID-19 address missing data for follow-up PVR. 11 The MI was
pandemic between 2020 and 2021, many patients performed in the following 3 steps: 1) the missing
were not able to get these tests or right heart cathe- follow-up PVR data (n ¼ 13 [16.9%]) were filled in 100
terization as they were evaluated virtually via video times to generate 100 complete data sets; 2) the 100
telehealth encounter. data sets were analyzed by paired t-tests to compare
6 Bashir et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Refined Balloon Pulmonary Angioplasty in CTEPH MAY 2023:100291

were female. Twenty-one (27.3%) patients had un-

T A B L E 3 Procedural Characteristics (N ¼ 211)
dergone prior PTE, 33 (42.9%) had inoperable CTEPH,
Mean number of BPA sessions per patient 2.7
1.7
21 (27.3%) were determined to be at prohibitively high
Mean number of lesions treated per BPA session 5.03
3.35
surgical risk due to comorbidities, and 2 (2.6%)
No. of patients treated with $3 BPA sessions 35 (45.0)
Mean number of lesions treated per patient 12.8 (9.3) declined PTE. Fifteen (19.2%) patients are still un-
Mean number of CTOs treated per patient 0.9 (1.3) dergoing their BPA treatment course.
Median contrast used per BPA session (cc) 270 (200-330) The remaining 62 patients completed all their BPA
Median fluoroscopy time per session (min) 59 (45-70) sessions, which was determined by having achieved
Median dose of radiation (mGy) 429 (258-628) clinical goals, like WHO functional class 2 or less, PVR
Guide catheters used (in the order of frequency) MP, JR4, and AL1
of <3, right ventricle (RV) function by echocardiog-
Wires used most commonly (in the order of frequency) Radi, BMW, and Runthrough
raphy normal or near normal, or comorbidities that
Mean balloon size (minimum to maximum), mm 2.75
1.2 (1.5-9.0)
would limit any further functional improvement or
Values are mean
SD, n (%), or median (IQR [25th-75th]). patient refused any further BPA sessions.
BMW ¼ balance middle weight; BPA ¼ balloon pulmonary angioplasty; CTO ¼ chronic total occlusions. Forty-eight of the 77 patients (62.3%) were
receiving PH medical therapies, the most common
being riociguat (39.0%), with phosphodiesterase 5
inhibitor (PDE5i) therapy (sildenafil or tadalafil) in
baseline and post-BPA PVR values; and 3) the results
22.1% of patients (Table 2). Twenty-two (28.5%) pa-
from the 100 complete data sets were combined for
tients were on warfarin, 47 (61%) on direct oral anti-
the MI inference. 11 SAS PROC MI and PROC MIANA-
coagulants, and 6 (7.8%) were on enoxaparin. Eleven
LYZE were used for the multiple-imputation analysis.
(14.3%) patients were also taking aspirin; 2 (2.6%)
The software used for all the analysis was SAS version
were taking clopidogrel. Eleven patients had an
9.4 (SAS Institute Inc).
invasive CPET pre-BPA (mild to moderate CTEPH ¼ 7
RESULTS and CTEPD ¼ 4), the findings of which are in the
Supplemental Table 1.
Seventy-seven patients underwent a total of 211 BPA The average number of BPA sessions per patient
sessions at Temple University Hospital between was 2.7
1.7 sessions (Figure 2). The mean number of
August 2015 and May 2022. Baseline characteristics of lesions treated per session was 5.03
3.35 lesions.
the study population are shown in Table 1. The mean Thirty five patients (45.4%) underwent >3 sessions
age of the patients was 62
14.1 years, and 40 (51.9%) and 42 patients had only 1 or 2 BPA sessions. More
than 60% of the lesions were crossed with pressure
wire, with pressure wire measurements being mostly
F I G U R E 3 WHO Functional Class Before and After Balloon Pulmonary Angioplasty qualitative, given the majority of patients had a
pressure waveform distal to the stenosis similar to
pulmonary capillary wedge pressure tracing before
BPA intervention. Following BPA intervention, the
distal pressure waveform would change into a pul-
satile tracing, and it was this pressure we monitored
with progressively increasing balloon sizes, with the
goal to keep distal mean PA pressure <35 mm Hg in
those patients whose mean pulmonary arterial pres-
sure (mPAP) was >35 mm Hg. 8 The balloon sizes
ranged from 1.5 to 9 mm (mean: 2.75
1.2 mm). In the
initial sessions we undersized the balloon just to
restore the flow into an occluded segment and then in
subsequent sessions we used appropriately sized
balloons. We use intravascular ultrasound for balloon
sizing for proximal lesions in the interlobar or truncus
lesions but avoid this for the distal lesions, where we
use visual estimation. Post-PTE patients accounted

This figure shows the improvement in WHO functional class before and after the balloon
for 24.4% (52/211) of all the sessions. Other procedural
pulmonary angioplasty. BPA ¼ balloon pulmonary angioplasty; WHO ¼ World Health characteristics are shown in Table 3.
Organization. The primary endpoint of the change in mean PVR
showed a decrease from 6.5
3.4 WU to 4.8
2.9 WU
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Bashir et al 7
MAY 2023:100291 Refined Balloon Pulmonary Angioplasty in CTEPH

T A B L E 4 Clinical and Hemodynamic Data Before and After Balloon Pulmonary Angioplasty

Baseline Post BPA Change P Value

WHO functional class (n ¼ 65) (n ¼ 65)

Median III II I
Mean
SD 2.85
0.85 1.89
0.92 0.95
0.94 <0.001
I 4 (6.2) 26 (40.0)
II 17 (26.2) 25 (38.5)
III 29 (44.6) 9 (13.8)
IV 15 (23.1) 5 (7.7)
6 min walk distance (m) (n ¼ 40) (n ¼ 40)
354.5
131.4 426.2
124.3 71.7
100.9 <0.001
B-type natriuretic peptide (pg/mL) (n ¼ 48) (n ¼ 48)
Median (IQR, 25-75) 114 (53-235) 40 (21-98.6)
Mean
SD 206.7
197.8 83.5
93.7 123.2
183.2 <0.001
Hemodynamics (n ¼ 65) (n ¼ 65)
Right atrial pressure (mm Hg) 9.0
4.4 9.0
11.9 0.0
12.0 1.00
Systolic PA pressure (mm Hg) 65.8
19.2 57.6
20.8 8.2
16.2 <0.001
Diastolic PA pressure (mm Hg) 24.0
8.0 21.0
7.2 3.0
7.7 0.0028
Mean PA pressure (mm Hg) 41.0
13.6 34.6
11.3 6.4
13.3 <0.001
Pulmonary capillary wedge pressure (mm Hg) 11.9
3.5 11.8
3.9 0.12
4.2 0.81
Pulmonary vascular resistance (WU) 6.5
3.4 4.8
2.9 1.7
2.8 <0.001
Systemic vascular resistance (dynes/s/cm5) 1,299
631 1,364
318 65
631 0.51
Cardiac output (L/min) 4.72
0.98 4.96
1.1 0.24
0.98 0.05
Cardiac index (L/min/m2) 2.43
0.5 2.53
0.4 0.11
0.48 0.07
PA compliance (mL/mm Hg) 1.89
0.9 2.34
1.2 0.45
1.2 0.003

Values are mean
SD or n (%) unless otherwise indicated. Since these are paired tests and the number of pairs that data was available and tested for comparison are shown.
BPA ¼ balloon pulmonary angioplasty; PA ¼ pulmonary artery; WHO ¼ World Health Organization.

(mean change: 1.7
2.8; P < 0.001). The mean 6MWD others were from motor vehicle accident, recreational
improved from 354.5
131.4 m to 426.2
124.3 m drug overdose, severe left heart failure, and 2 were
(P < 0.0001). WHO functional class improved from a from COVID-19. Seven (47%) of the 15 patients who
median of 3 at baseline to 2 after BPA (P < 0.0001), were on home oxygen before the BPA, were able to
Figure 3. The mean BNP level decreased from come off the oxygen after the BPA. There was no
206.7 pg/mL to 83.5 pg/mL (P < 0.0001). PA compli- change in the overall number of PH medications
ance increased by 23.8% from 1.89
0.9 mL/mm Hg to before and after BPA. Three patients were able to
2.34
1.2 mL/mm Hg (increase of 0.45
1.2; P ¼ 0.003) discontinue all their PH medications whereas 3 pa-
(Central Illustration). Cardiac index increased from tients with incomplete BPA’s were treated with one
2.43
0.5 L/min/m 2 to 2.53
0.4 L/min/m 2 (increase additional PH medication (Figure 4).
of 0.11
0.48) which was not statistically significant PROCEDURE-RELATED COMPLICATIONS. Three pa-
(P ¼ 0.07). The mean PA pressure decreased from tients (1.4%) had major procedure-related adverse
41.0
13.6 mm Hg to 34.6
11.3 mm Hg (P < 0.0001) events, which included 1 death related to reperfusion
(Table 4). lung injury, 1 pulmonary hemorrhage, and 1 mild
While follow-up echocardiography data were hemoptysis that was treated with mechanical venti-
limited to a smaller cohort within our study, the lation. Ten BPA sessions (4.7%) were complicated by
available data in 41 patients demonstrated significant hemoptysis, and all occurred at the time of the pro-
improvements in the degree of RV function as quan- cedure (Table 6). The hemoptysis rate was 5.7% (3/52)
tified by tricuspid annular systolic excursion, RV size, in post-PTE patients and it was 4.3% (7/161) in pa-
degree of tricuspid regurgitation, PA systolic pressure tients who did not have prior PTE. Nine patients had
estimation, the degree of systolic interventricular mild hemoptysis of <50 cc by visual estimation and
septal flattening, and the RV base:apex ratio (Table 5). one had major hemoptysis (pulmonary hemorrhage)
The average follow-up duration after the first BPA of >50 cc which required mechanical ventilation and
session was 15.0
17.9 months. Patient overall sur- embolization of the PA branch which was performed
vival rate was 92%. There were 6 deaths in total of immediately by the interventional cardiologists per-
which 1 was related to BPA and CTEPH, while as 5 forming the BPA. The only other patient that required
8 Bashir et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Refined Balloon Pulmonary Angioplasty in CTEPH MAY 2023:100291

capabilities were so severely impaired despite PH

T A B L E 5 Echocardiographic Outcomes
medical therapy that the patient and the multidisci-
Baseline Post BPA P Value plinary CTEPH team opted to pursue BPA. None of the
(n ¼ 41) (n ¼ 41) (Paired t-Test)
patients developed acute kidney injury.
Pulmonary artery systolic pressure (mm Hg) 53.4
20.9 48.4
17.8 <0.001
Left ventricular ejection fraction (%) 61.8
3.6 63.8
4.5 <0.001 FACTORS ASSOCIATED WITH HEMODYNAMIC AND/OR
TAPSE (CM) 1.83
0.34 2.1
0.38 <0.001 FUNCTIONAL RESPONSE. The factors associated with a
Septal flattening 0.003 favorable hemodynamic and/or functional response
None 4 (9.8) 3 (18.8)
to the BPA included the following: use of riociguat
Mild 17 (41.5) 11 (68.8)
prior to BPA, higher number of BPA sessions (for
Moderate 13 (31.7) 2 (12.5)
every additional BPA session the odds ratio of func-
Severe 6 (14.6) 0 (0)
RV size 0.351
tional and/or hemodynamic response was 3.3-fold
Normal 2 (4.9) 8 (50.0) higher), lower baseline PA compliance (an increase of
Mild enlargement 7 (17.1) 1 (6.5) 1 in baseline PA compliance reduces the chances of
Moderate enlargement 26 (63.4) 6 (37.5) being a responder by 55%) (Table 7). The baseline PVR
Severe enlargement 5 (12.2) 4 (6.3) >4.0 was also associated with a positive response;
RV apex to base ratio 0.012
however, it was not statistically significant (Figure 5).
Normal 10 (24.4) 10 (62.5)
On multivariate logistic regression, the independent
Mild 12 (29.3) 5 (31.3)
variables associated with a hemodynamic and/or
Moderate 17 (41.5) 1 (6.3)
Severe 1 (2.4) 0 (0) functional response included the number of BPA
RV function 0.974 sessions the patient had and baseline riociguat
Normal 5 (7.0) 7 (43.8) treatment prior to BPA. The median number of ses-
Mildly dysfunctional 32 (45.1) 8 (50) sions amongst responders was 3.0 and in non-
Moderately dysfunctional 24 (33.8) 1 (6.3) responders was 1.0 BPA sessions, which most likely
Severely dysfunctional 10 (14.1) 0 (0)
reflects the burden of PA obstruction and its role in
RVOT notch N/A
causing patient’s symptoms. The hemodynamic
None 6 (14.6) 12 (80.0)
Late systolic notching 25 (60.9) 1 (6.7)
and/or functional responders had an increase in PA
Mid systolic notching 9 (21.9) 2 (13.3) compliance of 0.5 mL/mm Hg as compared to non-
Tricuspid valve regurgitation 0.035 responders whose PA compliance increased by only
Mild 25 (60.9) 7 (43.8) 0.2 mL/mm Hg after BPA. The variable associated
Moderate 9 (21.9) 8 (50.0) with a poor response was the normal baseline PA
Severe 1 (19.4) 1 (6.3)
compliance. The mean reduction in PVR in hemody-

Values are mean
SD or n (%).

namic responders was 34.3%
18.8% (median 32.8%,
BPA ¼ balloon pulmonary angioplasty; RV ¼ right ventricle; RVOT ¼ right ventricular outflow tract; IQR: 17.3%-48.8%)
TAPSE ¼ tricuspid annular systolic excursion.
SENSITIVITY ANALYSIS AND MISSINGNESS. There
are 65 patients with follow-up PVR and 12 patients
without follow-up PVR. Demographics and clinical
mechanical ventilation experienced mild hemoptysis
baseline characteristics were generally similar be-
and was intubated as a precautionary measure, as this
tween the 2 groups. We addressed the missingness of
was very early in our BPA experience. This patient
the data using MIs method (see the results in
later returned for repeat BPA sessions without any
Supplemental Appendix). The results of our study
subsequent adverse events. Both patients were
were similar by MI.
extubated within 24 hours and discharged home.
None of the patients had hemoptysis after 24 hours of DISCUSSION
BPA. The third serious complication occurred in one
patient that led to death and was related to reperfu- This single center observational study represents one
sion lung injury requiring extracorporeal membrane of the largest, reported experiences of American pa-
oxygenation. This patient had been turned down tients treated with BPA. We found that BPA using
multiple times by the surgical team due to comor- refined techniques was associated with a significant
bidities including a body mass index of 80 kg/m 2, improvement in pulmonary hemodynamic parame-
obesity hypoventilation, reactive airway disease, and ters as defined by a reduction in both PVR and mPAP.
severe physical deconditioning placing her at an un- Patients treated with BPA experienced improvement
acceptably high operative risk. Although her antici- in functional class and exercise capabilities, with
pated BPA risk was also high due to comorbidities, her relatively low rates of procedure-related complica-
hemodynamic status, RV function, and exercise tions. The use of riociguat prior to BPA was associated
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Bashir et al 9
MAY 2023:100291 Refined Balloon Pulmonary Angioplasty in CTEPH

with a beneficial response. A novel finding in our

F I G U R E 4 Pulmonary Hypertension Medications Before and After Balloon Pulmonary
study was that normal baseline PA compliance of Angioplasty
>2.3 mL/mm Hg was associated with lack of hemo-
dynamic and functional response to BPA.
BPA was associated with significant improvements
in invasive hemodynamic parameters with a 26% (1.7
WU) improvement in PVR and a 15% (6.4 mm Hg)
reduction in mPAP. Though these data are similar to
the results from recently published randomized
controlled studies, the degree of improvement in
those studies were larger with regards to reduction in
PVR (26% vs 60%) and mPAP (6.4 mm Hg vs 15 mm Hg
improvement).12-14 This difference may be due to
several reasons, the most important one being the
lower baseline PVR and lesser number of BPA ses-
sions in our cohort. In our cohort, 30% of the subjects
had a PVR <4 WU, while comparative studies have
only included patients with a PVR >4 WU. In the
current study the average number of BPA sessions
was 2.7 as compared 4.7 sessions in multicenter ran-
domized controlled trial based on BPA and 7.7 ses-
sions per patient in the RACE (Rate Control vs
Electrical Cardioversion for Persistent Atrial Fibrilla-
tion Study) trial. 13,14 Other possible factors include This figure shows that the number pulmonary hypertension medications were similar
operator experience, differences in patient popula- before and after the Balloon pulmonary angioplasty. BPA ¼ balloon pulmonary

tion, and differences in inoperability criteria. Addi- angioplasty.

tionally 27% of patients in our cohort were post PTE

patients, a group that has traditionally lower magni-
tude of reduction in PVR.15 and function. Published echo-Doppler outcomes after
Recently, pulmonary arterial compliance (PAC) has BPA are limited. Prior studies have shown these var-
been shown to be an important measure of RV after- iables correlate favorably with clinical and hemody-
load along with PVR, and closely correlates with namic outcomes including survival and functional
prognosis and functional improvement.16-19 This capacity in patients with CTEPH. 21 In this study echo-
study demonstrated an improvement in PAC after Doppler data seem to parallel our PTE experience, 22
BPA, which further supports the hemodynamic as and support the notion that BPA intervention led to
well as functional status efficacy of BPA intervention. physiologically significant improvements in right
This is especially relevant considering that thrombo- heart afterload, and in turn improved RV systolic
embolic disease is often a multicompartmental pul- function and reverse remodeling of the right heart.
monary vascular pathology, affecting both small and BPA treatment was associated with significant im-
large arterial segments, with reduced PAC being pre- provements in WHO functional class and 6MWD
sent in some CTEPH patients despite a lower or
normal resting PVR.20 One novel finding of this study
was that normal PA compliance at baseline was T A B L E 6 Procedural Complications (N ¼ 211)

associated with lack of hemodynamic and functional Death related to BPA 1/77 (1.3)
status benefit. Future studies should consider evalu- Hemoptysis, mild 9/211 (4.2)
ating this parameter prospectively to assess if a Hemoptysis, moderate to severe 1/211 (0.47)
Mechanical ventilation 2/211 (0.95)
normal compliance may suggest futility of BPA in
Use of ECMO 1/211 (0.47)
these patients.
Delayed onset lung injury 1/77 (1.3)
All qualitative echo-Doppler features that reflect a
Acute kidney injury 0/211 (0)
reduction in PVR including RV base:apex ratio, right Major adverse event related to BPA 3/77 (3.8)
ventricular outflow tract pulse wave Doppler notch-
ing, and systolic interventricular septal flattening, Values are n/N (%).
BPA ¼ balloon pulmonary angioplasty; ECMO ¼ extracorporeal membrane
showed improvements in this study. In addition, oxygenation.
there were significant improvements noted in RV size
10 Bashir et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Refined Balloon Pulmonary Angioplasty in CTEPH MAY 2023:100291

T A B L E 7 The Univariate Factors of Hemodynamic and or Functional Response to Balloon Pulmonary Angioplasty

Overall Responder Nonresponders Odds Ratio

(N ¼ 72) (n ¼ 61) (n ¼ 11) P Value (95% CI) Increment

Riociguat use prior to BPA 27 (37.5) 27 (100.0) 0 (0.0) 0.0034 11.6 (2.28–342)
Baseline PVR wood units 6.2
3.3 6.5
3.4 4.3
2.4 0.048 1.85 (1.01–3.41) 2
Baseline PVR >4.0 WU 50 (69.4) 45 (90) 5 (10.0) 0.07 3.38 (0.90–12.59) 2
Number of sessions per patient 2.0 (2.0-4.0) 3.0 (2.0-4.0) 1.0 (1.0-2.0) 0.008 3.33 (1.36–8.13) 1
Baseline PA compliance 1.9
1.0 1.8
0.7 2.8
1.8 0.015 0.45 (0.23–0.86) 1
Baseline PA compliance >2.3 mL/mm Hg 21 (29.3) 15 (71.4) 6 (28.6) 0.053 0.27 (0.07–1.02) 1
No. of segments per patient 8.2
4.4 8.7
4.3 5.5
4.3 0.033 1.78 (1.05–3.04) 3
No. of vessels per patient 12.8
9.3 13.8
9.4 7.7
7.6 0.061 1.93 (0.97–3.84) 7
No. of CTO per patient 0.9
1.3 1.0
1.4 0.4
0.5 0.18 1.72 (0.77–3.86) 1

Values are n (%), mean
SD, or median (IQR).

BPA ¼ balloon pulmonary angioplasty; CTO ¼ chronic total occlusion; PA ¼ pulmonary artery; PVR ¼ pulmonary vascular resistance.

without any change in the number of PH specific physically active patients. In addition almost one-half
medications. There was an improvement in mean of the patients that were on home oxygen were able
6MWD by 71.7 m after BPA, and an improvement in to come off oxygen, which can dramatically improve
mean WHO class by 1 level. These results are consis- the quality of life in these patients. 29
7,8,12,23-28
tent with those of previous studies. The One of the most notable findings of this study is the
improvements we observed in exercise and func- low procedure-related complications with a hemop-
tional capacity after BPA are promising, and at times tysis rate of 4.7% and procedure related mortality of
life-changing, particularly in the young and (1.3%). Other studies have reported a hemoptysis rate
ranging from 6% to 22%. 13,14,30,31 Refinements of
technique have made BPA a safer procedure. 8 The low
rates of adverse events in this study may be related to
F I G U R E 5 Factors Associated With a Functional and Hemodynamic Response
the routine use of pressure wire guidance in our in-
Following Balloon Pulmonary Angioplasty
terventions. Prior studies have shown that treating
target lesions to a goal distal mean PA pressure of
35 mm Hg or less resulted in fewer complica-
tions.8,32,33 Contemporary published experiences in
BPA do not utilize this approach. 7,8,12,23,24,26-28 Our
center attempts to always start with a pressure wire
and use other wires when a 300 cm pressure wire with
a balloon support cannot cross a lesion. We adhere to
this approach, performing multiple staged pressure
wire-guided BPA sessions in close proximity, with
progressive dilatation using larger balloon diameters
during the subsequent sessions. Pretreating patients
medically in an attempt to achieve lower PVR prior to
BPA could be another reason for reduced procedural
complications.14
Prior use of riociguat was a predictor of a favorable
response to BPA in our cohort. This finding is
consistent with the finding of a recently published
randomized controlled RACE trial, 14 and may speak to
This is a Forrest plot showing the log of the odds ratios associated with the functional and both a more robust PH phenotype seen in patients
hemodynamic response after a balloon pulmonary angioplasty in patients with CTEPH. selected to receive riociguat prior to BPA as well as a
In the multivariable logistic regression analysis, the independent variables of this
potential salient effect of medication pretreatment
response were use of riociguat and number of BPA sessions. Another variable associated
with a lack of response was the normal pulmonary artery compliance. BPA ¼ balloon
prior to BPA. In addition, the greater the number of
pulmonary angioplasty; CTO ¼ chronic total occlusion; PA ¼ pulmonary artery; sessions, number of segments, number of vessels
PVR ¼ pulmonary vascular resistance. revascularized, the better the hemodynamic and
functional response.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Bashir et al 11
MAY 2023:100291 Refined Balloon Pulmonary Angioplasty in CTEPH

STUDY LIMITATIONS. First, it is a retrospective, single ACKNOWLEDGMENT The authors would like to
center cohort study with well-known limitations of thank Dr Hiromi Matsubara for his guidance and
this study design. Second, this is an initial report of our mentoring, while we were getting our BPA program
ongoing experience with BPA. The modest PVR started at Temple University Hospital.
reduction seen in this study may be related to few BPA
FUNDING SUPPORT AND AUTHOR DISCLOSURES
sessions per patient and the fact that some of the pa-
tients had there last invasive hemodynamics assessed The study was supported by Temple Heart and Vascular Institute. Dr
prior to the last BPA session. This is corroborated by Bashir is the co-inventor of the Bashir endovascular catheter and has
the significant improvements noted in the functional, equity interest in Thrombolex, Inc; and is supported by a NHLBI grant
under the SBIR grant mechanism. Dr Brisco-Bacik is currently
echocardiographic, and biomarkers observed weeks or
employed by GSK plc. Dr Auger is a consultant for Janssen pharma-
months after the last recorded BPA session. Third, a ceuticals and Neptune Medical; and is an invited speaker for MERCK
portion of our patients have yet to complete their BPA and Bayer. All other authors have reported that they have no re-
treatment course or follow-up studies, with many lationships relevant to the contents of this paper to disclose.

currently due for further BPA sessions. In an attempt to

remedy this, future studies may contain a subgroup ADDRESS FOR CORRESPONDENCE: Prof Riyaz
analysis dedicated to the cohort of patients who have Bashir, Vascular and Endovascular Medicine, Division
completed revascularization in order to best assess the of Cardiovascular Diseases, Department of Medicine,
clinical and hemodynamic effects of BPA. Another Temple University Hospital, 3401 North Broad Street
limitation of this nonblinded study design with follow- (9PP), Philadelphia, Pennsylvania 19140, USA. E-mail:
up at varied time points is the interventionalist’s bias [email protected].
regarding start and end times of revascularizations.
Lastly, our study was conducted during COVID 19 PERSPECTIVES
pandemic which significantly affected our in-person
clinical follow-up of these patients, and therefore,
COMPETENCY IN PATIENT CARE AND PROCEDURAL
missing information on tests like 6MWD, echocardio-
SKILLS: BPA is a rapidly evolving therapeutic option for patients
gram, or invasive hemodynamics.
with nonsurgical CTEPH. There are few U.S. studies that have

CONCLUSIONS reported on the outcomes of this novel therapeutic option using

refined techniques. The BPA with refined techniques in the US

In this single center early BPA experience within the patients with CTEPH was safe and was associated with significant

United States, eligible patients undergoing BPA improvements in pulmonary vascular resistance, right ventricular

experienced significant improvements in PVR, function, and functional capacity.

6MWD, WHO functional class, and BNP levels

TRANSLATIONAL OUTLOOK: To assess comparative effec-
following the treatment. This refined BPA technique
tiveness of refined BPA with PH-specific medical therapy as
was safe and well tolerated with a relatively low rate
compared to PH-specific medical therapy alone in patients with
of complications. In conjunction with medical ther-
nonsurgical chronic thromboembolic pulmonary disease and
apy, BPA could be considered a viable treatment op-
CTEPH using a randomized controlled trial design.
tion for the inoperable CTEPH/CTEPD and recurrent
or persistent PH after PTE.

REFERENCES

1. Humbert M, Kovacs G, Hoeper MM, et al. 2022 International Society for Heart and Lung Trans- results from an international prospective registry.
ESC/ERS guidelines for the diagnosis and treat- plantation (ISHLT). Eur Respir J. 2015;46:903– J Thorac Cardiovasc Surg. 2011;141:702–710.
ment of pulmonary hypertension. Eur Respir J. 975.
5. Humbert M, Kovacs G, Hoeper MM, et al. 2022
2023;61:2200879.
3. Lang IM, Pesavento R, Bonderman D, Yuan JX. ESC/ERS guidelines for the diagnosis and treat-
2. Galie N, Humbert M, Vachiery JL, et al. 2015
Risk factors and basic mechanisms of chronic ment of pulmonary hypertension. Eur Heart J.
ESC/ERS guidelines for the diagnosis and treat-
thromboembolic pulmonary hypertension: a cur- 2022;43:3618–3731.
ment of pulmonary hypertension: the joint task
rent understanding. Eur Respir J. 2013;41:462–
force for the diagnosis and treatment of pulmo- 6. Feinstein JA, Goldhaber SZ, Lock JE,
468.
nary hypertension of the European Society of Ferndandes SM, Landzberg MJ. Balloon pulmonary
Cardiology (ESC) and the European Respiratory 4. Mayer E, Jenkins D, Lindner J, et al. Surgical angioplasty for treatment of chronic thromboem-
Society (ERS): Endorsed by: Association for Euro- management and outcome of patients with bolic pulmonary hypertension. Circulation.
pean Paediatric and Congenital Cardiology (AEPC), chronic thromboembolic pulmonary hypertension: 2001;103:10–13.
12 Bashir et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Refined Balloon Pulmonary Angioplasty in CTEPH MAY 2023:100291

7. Mizoguchi H, Ogawa A, Munemasa M, 17. Dupont M, Mullens W, Skouri HN, et al. Prog- 27. Andreassen AK, Ragnarsson A, Gude E,
Mikouchi H, Ito H, Matsubara H. Refined balloon nostic role of pulmonary arterial capacitance in Geiran O, Andersen R. Balloon pulmonary angio-
pulmonary angioplasty for inoperable patients advanced heart failure. Circ Heart Fail. 2012;5:778– plasty in patients with inoperable chronic throm-
with chronic thromboembolic pulmonary hyper- 785. boembolic pulmonary hypertension. Heart.
tension. Circ Cardiovasc Interv. 2012;5:748–755. 2013;99:1415–1420.
18. Godinas L, Bonne L, Budts W, et al. Balloon
8. Inami T, Kataoka M, Shimura N, et al. Pressure- pulmonary angioplasty for the treatment of non- 28. Kataoka M, Inami T, Kawakami T, Fukuda K,
wire-guided percutaneous transluminal pulmonary operable chronic thromboembolic pulmonary hy- Satoh T. Balloon pulmonary angioplasty (percu-
angioplasty: a breakthrough in catheter- pertension: single-center experience with low taneous transluminal pulmonary angioplasty) for
interventional therapy for chronic thromboem- initial complication rate. J Vasc Interv Radiol. chronic thromboembolic pulmonary hypertension:
bolic pulmonary hypertension. J Am Coll Cardiol 2019;30:1265–1272. a Japanese perspective. J Am Coll Cardiol Intv.
Intv. 2014;7:1297–1306. 2019;12:1382–1388.
19. Akaslan D, Atas H, Aslanger E, et al. Change in
9. Anand V, Frantz RP, DuBrock H, et al. Balloon pulmonary arterial compliance and pulmonary 29. Sakamoto H, Goda A, Tobita K, et al.
pulmonary angioplasty for chronic thromboem- pulsatile stress after balloon pulmonary angio- EmPHasis-10 health-related quality of life and
bolic pulmonary hypertension: initial single-center plasty. Anatol J Cardiol. 2022;26:43–48. exercise capacity in chronic thromboembolic pul-
experience. Mayo Clin Proc Innov Qual Outcomes. monary hypertension after balloon angioplasty.
20. Umemoto S, Abe K, Hosokawa K, et al.
2019;3:311–318. J Am Heart Assoc. 2022;11:e026400.
Increased pulmonary arterial compliance after
10. Mahmud E, Patel M, Ang L, Poch D. Advances balloon pulmonary angioplasty predicts exercise 30. Wang W, Wen L, Song Z, Shi W, Wang K,
tolerance improvement in inoperable CTEPH pa- Huang W. Balloon pulmonary angioplasty vs rio-
in balloon pulmonary angioplasty for chronic
tients with lower pulmonary arterial pressure. ciguat in patients with inoperable chronic throm-
thromboembolic pulmonary hypertension. Pulm
Heart Lung. 2022;52:8–15. boembolic pulmonary hypertension: a systematic
Circ. 2021;11:20458940211007385.
review and meta-analysis. Clin Cardiol. 2019;42:
21. Vaidya A, Golbus JR, Vedage NA, Mazurek J,
11. Rubin DB. Multiple Imputation for Nonresponse 741–752.
Raza F, Forfia PR. Virtual echocardiography
in Surveys. John Wiley & Sons, Inc; 1987.
screening tool to differentiate hemodynamic pro- 31. Gerges C, Friewald R, Gerges M, et al. Efficacy
12. Ogawa A, Satoh T, Fukuda T, et al. Balloon files in pulmonary hypertension. Pulm Circ. and safety of percutaneous pulmonary artery
pulmonary angioplasty for chronic thromboem- 2020;10:2045894020950225. subtotal occlusion and chronic total occlusion
bolic pulmonary hypertension: results of a multi- intervention in chronic thromboembolic pulmo-
22. Raza F, Vaidya A, Lacharite-Roberge AS, et al.
center registry. Circ Cardiovasc Qual Outcomes. nary hypertension. Circ Cardiovasc Interv. 2021;14:
Initial clinical and hemodynamic results of a
2017;10:e004029. e010243.
regional pulmonary thromboendarterectomy pro-
13. Kawakami T, Matsubara H, Shinke T, et al. gram. J Cardiovasc Surg (Torino). 2018;59:428– 32. Kinutani H, Shinke T, Nakayama K, et al. High
Balloon pulmonary angioplasty versus riociguat in 437. perfusion pressure as a predictor of reperfusion
inoperable chronic thromboembolic pulmonary pulmonary injury after balloon pulmonary angio-
23. Wiedenroth CB, Ghofrani HA, Adameit MSD,
hypertension (MR BPA): an open-label, rando- plasty for chronic thromboembolic pulmonary
et al. Sequential treatment with riociguat and
mised controlled trial. Lancet Respir Med. hypertension. Int J Cardiol Heart Vasc. 2016;11:1–6.
balloon pulmonary angioplasty for patients with
2022;10:949–960.
inoperable chronic thromboembolic pulmonary 33. Roik M, Wretowski D, Labyk A, et al. Refined
14. Jais X, Brenot P, Bouvaist H, et al. Balloon hypertension. Pulm Circ. 2018;8: balloon pulmonary angioplasty driven by com-
pulmonary angioplasty versus riociguat for the 2045894018783996. bined assessment of intra-arterial anatomy and
treatment of inoperable chronic thromboembolic physiology–Multimodal approach to treated le-
24. Minatsuki S, Kiyosue A, Kodera S, et al.
pulmonary hypertension (RACE): a multicentre, sions in patients with non-operable distal chronic
Effectiveness of balloon pulmonary angioplasty in
phase 3, open-label, randomised controlled trial thromboembolic pulmonary hypertension–Tech-
patients with inoperable chronic thromboembolic
and ancillary follow-up study. Lancet Respir Med. nique, safety and efficacy of 50 consecutive an-
pulmonary hypertension despite having lesion
2022;10:961–971. gioplasties. Int J Cardiol. 2016;203:228–235.
types suitable for surgical treatment. J Cardiol.
15. Hug KP, Gerry Coghlan J, Cannon J, et al. Serial 2020;75:182–188.
right heart catheter assessment between balloon 25. Brenot P, Jais X, Taniguchi Y, et al. French KEY WORDS balloon pulmonary
pulmonary angioplasty sessions identify proce- experience of balloon pulmonary angioplasty for angioplasty, chronic thromboembolic
dural factors that influence response to treatment. chronic thromboembolic pulmonary hypertension. pulmonary disease, chronic thromboembolic
J Heart Lung Transplant. 2021;40:1223–1234. Eur Respir J. 2019;53:1802095. pulmonary hypertension
16. Mahapatra S, Nishimura RA, Sorajja P, Cha S, 26. Olsson KM, Wiedenroth CB, Kamp JC, et al.
McGoon MD. Relationship of pulmonary arterial Balloon pulmonary angioplasty for inoperable
capacitance and mortality in idiopathic pulmonary patients with chronic thromboembolic pulmonary A PPE NDI X For supplemental materials and
arterial hypertension. J Am Coll Cardiol. 2006;47: hypertension: the initial German experience. Eur table, please see the online version of this
799–803. Respir J. 2017;49:1602409. paper.
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHOR. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

EDITORIAL COMMENT

Balloon Pulmonary Angioplasty for

Chronic Thromboembolic Pulmonary
Hypertension
Here to Stay, More to Learn*

Robert P. Frantz, MD

T he world view of balloon pulmonary angio-

plasty for treatment of inoperable or residual
chronic thromboembolic hypertension has
evolved from skepticism to cautious enthusiasm
ventricular function by echocardiography were also
confirmed. The authors are to be congratulated on
these excellent outcomes. This was achieved within
the construct of a methodical chronic thromboem-
over the past 2 decades. This evolution reflects refine- bolic pulmonary hypertension (CTEPH) team-based
ment of technique, increasing number of case series approach to evaluation, decision-making, and proce-
documenting reasonable safety and success, and dural technique, each element of which is critical to
completion of the randomized clinical trial of balloon successful incorporation of balloon pulmonary an-
pulmonary angioplasty vs riociguat therapy demon- gioplasty into the pathway of CTEPH care. This in-
strating superiority of balloon pulmonary angio- cludes treatment with riociguat prior to embarking on
plasty.1 Yet as the number of centers performing a course of balloon pulmonary angioplasty, particu-
this challenging procedure expands, it remains larly in patients with baseline mean pulmonary artery
incumbent upon the community to continue to report pressure (mPA) >35 mm Hg. The authors emphasize
outcomes and to describe evolving best practices. the consistent use of pressure wires to facilitate
In this issue of JACC: Advances, Bashir et al 2 interrogation and guide graded performance of
describe methods and outcome in a cohort of 77 pa- balloon pulmonary angioplasty. Not all centers follow
tients undergoing balloon pulmonary angioplasty at a this approach, some choosing to maximally dilate
single center in the United States. The findings each lesion while limiting the number of lesions per
confirm reasonable safety; only 3 patients experi- session to reduce the risk of clinically relevant
enced major procedure-related complications, reperfusion edema. Centers choosing the latter
including one death, and hemoptysis occurred in only approach would be well served to benchmark their
4.7% of procedures. Statistically significant improve- results against the excellent safety profile exhibited
ments in 6-minute walk distance, functional class, in the current report.
and brain natriuretic peptide were observed. Hemo- Nonetheless, the findings raise additional unre-
dynamic improvement and improvement in right solved questions. The average extent of improvement
in pulmonary vascular resistance of 26% (1.7 WU) and
reduction in mPA of 15% (6.4 mm Hg) is numerically
less than in some other series. The authors offer
*Editorials published in JACC: Advances reflect the views of the authors
and do not necessarily represent the views of JACC: Advances or the reasonable explanations for this result including on
American College of Cardiology. average less severely impaired baseline hemody-
From the Department of Cardiovascular Medicine, Mayo Clinic College of namics, several patients who only had 1 or 2 sessions,
Medicine and Science, Rochester, Minnesota, USA. follow-up hemodynamics obtained before the last
The author attests they are in compliance with human studies commit-
session of balloon pulmonary angioplasty, many pa-
tees and animal welfare regulations of the author’s institutions and Food
and Drug Administration guidelines, including patient consent where
tients who did not return for additional sessions, and
appropriate. For more information, visit the Author Center. others who were still in the process of completing

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100348

2 Frantz JACC: ADVANCES, VOL. 2, NO. 3, 2023

Balloon Pulmonary Angioplasty for CTEPH MAY 2023:100348

their sessions. Yet the mean residual pulmonary outcome, and counsel the patient accordingly to
vascular resistance was significantly elevated at encourage completion of BPA therapy if this concern
4.8  2.9 WU. In addition, the percentage of patients is present. Further long-term outcome studies will be
treated with vasodilator therapy remained un- important to defining this aspect of optimal man-
changed, which seems disappointing. One would agement. Sufficient hemodynamic improvement to
hope that the need for pulmonary vasodilator therapy allow withdrawal of pulmonary vasodilator therapy is
would on average diminish with a successful course also considered to be a useful goal. 3 Failure for that to
of balloon pulmonary angioplasty therapy, as has occur can either mean untapped opportunity to treat
been shown in other series. 3 The only U.S. Food and additional segments, presence of lesions that are not
Drug Administration approved therapy for inoper- amenable to balloon pulmonary angioplasty, and/or a
able/residual chronic thromboembolic pulmonary particular patient phenotype where vasoconstriction
hypertension is the soluble guanylate cyclase stimu- or other causes of pulmonary hypertension is
lator riociguat, while parenteral subcutaneous tre- continuing to play an important role despite opti-
prostinil is approved for use in the European Union; mally successful technical result. The authors of the
both are significantly expensive and have potential present report do not specifically address their
for side effects. Prior to embarking on a course of BPA, approach to this issue. Systematic efforts to withdraw
it is necessary to discuss with the patient the impor- vasodilator therapy in patients achieving excellent
tance of commitment to completing the course of hemodynamic improvement seem appropriate and
treatment, to optimize opportunity for major benefit. are part of our institution’s approach. On the other
How much is enough? Determining when the course hand, there is potential for clinical worsening with
of balloon pulmonary angioplasty should be declared withdrawal of riociguat, so patient counseling and
complete is a matter of both judgment and debate. systematic reassessment are important. It is hoped
The European Respiratory Society statement on that future reports of balloon pulmonary angioplasty
chronic thromboembolic pulmonary hypertension experiences will systematically incorporate this
cites identification of treatment goals and key end aspect as a component.
points for completing balloon pulmonary angioplasty Predictors of response to balloon pulmonary an-
interventions as a relevant research priority.4 Com- gioplasty included more severe disease (pulmonary
plete revascularization of all amenable lesions is cited vascular resistance >4 WU, more impaired pulmonary
as the general approach for balloon pulmonary an- artery compliance), preoperative use of riociguat
gioplasty in the International Society for Heart and (which may be another marker for more severe dis-
Lung Transplantation consensus statement. 3 Suffi- ease, plus perhaps its role in reducing risk of com-
cient clinical improvement for the patient to declare plications that might discourage additional sessions),
that they are satisfied with their activity tolerance and number of sessions (another marker for extent of
and quality of life is certainly also an important CTEPH plus more complete revascularization). For
metric. Sometimes patients ‘vote with their feet’ once patients with less severe CTEPH and with chronic
they have experienced improvement and may choose thromboembolic pulmonary disease (CTEPD), the
not to return for additional sessions even though not study provides important messages. The demonstra-
all approachable lesions have been treated. The tion of marked rise in pulmonary artery pressure with
extent of improvement in hemodynamics to create invasive cardiopulmonary exercise testing in these
this outcome depends on the patient. For a young patients as reported in the appendix (resting mPA
athlete, optimal treatment of all potential lesions is 27.3  6.0, exercise mPA 47.7  6.0) is helpful in
generally necessary to permit return to relatively illustrating the clinically important relevance of what
unhindered athletic performance. For an older or seems at first view to be mild disease. This fact adds
more sedentary patient, a less aggressive extent of urgency to considering CTEPD in the differential of
treatment leaving greater residual pulmonary hyper- patients presenting with unexplained dyspnea, since
tension may be considered clinically acceptable, resting echo findings may be unimpressive, and the
although presumably the residual pulmonary hyper- diagnosis is easily missed unless appropriate testing
tension will to some degree be associated with less including ventilation/perfusion lung imaging is un-
optimal long-term survival and may result in greater dertaken. An additional important message is the
risk of adverse outcome with subsequent illnesses value of cardiopulmonary exercise testing (noninva-
and surgeries. This needs to be weighed against the sive or invasive) in assessing efficacy of balloon pul-
procedural risk and cost. It is important for the monary angioplasty in patients with mild CTEPH or
clinician to consider whether sufficient residual pul- CTEPD, where ventilatory efficiency and maximal
monary hypertension is present to jeopardize patient oxygen consumption measures are useful. Such
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Frantz 3
MAY 2023:100348 Balloon Pulmonary Angioplasty for CTEPH

follow-up data are unfortunately not consistently continue to advance the field for the benefit of our
obtained. patients.
Balloon pulmonary angioplasty is an exciting,
rewarding, increasingly safe therapy for inoperable or FUNDING SUPPORT AND AUTHOR DISCLOSURES
residual chronic thromboembolic pulmonary hyper-
Dr Frantz has received research funding to his institution from Bayer.
tension. It is anticipated that further refinements in
technique, including evolving approaches to total
occlusions, combined with the commitment of the ADDRESS FOR CORRESPONDENCE: Dr Robert P.
CTEPH community to diligent collection and report- Frantz, Mayo Clinic, 200 First Street, SW, Rochester,
ing of long-term management and outcomes, will Minnesota 55905, USA. E-mail: [email protected].

REFERENCES

1. Jais X, Brenot P, Bouvaist H, et al. Balloon pulmo- boembolic pulmonary hypertension: initial results 4. Delcroix M, Torbicki A, Gopalan D, et al. ERS
nary angioplasty versus riociguat for the treatment of of US regional program. JACC: Adv. 2023;2: statement on chronic thromboembolic pulmo-
inoperable chronic thromboembolic pulmonary hy- 100291. nary hypertension. Eur Respir J. 2021;57(6):
pertension (RACE): a multicentre, phase 3, open-label, 2002828.
3. de Perrot M, Gopalan D, Jenkins D, et al. Evaluation
randomised controlled trial and ancillary follow-up
and management of patients with chronic thrombo-
study. Lancet Respir Med. 2022;10(10):961–971.
embolic pulmonary hypertension: consensus state- KEY WORDS balloon pulmonary angioplasty,
2. Bashir R, Noory A, Oliveros E, et al. Refined ment from the ISHLT. J Heart Lung Transplant. chronic thromboembolic pulmonary hypertension,
balloon pulmonary angioplasty in chronic throm- 2021;40(11):1301–1326. disease management outcome, riociguat
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ORIGINAL RESEARCH

CRITICAL CARE CARDIOLOGY

Treatment Intensity for the Management

of Cardiogenic Shock
Comparison Between STEMI and Non-STEMI

Shashank S. Sinha, MD, MSC,a Mohit Pahuja, MD,b Rachna Kataria, MD,c Vanessa Blumer, MD,d
Jaime Hernandez-Montfort, MD, MPH,e Manreet Kanwar, MD,f A. Reshad Garan, MD,g Yijing Zhang, MA,h
Jeffrey A. Marbach, MBBS, MS,g Adnan Khalif, MD,f Saraschandra Vallabhajosyula, MD, MSC,i
Sandeep Nathan, MD, MSC,j Jacob Abraham, MD,k Borui Li, MA,h Katherine L. Thayer, MPH,h Paulina Baca, MS,h
Fatou Dieng, BA,h Neil M. Harwani, MS,h Michael Y. Yin, MD, MSC,h Anthony J. Faugno, MD,h Haroon A. Faraz, MD,l
Maya Guglin, MD, PHD,m Gavin W. Hickey, MD,n Detlef Wencker, MD,e Shelley Hall, MD,o
Andrew D. Schwartzman, MD,p Wissam Khalife, MD,q Song Li, MD,r Claudius Mahr, DO,r Ju H. Kim, MD,s
Arvind Bhimaraj, MD, MPH,s Van-Khue Ton, MD, PHD,t Esther Vorovich, MD,u Daniel Burkhoff, MD, PHD,v
Navin K. Kapur, MDh

ABSTRACT

BACKGROUND Cardiogenic shock is a leading cause of mortality in patients with acute myocardial infarction.

OBJECTIVES The authors sought to compare clinical characteristics, hospital trajectory, and drug and device use be-
tween patients with ST-segment elevation myocardial infarction-related cardiogenic shock (STEMI-CS) and those without
(non-ST-segment elevation myocardial infarction complicated by cardiogenic shock [NSTEMI-CS]).

METHODS We analyzed data from 1,110 adult admissions with cardiogenic shock complicating acute myocardial
infarction (AMI-CS) across 17 centers within Cardiogenic Shock Working Group. The primary end point was in-hospital
mortality.

RESULTS Our study included 1,110 patients with AMI-CS, of which 731 (65.8%) had STEMI-CS and 379 (34.2%) had
NSTEMI-CS. Most patients were male (STEMI-CS: 71.6%, NSTEMI-CS: 66.5%) and White (STEMI-CS: 53.8%, NSTEMI-CS:
64.1%). In-hospital mortality was 41% and was similar among patients with STEMI-CS and NSTEMI-CS (43% vs 39%,
P ¼ 0.23). Patients with out-of-hospital cardiac arrest had higher in-hospital mortality in patients with NSTEMI-CS (63%
vs 36%, P ¼ 0.006) as compared to patients with STEMI-CS (52% vs 41%, P ¼ 0.16). Similar results were observed for
in-hospital cardiac arrest in patients with STEMI-CS (63% vs 33%, P < 0.001) and NSTEMI-CS (60% vs 32%, P < 0.001).
Only 27% of patients with STEMI-CS and 12% of NSTEMI-CS received both a drug and temporary mechanical circulatory
support device during the first 24 hours, which increased to 78% and 61%, respectively, throughout the course of the
hospitalization (P < 0.001 for both).

CONCLUSIONS Despite increasing use of inotropic and vasoactive support and mechanical circulatory support
throughout the hospitalization, both patients with STEMI-CS and NSTEMI-CS remain at increased risk for in-hospital
mortality. Randomized controls trials are needed to elucidate whether timing and sequence of escalation of support
improves outcomes in patients with AMI-CS. (JACC Adv 2023;2:100314) © 2023 The Authors. Published by Elsevier on
behalf of the American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100314

2 Sinha et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

A Comparison Between STEMI-CS and NSTEMI-CS MAY 2023:100314

C
ABBREVIATIONS ardiogenic shock complicating acute Cardiogenic Shock Working Group (CSWG) registry is
AND ACRONYMS myocardial infarction (AMI-CS) re- a large, robust, multicenter North American registry
mains a morbid and lethal complica- of patients with cardiogenic shock (CS). Herein, we
AMI-CS = cardiogenic shock
complicating acute myocardial
tion, occurring in 5% to 10% of acute leveraged the CSWG registry to compare clinical
infarction myocardial infarction cases, and serving as characteristics, hemodynamic and metabolic pheno-
CA = cardiac arrest the leading cause of in-hospital mortality, types, in-hospital mortality, and elucidate drug- and
CABG = coronary artery bypass ranging from 35% to 50%.1,2 Despite advances device-based utilization between patients with
grafting in revascularization techniques and STEMI-CS and NSTEMI-CS.
CS = cardiogenic shock increasing use of temporary mechanical cir-
CSWG = Cardiogenic Shock culatory support (tMCS) over the past 2 de- METHODS
Working Group cades, randomized clinical trials (RCTs)
IABP = intra-aortic balloon have failed to identify treatment strategies DATA SOURCE. The CSWG registry was initiated in
pump
that improve mortality in AMI-CS. 3-7 Data 2016 with 17 clinical sites across the United States
IHCA = in-hospital cardiac from RCTs in the contemporary era of AMI- contributing patient with CS data. We previously
arrest
CS management have been limited by lack detailed definitions of database parameters and clin-
MAP = mean arterial pressure
of uniformity with shock definitions, focus ical outcomes, and methods for data entry and
NSTEMI = non–ST-segment
on ST-segment elevation myocardial monitoring.12 Briefly, participating sites include
elevation myocardial infarction
infarction–related cardiogenic shock community and university hospitals who contribute
NSTEMI-CS = non–ST-segment
elevation myocardial infarction
(STEMI-CS) and variability of reporting of real-world data to the registry, which includes a
complicated by cardiogenic key outcome predictors among other consid- standardized set of data elements (patient, proce-
shock erations. 8-10 These trials have primarily been dural, and outcomes) that were predefined by CSWG
OHCA = out-of-hospital performed outside the United States, steering committee members with input from prin-
cardiac arrest
traditionally enrolling fewer women and cipal investigators at each site and collected retro-
RCT = randomized clinical trial
lower-risk patients, thus challenging their spectively. Patient demographic, laboratory, and
SBP = systolic blood pressure generalizability in daily practice. Finally, hemodynamic data were collected at a single time
SCAI = Society for there is a paucity of data on clinical charac- point as close to admission as possible, as well as
Cardiovascular Angiography
teristics, hemodynamic profiles, and out- across hospitalization at time points that are clinically
and Interventions
comes in patients with non–ST-segment relevant to patient with CS progression where appli-
STEMI = ST-segment elevation
myocardial infarction elevation myocardial infarction complicated cable including: first vasopressor/inotrope adminis-
STEMI-CS = ST-segment
by cardiogenic shock (NSTEMI-CS).11 Hence, tration, every mechanical circulatory support
elevation myocardial infarction– a better understanding of the contemporary placement, pulmonary artery catheter placement,
related cardiogenic shock patient population with AMI-CS is needed. 24 hours after last device placement, and discharge.
tMCS = temporary mechanical In an effort to bridge knowledge gaps, The diagnosis of CS was physician-adjudicated at
circulatory support
multiple retrospective and prospective na- each site and defined as a sustained episode of at
tional registries have been designed to help delineate least one of the following: systolic blood pressure
AMI-CS phenotypes and inform clinical-decision (SBP) <90 mm Hg for at least 30 minutes, use of
making in patients presenting with AMI-CS. 12-16 The vasoactive agents, cardiac index <2.2 L/min/m 2 in the

From the aInova Heart and Vascular Institute, Inova Fairfax Medical Campus, Falls Church, Virginia, USA; bUniversity of Okla-
homa Health Science Center, Oklahoma City, Oklahoma, USA; cBrown University, Lifespan Cardiovascular Center, Providence,
Rhode Island, USA; dDepartment of Cardiovascular Medicine, Heart and Vascular Institute, Kaufman Center for Heart Failure,
Cleveland Clinic, Cleveland, Ohio, USA; eAdvanced Heart Disease Program, Baylor Scott & White Health, Temple, Texas, USA;
f
Cardiovascular Institute at Allegheny Health Network, Pittsburgh, Pennsylvania, USA; gBeth Israel Deaconess Medical Center,
Boston, Massachusetts, USA; hThe CardioVascular Center, Tufts Medical Center, Boston, Massachusetts, USA; iWake Forest
University School of Medicine, Winston-Salem, North Carolina, USA; jUniversity of Chicago Medicine, Chicago, Illinois, USA;
k
Providence Heart Institute, Center for Cardiovascular Analytics, Research, and Data Science (CARDS), Providence St. Joseph
Health, Portland, Oregon, USA; lHackensack Meridian Health, Hackensack, New Jersey, USA; m
Indiana University Health
Advanced Heart and Lung Care, Indianapolis, Indiana, USA; nUniversity of Pittsburgh Medical Center, Pittsburgh, Pennsylvania,
USA; oBaylor Scott & White Advanced Heart Failure Clinic, Dallas, Texas, USA; pMaine Medical Center, Portland, Maine, USA;
q
University of Texas Medical Branch, Galveston, Texas, USA; rUniversity of Washington Medical Center, Seattle, Washington, USA;
s
Houston Methodist Research Institute, Houston, Texas, USA; tMassachusetts General Hospital, Boston, MA, USA; uNorthwestern
Medicine, Chicago, Illinois, USA; and the vCardiovascular Research Foundation, New York, New York, USA.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received December 27, 2022; accepted February 1, 2023.

JACC: ADVANCES, VOL. 2, NO. 3, 2023 Sinha et al 3
MAY 2023:100314 A Comparison Between STEMI-CS and NSTEMI-CS

absence of hypovolemia, determined to be secondary defined SCAI stages retrospectively at the data coor-
to cardiac dysfunction or use of a tMCS device for dinating center as follows: stage B as patients having
clinically-suspected CS. Treatment of CS was left to either isolated hypoperfusion (lactate 2-5 mmol/L or
the discretion of the clinicians at each center and not alanine aminotransferase (ALT) 200-500 U/L) or hy-
guided by a prescribed algorithm. Quality assurance potension (SBP 60-90 mm Hg or mean arterial pres-
was achieved through adjudication at each site by the sure [MAP] 50-65 mm Hg) without the use of drug or
respective clinical coordinators and principal inves- device therapy. SCAI stage C patients were defined as
tigator. Values were centrally audited and screened having hypoperfusion and hypotension using the
by the CSWG research team for any discrepancies or same criteria as for SCAI stage B or those patients who
major outliers which were resolved with the respec- are being treated for CS with 1 drug (eg, a vasopressor
tive sites. Institutional Review Board approval was or an inotrope) or 1 circulatory support device. For
obtained by individual sites to access these de- the purposes of this analysis, inotropes included
identified data from medical records and patient dobutamine and milrinone. Vasopressors included
consent was not required. Data abstraction and entry norepinephrine, epinephrine, vasopressin, and
into the registry electronic database were performed phenylephrine. SCAI stage D patients were defined as
according to the requirements from individual Insti- having hypotension (SBP 60-90 mm Hg or MAP 50-
tutional Review Boards, which varied from site-to- 65 mm Hg) and hypoperfusion (lactate 5-10 mmol/L or
site. In all cases investigative staff members ALT >500 U/L) while also receiving 2 to 5 drugs
accessed medical records and removed patient iden- or devices. Stage D also included patients on 1 drug or
tifiers. Waivers of informed consent were granted at device with hypotension or hypoperfusion that per-
all sites for data collected retrospectively from elec- sisted despite treatment. SCAI stage E patients were
tronic medical records. Consent was obtained from defined as having hypotension (SBP <60 mm Hg or
patients for all prospective follow up data. MAP <50 mm Hg) or hypoperfusion (lactate
STUDY POPULATION. Between years 2016 and 2020, >10 mmol/L or pH #7.2) or who were receiving more
data from 1,110 adult patient hospital admissions with than 3 drugs and/or 3 devices. All patients who
the diagnosis of AMI-CS were collected. CS cause was experienced out-of-hospital cardiac arrest (OHCA)
reported by each site as due to ST-segment elevation were included in stage E.
myocardial infarction (STEMI) or non–ST-segment Using this modified version of the SCAI staging
elevation myocardial infarction (NSTEMI) based on approach, patients with available data within
previously published consensus definitions. NSTEMI 24 hours of admission (n ¼ 1,110) were stratified ac-
was defined by the presence of acute myocardial cording to SCAI Stages using the first set of values
injury detected by abnormal cardiac biomarkers in acquired. This baseline SCAI stage, based on onset of
the setting of evidence of acute myocardial ischemia CS as adjudicated by the site investigators, was then
without evidence of ST-segment elevation. For mor- analyzed for associations with in-hospital mortality.
tality analyses, individuals with unknown mortality Maximum SCAI stage reached during hospitalization
status at the time of hospital discharge (n ¼ 23, 2.1%) using the same clinical criteria was assigned to each
were excluded. The same inclusion and exclusion patient and analyzed for association with mortality.
criteria were the same for all 1,110 consecutively Lastly, the association between time from admission
enrolled patients included into the registry. The first to maximum SCAI stage and in-hospital mortality was
cohort of 517 patients was enrolled during the first reported. For the purposes of this manuscript, esca-
version of the registry in which data were not avail- lation to refer to intensification of treatment strate-
able to assess changes from hospital admission over gies and transition or progression refers to upward
time. The second cohort of 593 patients was enrolled movement between SCAI stages.
in the second version of the registry, using same in- STATISTICAL ANALYSES. For all the analyses
clusion/exclusion criteria, after additional data ele- described above, continuous variables were reported
ments were added. Accordingly, the current analysis as mean  SD and categorical variables were reported
includes all patients with necessary data available. as frequencies and percentages. The means of nor-
DEFINING PARAMETERS FOR SCAI STAGING. The mally distributed continuous variables were
criteria for Society for Cardiovascular Angiography compared using independent t-tests and medians of
and Interventions (SCAI) stages have been published non-normally distributed variables were compared
previously (reproduced in Supplemental Figure 1).17 using Wilcoxon’s test. Categorical variables were
Based on the recently updated SCAI consensus compared using chi-squared analyses. Relative com-
document, 18 we employed the CSWG interpretation of parisons were reported as odds ratio (95% CI).
4 Sinha et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

A Comparison Between STEMI-CS and NSTEMI-CS MAY 2023:100314

T A B L E 1 Baseline Demographic, Metabolic, and Hemodynamic Characteristics of the Study Population

Total MI-CS (N ¼ 1,110) STEMI-CS (N ¼ 731) NSTEMI-CS (N ¼ 379)

n (%) n (%) n (%)

n Mean  SD n Mean  SD n Mean  SD P Value

Demographic
Non-survivor 449 (40.5) 305 (41.7) 144 (38.0) 0.49
Male 775 (69.8) 523 (71.6) 252 (66.5) 0.23
Race <0.001
White 636 (57.3) 393 (53.8) 243 (64.1)
Asian 53 (4.8) 38 (5.2) 15 (4.0)
Black 42 (3.8) 22 (3.0) 20 (5.3)
Other 23 (2.1) 15 (2.1) 8 (2.1)
Age (y) 1,109 65.6  12.5 730 63.8  12.3 379 69.2  12.3 <0.001
Weight (kg) 593 86.3  22.0 349 87.0  21.1 244 85.4  23.3 0.10
Body mass index (kg/m2) 588 29.4  7.0 346 29.4  6.6 242 29.3  7.5 0.49
Medical history
Hypertension 721 (65.0) 443 (60.6) 278 (73.4) <0.001
Diabetes mellitus 482 (43.4) 280 (38.3) 202 (53.3) <0.001
Atrial fibrillation/atrial flutter 125 (11.3) 66 (9.0) 59 (15.6) <0.001
Chronic kidney disease (any stage) 106 (9.6) 41 (5.6) 65 (17.2) <0.001
Peripheral vascular disease 86 (7.8) 43 (5.9) 43 (11.4) <0.001
Chronic obstructive pulmonary disease 104 (9.4) 52 (7.1) 52 (13.7) <0.001
CVA/TIA 123 (11.1) 57 (7.8) 66 (17.4) <0.001
Valvular disease 80 (7.2) 36 (4.9) 44 (11.6) <0.001
Percutaneous coronary intervention 297 (26.8) 201 (27.5) 96 (25.3) 0.07
Coronary artery bypass grafting 92 (8.3) 48 (6.6) 44 (11.6) 0.003
Implantable cardioverter defibrillator 44 (4.0) 21 (2.9) 23 (6.1) <0.001
Metabolic
Alanine aminotransferase (U/L) 858 270.6  689.5 554 281.7  729.6 304 250.5  610.5 <0.001
Blood urea nitrogen (mg/dL) 1,017 28.7  18.6 651 26.4  17.1 366 32.9  20.4 <0.001
Lactate (mmol/L) 703 4.7  4.3 473 5.2  4.5 230 3.8  3.7 <0.001
Sodium bicarbonate (mEq/L) 819 20.2  5.8 557 19.9  6.1 262 20.9  5.2 <0.001
Serum creatinine (mg/dL) 1,050 1.7  1.3 683 1.7  1.3 367 1.8  1.3 0.03
pH 652 7.3  0.2 463 7.3  0.2 189 7.3  0.1 0.11
Hemodynamic
Left ventricular ejection fraction (%) 803 30.6  16.2 492 30.0  16.1 311 31.6  16.3 0.20
Right atrial pressure (mm Hg) 403 14.2  6.4 281 14.5  6.3 122 13.6  6.7 0.10
PCWP (mm Hg) 300 24.0  9.2 209 23.7  8.8 91 24.7  10.1 0.41
Mean PAP (mm Hg) 439 29.9  9.7 307 29.1  9.9 132 31.7  8.9 0.001
Cardiac output (L/min) 412 4.0  2.1 296 4.0  2.3 116 4.0  1.7 0.83
Cardiac power output (W) 396 0.7  0.4 287 0.7  0.5 109 0.7  0.3 0.57
Heart rate (beats/min) 1,015 91.2  23.4 649 91.5  24.1 366 90.7  22.2 0.62
Cardiac index 415 2.0  0.7 301 2.0  0.7 114 2.1  0.8 0.37
Mean arterial pressure (mm Hg) 1,053 81.5  20.5 685 80.7  21.0 368 82.8  19.5 0.08
Systolic blood pressure (mm Hg) 1,002 109.6  28.4 637 107.7  28.3 365 113.0  28.2 0.004
Pulmonary artery pulsatility index 80 1.7  1.2 54 1.5  0.9 26 2.1  1.7 0.04

All boldface and italics values are statistically significant.

CVA ¼ cerebrovascular accident; MI ¼ myocardial infarction; NSTEMI ¼ non–ST-segment elevation myocardial infarction; PAP ¼ pulmonary artery pressure; PCWP ¼ pulmonary capillary wedge pressure;
STEMI ¼ ST-segment elevation myocardial infarction; TIA ¼ transient ischemic attack.

Univariate logistic regressions were run to test all for all analyses. SAS version 9.4 (SAS Institute Inc,
associations with mortality. In a secondary analysis, Cary, North Carolina) was used for all data analyses.
interaction testing was performed using a logistic
regression model and multivariate logistic re- RESULTS
gressions including all significant univariate pre-
dictors of mortality were performed. Statistical BASELINE CHARACTERISTICS. Our study cohort
significance was determined using an alpha of 0.05 included 1,110 patients with AMI-CS out of which 731
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Sinha et al 5
MAY 2023:100314 A Comparison Between STEMI-CS and NSTEMI-CS

F I G U R E 1 Distribution of Patients With Cardiogenic Shock Stratified by Etiology of Acute Myocardial Infarction and Associated
In-Hospital Mortality with or without Cardiac Arrest

(A) Distribution of patients with cardiogenic shock stratified by etiology of acute myocardial infarction. (B) Unadjusted in-hospital mortality
rate among total MI-CS, STEMI-CS, and NSTEMI-CS cohorts. (C) In-hospital mortality rate among patients with vs Without OHCA among total
MI-CS, STEMI-CS, and NSTEMI-CS cohorts. (D) In-hospital mortality rate among patients with vs without IHCA among total MI, STEMI, and
NSTEMI cohorts. IHCA ¼ in-hospital cardiac arrest; MI-CS ¼ cardiogenic shock complicating myocardial infarction; NSTEMI-CS ¼ cardiogenic
shock complicating non–ST-segment elevation myocardial infarction; OHCA ¼ out-of-hospital cardiac arrest; STEMI-CS ¼ cardiogenic shock
complicating ST-segment elevation myocardial infarction.

(65.8%) had STEMI-CS and 379 (34.2%) had NSTEMI- (CABG) (11.6% vs 6.6%, P ¼ 0.003) but had similar
CS. Baseline characteristics are summarized in rates of percutaneous coronary interventions.
Table 1. The majority of the patients were male Compared to STEMI-CS survivors, patients with
(STEMI-CS: 71.6%, NSTEMI-CS: 66.5%) and White STEMI-CS who died were older (65 vs 62.7 years,
(STEMI-CS: 53.8%, NSTEMI-CS: 64.1%). Compared P ¼ 0.01), had higher prevalence of hypertension
with patients with NSTEMI-CS, patients with STEMI- (65.9% vs 56.5%, P ¼ 0.02), diabetes (46.2% vs 32.6%,
CS were younger (63.8  12.3 vs 69.2  12.3, P ¼ 0.001), chronic kidney disease (8.9% vs
P < 0.001). Patients with NSTEMI-CS had higher 3.4%, P ¼ 0.002), peripheral vascular disease (6.6% vs
prevalence of hypertension (73.4% vs 60.6%, 5.1%, P ¼ 0.01), prior history of CABG (9.2% vs 4.6%,
P < 0.001), diabetes (53.3% vs 38.3%, P < 0.001), P ¼ 0.049), and lower left ventricular ejection fraction
chronic kidney disease (17.2% vs 5.6%, P < 0.001), (26.5% vs 32.6%, P < 0.001) as shown in Supplemental
peripheral vascular disease (11.4% vs 5.9%, Table 1A. A comparison of survivors vs non-survivors
P < 0.001), stroke (cerebrovascular accident or tran- in patients with NSTEMI-CS are described in
sient ischemic attack) (17.4% vs 7.8%, P < 0.001), and Supplemental Table 1B. Non-survivors were older
prior history of coronary artery bypass grafting (71.2 vs 67.8 years, P ¼ 0.005) and had higher
6 Sinha et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

A Comparison Between STEMI-CS and NSTEMI-CS MAY 2023:100314

F I G U R E 2 Association of In-Hospital Mortality Across Baseline and Maximum SCAI Stage Among Total MI-CS, STEMI-CS, and NSTEMI-CS Patients

(A) Bar graphs illustrating the association of in-hospital mortality rate across baseline SCAI stages among total MI-CS, STEMI-CS, and NSTEMI-CS cohorts. (B) Bar
graphs illustrating the association of in-hospital mortality rate across maximum SCAI stages among total MI-CS, STEMI-CS, and NSTEMI-CS cohorts.
MI-CS ¼ cardiogenic shock complicating myocardial infarction; NSTEMI-CS ¼ cardiogenic shock complicating non–ST-segment elevation myocardial infarction;
SCAI ¼ Society for Cardiovascular Angiography and Interventions; STEMI-CS ¼ cardiogenic shock complicating ST-segment elevation myocardial infarction.

prevalence of atrial fibrillation/flutter (21.5% vs 11.5%, namely serum creatinine (1.8 vs 1.7 mg/dL, P ¼ 0.03),
P ¼ 0.03). were worse in patients with STEMI-CS as compared to
patients with NSTEMI-CS. Non-survivors among both
CLINICAL OUTCOMES. Overall unadjusted in-
patients with STEMI-CS and NSTEMI-CS were noted
hospital mortality was 41%. Mortality was similar
to have higher lactate level (STEMI-CS: 6.6 vs 3.9,
among patients with STEMI-CS and NSTEMI-CS (43%
P < 0.001; NSTEMI-CS: 4.6 vs 3.2 mmol/L, P < 0.001),
vs 39%, P ¼ 0.23) (Figure 1B). Cardiac arrest (CA) was
ALT levels (STEMI-CS: 377.9 vs 201.4 U/L, P < 0.001;
an effect modifier that was associated with worse
NSTEMI-CS: 297.6 vs 223 U/L, P ¼ 0.13), and serum
prognosis. Patients with OHCA had higher in-hospital
creatinine (STEMI-CS: 1.9 vs 1.4, P < 0.001; NSTEMI-
mortality (55% vs 39%, P ¼ 0.002), but this was more
CS: 2.1 vs 1.6 mg/dL, P < 0.001) (Supplemental
prominently seen among patients with NSTEMI-CS
Tables 1A and 1B). Among the hemodynamic param-
(63% vs 36%, P ¼ 0.006) as compared to patients
eters in patients with STEMI-CS, non-survivors had
with STEMI-CS (52% vs 41%, P ¼ 0.16). Similar results
higher right atrial pressure (16.3 vs 13.3, P < 0.001)
were observed for patients with in-hospital cardiac
and lower pulmonary artery pulsatility index (1.3 vs
arrest (IHCA) (62% vs 33%, P < 0.001) with signifi-
1.7, P ¼ 0.03).
cantly high mortality in both patients with STEMI-CS
Supplemental Tables 2 and 3 summarize the labo-
(63% vs 33%, P < 0.001) and NSTEMI-CS (60% vs 32%,
ratory and hemodynamic parameters for patients
P < 0.001) (Figures 1C and 1D). We also performed
with STEMI-CS and NSTEMI-CS at baseline and
interaction testing using a logistic regression model
maximum SCAI stages, respectively. Lactate levels
of OHCA and etiology of MI (ie, STEMI-CS, NSTEMI-
were higher in patients with SCAI stage D and E CS in
CS) predicting in-hospital mortality. The result
both patients with STEMI-CS and NSTEMI-CS at
turned out to be not significant for the interaction
baseline and maximum SCAI stage. At baseline, ALT
(P ¼ 0.12), thereby suggesting the difference observed
was highest in stage D CS for both STEMI-CS and
between the status of OHCA is consistent over etiol-
NSTEMI-CS whereas serum creatinine increased from
ogy. We performed a similar analysis for IHCA, and
stage B to stage E for STEMI-CS, but in the NSTEMI-CS
noted the interaction is also not significant (P ¼ 0.89).
cohort, it was found to be highest in stage D CS.
HEMO-METABOLIC AND HEMODYNAMIC PROFILES. The in-hospital mortality rate across baseline and
Laboratory markers for perfusion such as lactate maximum SCAI stages for total myocardial infarction
levels were higher in STEMI-CS as compared to pa- complicated by cardiogenic shock, STEMI-CS, and
tients with NSTEMI-CS (5.2 vs 3.8 mmol/L, P < 0.001) NSTEMI-CS is shown in Figures 2A and 2B, respec-
(Table 1). Similarly, laboratory parameters for end- tively. Notably, the SCAI C patients with STEMI were
organ dysfunction such as liver dysfunction, ALT associated with the lowest mortality at baseline (19%)
(281 vs 250.5 U/L, P < 0.001), and renal dysfunction, and at maximum (4%) across all SCAI stages.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Sinha et al 7
MAY 2023:100314 A Comparison Between STEMI-CS and NSTEMI-CS

F I G U R E 3 Progression to Advanced SCAI Stages Among Total MI-CS, STEMI-CS and NSTEMI-CS Patients

* Baseline Maximum
A 100%
*
*
C
88% 89% Total MI-CS
86%

B B C D E
90%
STEMI-CS
80% 72%
71% 69% NSTEMI-CS
70%
Escalation (%)

60% Total MI-CS 65 (28%) 8 (13%) 6 (17%) 42 (19%) 9 (89%)

50% STEMI-CS 37 (30%) 4 (25%) 2 (0%) 24 (17%) 7 (86%)
NSTEMI-CS 28 (25%) 4 (100%) 4 (25%) 18 (22%) 2 (100%)
40%
28%
30% 23%

C C D E
B
20% 16%

10%
0%
B C D Total MI-CS 41 (20%) 12 (17%) 23 (9%) 6 (67%)
Baseline SCAI Stage STEMI-CS 16 (19%) 5 (0%) 9 (11%) 2 (100%)
NSTEMI-CS 25 (20%) 7 (29%) 14 (7%) 4 (50%)
B

C
250

D D E
B
*
Mean Time to Maximum SCAI Stage (hour)

*
204
200 Total MI-CS
178
STEMI-CS Total MI-CS 145 (47%) 112 (34%) 33 (91%)

150
NSTEMI-CS
STEMI-CS
NSTEMI-CS
82 (49%)
63 (44%)
59 (32%)
53 (36%)
23 (91%)
10 (90%)
Stage
116
n-number

C
E
100 (% Mortality)
59
45
50 35 38
31 26 Total MI-CS 197 (56%)
STEMI-CS 143 (57%)
NSTEMI-CS 54 (56%)
0
B C D
Baseline SCAI Stage

(A) Progression rate from baseline to maximum SCAI stages among total MI-CS, STEMI-CS, and NSTEMI-CS cohorts. (B) The mean time to progression from baseline to
maximum SCAI stages among total MI-CS, STEMI-CS, and NSTEMI-CS cohorts. (C) Illustration showing trajectory of SCAI stages during hospitalization among total
MI-CS, STEMI-CS, and NSTEMI-CS cohorts. The number of patients and percent in-hospital mortality for the maximum Society for Cardiovascular Angiography and
Interventions stage achieved stratified by baseline Society for Cardiovascular Angiography and Interventions stage is shown. MI-CS ¼ cardiogenic shock complicating
myocardial infarction; NSTEMI-CS ¼ cardiogenic shock complicating non–ST-segment elevation myocardial infarction; SCAI ¼ Society for Cardiovascular Angiography
and Interventions; STEMI-CS ¼ cardiogenic shock complicating ST-segment elevation myocardial infarction.

PROGRESSION FROM BASELINE TO MAXIMUM SCAI DRUG AND DEVICE UTILIZATION. The evolution of
SHOCK STAGE. Figures 3A and 3B summarize the tMCS device and drug utilization based on the first
overall progression of SCAI stages of patients with 24 hours and throughout the course of the hospitali-
STEMI-CS and NSTEMI-CS patients and the average zation is presented in Figure 4. Only 27% of patients
time to maximum SCAI stage during the course of the with STEMI-CS and 12% of NSTEMI-CS received both a
index hospitalization. Among patients with SCAI drug and tMCS device during the first 24 hours of the
stage B CS, 89% of patients with STEMI-CS and 86% of hospitalization, which increased to 78% and 61%,
patients with NSTEMI-CS transitioned to a higher respectively, throughout the course of the hospitali-
stage with average time to maximum SCAI stage of 35 zation. Conversely, 54% of all patients with STEMI-CS
and 59 hours, respectively. Similarly, among patients and 58% of patients with NSTEMI-CS received neither
presenting with stage C CS, 69% of patients with a drug nor tMCS device within the first 24 hours,
STEMI-CS and 72% of NSTEMI-CS transitioned to which decreased to 2% for both cohorts during the
higher stages with average time to maximum stage of hospitalization. Vasopressors (STEMI-CS: 21%;
38 and 26 hours, respectively. Among patients NSTEMI-CS: 20%), followed by vasopressors and
initially presenting in stage D CS, 28% of STEMI-CS inotropes (STEMI-CS: 11%; NSTEMI-CS: 6%), were the
and 16% of NSTEMI-CS progressed to stage E with most common drugs employed during the first
average time to progression of 204 and 116 hours, 24 hours (Supplemental Figure 3A). These were also
respectively. The highest in-hospital mortality was the most frequent drugs, or combinations thereof,
seen in patients presenting with baseline SCAI stage E that were employed in both cohorts during the hos-
or those progressing to maximum SCAI stage E. Pa- pitalization. Within 24 hours, the most common tMCS
tients that presented with baseline stage D CS were devices used were intra-aortic balloon pump (IABP)
noted to have higher mortality than those who tran- alone (STEMI-CS: 13%; NSTEMI-CS: 12%) and Impella
sitioned to the maximum SCAI stage D (Figure 3C). alone (STEMI-CS: 12%; NSTEMI-CS: 5%)
8 Sinha et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

A Comparison Between STEMI-CS and NSTEMI-CS MAY 2023:100314

F I G U R E 4 Drug and Device Usage Distribution Within 24 Hours and Throughout Hospitalization in STEMI-CS vs Patients With NSTEMI-CS

ns ¼ not statistically significant; NSTEMI-CS ¼ cardiogenic shock complicating non–ST-segment elevation myocardial infarction; STEMI-CS ¼ cardiogenic shock
complicating ST-segment elevation myocardial infarction.

(Supplemental Figure 3B). Throughout the hospitali- modifier, portending significantly worse mortality in
zation, however, there was a broader distribution of both patients with STEMI-CS and NSTEMI-CS. Sec-
device use. Among patients with STEMI-CS, 29% had ond, the majority of SCAI stage B and SCAI stage C
IABP, 28% 2 or more devices (including VA ECMO), patients with STEMI-CS and NSTEMI-CS progress
and 22% Impella alone. Among patients with during their hospitalization to SCAI stage D or E;
NSTEMI-CS, 28% used IABP, 21% Impella alone, and transition to a higher SCAI stage is associated with
11% 2 or more devices (including veno-arterial extra- worse in-hospital outcomes as compared to those
corporeal membrane oxygenation). who initially presented in SCAI E. Patients
withNSTEMI-CS SCAI B required a higher mean time
DISCUSSION to maximum SCAI stage as compared to patients
with STEMI-CS SCAI B; highlighting potential op-
To the best of our knowledge, this is one of the portunities for earlier recognition and perhaps more
largest, contemporary, real-world registry-based timely intervention in these patients. Finally, a
analyses characterizing AMI-CS specifically broad distribution of drug and device exposure oc-
comparing STEMI-CS and NSTEMI-CS across key curs within the first 24 hours and during the course
clinical characteristics, hemodynamic and metabolic of the hospitalization for both patients with STEMI-
parameters, in-hospital mortality, as well as drug- CS and NSTEMI-CS, which provides insights into the
and device-based strategies (Central Illustration). strategies for escalation of support. During the first
The main findings from this study are: first, in- 24 hours, patients with STEMI-CS were more likely
hospital mortality remains high (approximately to receive a drug and device, and patients with
40%-45%) across both STEMI-CS and NSTEMI-CS NSTEMI-CSwere more likely to receive drug only.
patients, suggesting that NSTEMI-CS patients also Our study provides an important contribution to
represent a vulnerable, high-risk cohort. Further- the rapidly evolving literature, as patients with
more, CA, irrespective of in-hospital and out-of- NSTEMI-CS have not been well studied in prior RCTs.
hospital presentation, is an adverse effect Approximately 17% of patients in the SHOCK (Should
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Sinha et al 9
MAY 2023:100314 A Comparison Between STEMI-CS and NSTEMI-CS

C ENTR AL I LL UST RA TI O N Comparison of ST-Segment-Elevation (STEMI-CS) and Non-ST-Segment-Elevation

(NSTEMI-CS) Myocardial Infarction Complicated by Cardiogenic Shock

Sinha SS, et al. JACC Adv. 2023;2(3):100314.

We sought to compare in-hospital mortality, clinical trajectories, and drug and device use between patients with ST-segment elevation myocardial infarction–related
cardiogenic shock (STEMI-CS) and those without (non–ST-segment elevation myocardial infarction complicated by cardiogenic shock [NSTEMI-CS]). In-hospital
mortality was similar among patients with STEMI-CS and NSTEMI-CS. Patients with out-of-hospital cardiac arrest had higher in-hospital mortality in patients with
NSTEMI-CS as compared to patients with STEMI-CS. Similar results were observed for in-hospital cardiac arrest in patients with STEMI-CS and NSTEMI-CS. Only 27%
of patients with STEMI-CS and 12% of NSTEMI-CS received both a drug and temporary mechanical circulatory support device during the first 24 hours, which
increased to 78% and 61%, respectively, throughout the course of the hospitalization. AMI-CS ¼ cardiogenic shock complicating acute myocardial infarction;
IHM ¼ in-hospital mortality; MI ¼ myocardial infarction; NSTEMI ¼ non–ST-segment elevation myocardial infarction; SCAI ¼ Society for Cardiovascular Angiography
and Interventions; STEMI ¼ ST-segment elevation myocardial infarction.
10 Sinha et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

A Comparison Between STEMI-CS and NSTEMI-CS MAY 2023:100314

We Emergently Revascularize Occluded Coronaries analyzed separately from patients with CS without CA
for Cardiogenic Shock) trial and 30% of patients in the in future clinical trials, given confounding previously
IABP-SHOCK (Intra-Aortic Balloon Pump in Cardio- reported in subgroup analyses of IABP-SHOCK II and
genic Shock) II trial presented with NSTEMI-CS.5,19 CULPRIT-SHOCK (Culprit Lesion Only Percutaneous
Patients with NSTEMI-CS are often heterogeneous Coronary Intervention Versus Multivessel Percuta-
reflecting a spectrum of cardiovascular risk. In the neous Coronary Intervention in Cardiogenic Shock).
The Global Utilization of Streptokinase and Tissue The present analysis also extends our prior work
Plasminogen Activator for Occluded Coronary Ar- employing the CSWG interpretation of the revised
teries (GUSTO)-IIb Registry, eg, patients with SCAI staging classification. 31 Both STEMI-CS and
NSTEMI-CS had later onset of CS than patients with NSTEMI-CS SCAI stage B (approximately 90% in both)
STEMI-CS, had more extensive coronary artery dis- and C (approximately 70% in both) patients transi-
ease, and more recurrent ischemia and infarction tioned to a higher SCAI stage during the course of
before developing shock compared with patients with their index hospitalization. Interestingly, patients
STEMI-CS. 20 The last American College of Cardiology with NSTEMI-CS SCAI B required a higher mean time
and American Heart Association Guidelines for the to maximum SCAI stage as compared to patients with
management of STEMI and NSTEMI were published in STEMI-CS SCAI B; the converse appeared to be pre-
2013 and 2014, respectively, and merit a timely up- sent for patients with STEMI-CS SCAI C as compared
date.21-24 More recently published American Heart to patients with NSTEMI-CS. When transition to SCAI
Association/American College of Cardiology/Heart E did occur, it was associated with worse outcomes
Failure Society of America (HFSA) guidelines on then de novo presentations in SCAI E perhaps due to
Heart Failure do not specifically address STEMI-CS or delay in recognition of onset of worsening or re-
NSTEMI-CS. 25-27 Thus, our findings highlight that fractory hemo-metabolic shock and multiorgan fail-
patients with NSTEMI-CS are an important high-risk ure. While these results are hypothesis-generating
cohort that merit inclusion in future clinical trials in given that the retrospective observational dataset
CS to determine the optimal treatment strategies in was not powered to adequately detect differences
these patients. among these subgroups, they highlight key clinical
CA remains a very important effect modifier of CS. questions that should inform future analyses. Spe-
Indeed, both OHCA and IHCA were associated with cifically, given the dynamic nature of SCAI stages,
higher in-hospital mortality in both STEMI-CS and clinicians must perform serial re-assessments to
NSTEMI-CS. OHCA was more prominently seen in the ascertain whether the maximum SCAI stage has been
NSTEMI-CS cohort, which likely represents a hetero- achieved when managing these patients with CS in
geneous syndrome that encompasses a broad spec- the cardiac intensive care unit.
trum of clinical presentations and hemo-metabolic Another novel and important finding from our
derangements due to acute coronary syndrome not analysis is the detailed understanding of drug and
involving acute plaque rupture. Our results are device exposure in a contemporary real-world CS
consistent with prior work,28 including that of Val- multicenter cohort. To the best of our knowledge,
labhajosyula et al,29 who have previously shown us- ours is the first detailed analysis of both drug and
ing National Inpatient Sample data collected from device utilization, both within the first 24 hours of
2000 to 2017 that the combined CS and CA cohort admission as well as throughout the hospitalization
have higher rates of multiorgan failure and in- for both patients with STEMI-CS and NSTEMI-CS.
hospital mortality as compared to CS only and CA Within the first 24 hours of the hospitalization, it is
only, respectively, in patients with STEMI-CS. Our remarkable that more than 50% of both patients with
analysis extends this work in a more contemporary STEMI-CS and NSTEMI-CS receive neither drug nor
cohort illustrating the impact on both STEMI-CS and device treatment. During this interval, patients with
now including patients with NSTEMI-CS. Matching STEMI-CS were more likely to receive a drug and
the right therapy to the right patient at the right time device (27% vs 12%) and patients with NSTEMI-CS
in this high-risk subgroup is especially challenging were more likely to receive drug only (22% vs 15%).
given the competing risk of death due to neurologic Perhaps not surprisingly, this distribution of drug and
injury.30 Nonetheless, our data supports the bur- device use evolved significantly for both patients
geoning evidence that patients with STEMI and with STEMI-CS and NSTEMI-CS during the course of
NSTEMI with both CS and CA represent a unique high- the index hospitalization. In particular, 98% of both
risk cohort with salient differences with respect to patients with STEMI-CS and NSTEMI-CS had exposure
pathophysiology, multiorgan sequelae, and causes of to either a drug or device or combination thereof
death. As noted by Jentzeret al, 30 these should be during the course of the hospitalization; notably, 78%
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Sinha et al 11
MAY 2023:100314 A Comparison Between STEMI-CS and NSTEMI-CS

of all patients with STEMI-CS were treated with a type and timing of revascularization (ie, percuta-
combination of drug and device as compared to 61% neous coronary intervention vs CABG) and tMCS (ie,
of patients with NSTEMI-CS. IABP was the most IABP, Impella, VA-ECMO, etc) likely serve as
common device employed in both STEMI-CS and important mediators of in-hospital outcomes. This
NSTEMI-CS, both within 24 hours and throughout the data have been identified as a priority for future data
course of the hospitalization. collection and investigation in our registry. Third,
These results are particularly noteworthy in light we did not use multiple imputation or other
of the IABP-SHOCK II trial, which showed the absence methods to account for the limited missing data as
of a significant effect of routine IABP on all-cause we could not confirm that the missing values were
mortality at 30 days, 12 months, and at 6-year missing at random. Thus, we chose to analyze only
follow-up in patients with AMI-CS. 5,32 Although the the complete records.
routine use of IABP in Europe has diminished since
the European Society of Cardiology STEMI Guidelines CONCLUSIONS
downgraded IABP from Class I to Class IIb in 2012 and
then Class IIIb in 2017, there has been clinical inertia We report one of the largest, contemporary, real-
to adopt these recommendations in the United world registry-based analyses characterizing AMI-CS
33,34 and specifically comparing STEMI-CS and NSTEMI-
States. Given that IABP still remains the most
ubiquitous tMCS device used in both the United CS across key clinical characteristics, hemodynamic
States and worldwide, future clinical trials are needed and metabolic parameters, in-hospital mortality, as
to elucidate the appropriate, select patient population, well as drug- and device-based strategies. Despite
severity, and phenotype of CS that would maximally increasing use of inotropic and vasoactive support
benefit from this therapy. In a similar vein, vasopres- and tMCS throughout the course of the hospitaliza-
sors were the most common drug exposure in STEMI- tion, both patients with STEMI-CS and NSTEMI-CS
CS and NSTEMI-CS, both within 24 hours and remain at increased risk for in-hospital mortality.
throughout the course of the hospitalization. Inter- Future RCTs are needed to elucidate whether timing
estingly, inotropes were more likely to be used in and sequence of escalation of support improves out-
combination with vasopressors than as single agents. comes in these critically ill patients.
This finding should also be interpreted in the context ACKNOWLEDGMENT The authors wish to express
of the Dobutamine Compared with Milrinone (DOR- their sincere gratitude and appreciation to Colleen
EMI) trial, which did not demonstrate any significant Allen, who serves as the Research Project Coordinator
difference between dobutamine and milrinone on a for our Cardiogenic Shock Working Group.
primary composite endpoint, which included all-
cause, in-hospital mortality, transient ischemic FUNDING SUPPORT AND AUTHOR DISCLOSURES
attack, stroke, and cardiovascular or renal events. 6
This work was supported by a NIH RO1 grants to Dr Kapur
This multicenter cohort study has several
(R01HL139785-01; R01HL159089-01) and institutional grants from
strengths, including robust data collection of real- Abiomed Inc, Boston Scientific Inc, Abbott Laboratories, Getinge Inc,
world patients with AMI-CS, use of invasive hemo- and LivaNova Inc to Tufts Medical Center. Sponsors had no input on
dynamics, and details regarding broad drug and collection, analysis, and interpretation of the data, nor in the prepa-
ration, review, or approval of the manuscript. Dr Kapur has received
device utilization throughout patients’ clinical tra-
consulting honoraria and institutional grant support from Abbott
jectories. However, there are several important lim- Laboratories, Abiomed Inc, Boston Scientific, Medtronic, LivaNova,
itations of this study, which should be Getinge, and Zoll. Dr Hernandez-Montfort is a consultant for Abiomed
acknowledged. First, we cannot exclude the possi- Inc. Dr Abraham is a consultant for Abbott Laboratories and Abiomed
Inc. Dr Burkhoff has received support from consulting services pro-
bility of residual confounding given the retrospec-
vided from Abiomed Inc. to Cardiovascular Research Foundation. All
tive observational design of this study. In addition, other authors have reported that they have no relationships relevant
the results may be influenced by local institutional to the contents of this paper to disclose.

clinical practices and operator expertise; no specific

algorithm was prescribed to guide CS management ADDRESS FOR CORRESPONDENCE: Dr Navin K.
due to the lack of randomized data and clear Kapur, Tufts Medical Center, 800 Washington Street,
guidelines. Second, we were not able to capture Box # 80, Boston, Massachusetts 02111, USA. E-mail:
granular information regarding temporary mechani- [email protected]. Twitter: @Navin-
cal circulatory support and revascularization char- Kapur4, @ShashankSinhaMD, @BurkhoffMd, @man-
acteristics, including timing and sequence of devices reetkanwar, @JHMontfort10, @reshadgaranmd,
and/or door to balloon time. We postulate that both @mohitpahujamd, @rachkataria, @vbluml.
12 Sinha et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

A Comparison Between STEMI-CS and NSTEMI-CS MAY 2023:100314

PERSPECTIVES

COMPETENCY IN PATIENT CARE: In-hospital mor- During the first 24 hours, patients with STEMI-CS were
tality remains high across both patients with STEMI-CS more likely to receive drug and device-based therapies
and NSTEMI-CS, suggesting that patients with NSTEMI- whereas patients with NSTEMI-CS were more likely to
CS also represent a vulnerable, high-risk cohort. CA is an receive drug therapies only, providing insights into the
adverse effect modifier for both cohorts of patients. The contemporary clinical practice of patients with AMI-CS.
majority of SCAI stage B and SCAI stage C patients with
STEMI-CS and NSTEMI-CS progress during their hospi- TRANSLATIONAL OUTLOOK: RCTs are needed to
talization to SCAI stage D or E; transition to a higher SCAI elucidate whether timing and sequence of escalation of
stage is associated with worse in-hospital outcomes as vasopressors, inotropes, and/or tMCS improves outcomes
compared to those who initially presented in SCAI E. in patients with STEMI-CS and NSTEMI-CS, respectively.

REFERENCES

1. Samsky MD, Morrow DA, Proudfoot AG, 10. Samsky MD, Krucoff MW, Morrow DA, et al. (ACC), the American Heart Association (AHA), the
Hochman JS, Thiele H, Rao SV. Cardiogenic shock Cardiac safety research consortium “shock II” think Society of Critical Care Medicine (SCCM), and the
after acute myocardial infarction: a review. JAMA. tank report: advancing practical approaches to Society of Thoracic Surgeons (STS) in April 2019.
2021;326(18):1840–1850. https://doi.org/10. generating evidence for the treatment of cardio- Catheter Cardiovasc Interv. 2019;94(1):29–37.
1001/jama.2021.18323 genic shock. Am Heart J. 2020;230:93–97. https:// https://doi.org/10.1002/ccd.28329
doi.org/10.1016/j.ahj.2020.09.015
2. van Diepen S, Katz JN, Albert NM, et al. 18. Naidu SS, Baran DA, Jentzer JC, et al. SCAI
Contemporary management of cardiogenic shock: a 11. Anderson ML, Peterson ED, Peng SA, et al. SHOCK stage classification expert consensus up-
scientific statement from the American Heart As- Differences in the profile, treatment, and prog- date: a review and incorporation of validation
sociation. Circulation. 2017;136(16):e232–e268. nosis of patients with cardiogenic shock by studies: this statement was endorsed by the
https://doi.org/10.1161/CIR.0000000000000525 myocardial infarction classification: a report from American College of Cardiology (ACC), American
NCDR. Circ Cardiovasc Qual Outcomes. 2013;6(6): College of Emergency Physicians (ACEP), Amer-
3. Thiele H, Akin I, Sandri M, et al. One-year out-
708–715. https://doi.org/10.1161/CIRCOUT- ican Heart Association (AHA), European Society of
comes after PCI strategies in cardiogenic shock.
COMES.113.000262 Cardiology (ESC) Association for Acute Cardio-
N Engl J Med. 2018;379(18):1699–1710. https://
12. Thayer KL, Zweck E, Ayouty M, et al. Invasive vascular Care (ACVC), International Society for
doi.org/10.1056/NEJMoa1808788
hemodynamic assessment and classification of in- Heart and Lung Transplantation (ISHLT), Society
4. Thiele H, Akin I, Sandri M, et al. PCI strategies in of Critical Care Medicine (SCCM), and Society of
hospital mortality risk among patients with
patients with acute myocardial infarction and Thoracic Surgeons (STS) in December 2021. J Am
cardiogenic shock. Circ Heart Fail. 2020;13(9):
cardiogenic shock. N Engl J Med. 2017;377(25): Coll Cardiol. 2022;79(9):933–946. https://doi.org/
e007099. https://doi.org/10.1161/CIRCH-
2419–2432. https://doi.org/10.1056/ 10.1016/j.jacc.2022.01.018
EARTFAILURE.120.007099
NEJMoa1710261
13. Basir MB, Kapur NK, Patel K, et al. Improved 19. Hochman JS, Sleeper LA, Webb JG, et al. Early
5. Thiele H, Zeymer U, Neumann FJ, et al. Intra- outcomes associated with the use of shock pro- revascularization in acute myocardial infarction
aortic balloon support for myocardial infarction tocols: updates from the national cardiogenic complicated by cardiogenic shock. SHOCK In-
with cardiogenic shock. N Engl J Med. shock initiative. Catheter Cardiovasc Interv. vestigators. Should we emergently revascularize
2012;367(14):1287–1296. https://doi.org/10. 2019;93(7):1173–1183. https://doi.org/10.1002/ occluded coronaries for cardiogenic shock. N Engl
1056/NEJMoa1208410 ccd.28307 J Med. 1999;341(9):625–634. https://doi.org/10.
6. Mathew R, Di Santo P, Jung RG, et al. Milrinone 1056/NEJM199908263410901
14. Berg DD, Barnett CF, Kenigsberg BB, et al.
as compared with dobutamine in the treatment of Clinical practice patterns in temporary mechanical 20. Holmes DR Jr, Berger PB, Hochman JS, et al.
cardiogenic shock. N Engl J Med. 2021;385(6):516– circulatory support for shock in the critical care Cardiogenic shock in patients with acute ischemic
525. https://doi.org/10.1056/NEJMoa2026845 cardiology trials network (CCCTN) registry. Circ syndromes with and without ST-segment eleva-
7. Ouweneel DM, Eriksen E, Sjauw KD, et al. Heart Fail. 2019;12(11):e006635. https://doi.org/ tion. Circulation. 1999;100(20):2067–2073.
Percutaneous mechanical circulatory support 10.1161/CIRCHEARTFAILURE.119.006635 https://doi.org/10.1161/01.cir.100.20.2067
versus intra-aortic balloon pump in cardiogenic 15. Berg DD, Bohula EA, van Diepen S, et al. 21. Amsterdam EA, Wenger NK, Brindis RG, et al.
shock after acute myocardial infarction. J Am Coll Epidemiology of shock in contemporary cardiac 2014 AHA/ACC guideline for the management of
Cardiol. 2017;69(3):278–287. https://doi.org/10. intensive care units. Circ Cardiovasc Qual Out- patients with non-ST-elevation acute coronary
1016/j.jacc.2016.10.022 comes. 2019;12(3):e005618. https://doi.org/10. syndromes: a report of the American College of
8. Tyler JM, Brown C, Jentzer JC, et al. Variability 1161/CIRCOUTCOMES.119.005618 Cardiology/American Heart Association Task Force
in reporting of key outcome predictors in acute 16. Garan AR, Kanwar M, Thayer KL, et al. Com- on Practice Guidelines. J Am Coll Cardiol.
myocardial infarction cardiogenic shock trials. plete hemodynamic profiling with pulmonary ar- 2014;64(24):e139–e228. https://doi.org/10.1016/
Catheter Cardiovasc Interv. 2022;99(1):19–26. tery catheters in cardiogenic shock is associated j.jacc.2014.09.017
https://doi.org/10.1002/ccd.29710 with lower in-hospital mortality. J Am Coll Cardiol 22. Amsterdam EA, Wenger NK, Brindis RG, et al.
HF. 2020;8(11):903–913. https://doi.org/10.1016/ 2014 AHA/ACC guideline for the management of
9. Samsky M, Krucoff M, Althouse AD, et al. Clin-
j.jchf.2020.08.012 patients with non-ST-elevation acute coronary
ical and regulatory landscape for cardiogenic
shock: a report from the Cardiac Safety Research 17. Baran DA, Grines CL, Bailey S, et al. SCAI clin- syndromes: executive summary: a report of the
Consortium ThinkTank on Cardiogenic Shock. Am ical expert consensus statement on the classifica- American College of Cardiology/American Heart
Heart J. 2020;219:1–8. https://doi.org/10.1016/j. tion of cardiogenic shock: this document was Association Task Force on Practice Guidelines.
ahj.2019.10.006 endorsed by the American College of Cardiology J Am Coll Cardiol. 2014;64(24):2645–2687.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Sinha et al 13
MAY 2023:100314 A Comparison Between STEMI-CS and NSTEMI-CS

https://www.jacc.org/doi/10.1016/j.jacc.2014.09. 27. Heidenreich PA, Bozkurt B, Aguilar D, et al. complicating acute myocardial infarction: long-
016 2022 AHA/ACC/HFSA guideline for the manage- term 6-year outcome of the randomized IABP-
ment of heart failure: executive summary: a report SHOCK II trial. Circulation. 2019;139(3):395–403.
23. O’Gara PT, Kushner FG, Ascheim DD, et al.
of the American College of Cardiology/American https://doi.org/10.1161/CIRCULATIONAHA.118.
2013 ACCF/AHA guideline for the management of
Heart Association Joint Committee on Clinical 038201
ST-elevation myocardial infarction: executive
Practice Guidelines. J Am Coll Cardiol.
summary: a report of the American College of 33. Task Force on the management of ST-segment
2022;79(17):e263–e421. https://www.jacc.org/
Cardiology Foundation/American Heart Associa- elevation acute myocardial infarction of the Eu-
doi/10.1016/j.jacc.2021.12.012
tion Task Force on Practice Guidelines. J Am Coll ropean Society of Cardiology (ESC), Steg PG,
Cardiol. 2013;61(4):485–510. https://doi.org/10. 28. Omer MA, Tyler JM, Henry TD, et al. Clinical James SK, et al. ESC guidelines for the manage-
1016/j.jacc.2012.11.018 characteristics and outcomes of STEMI patients ment of acute myocardial infarction in patients
24. O’Gara PT, Kushner FG, Ascheim DD, et al. with cardiogenic shock and cardiac arrest. J Am presenting with ST-segment elevation. Eur Heart
2013 ACCF/AHA guideline for the management of Coll Cardiol Intv. 2020;13(10):1211–1219. https:// J. 2012;33(20):2569–2619. https://doi.org/10.
ST-elevation myocardial infarction: a report of the doi.org/10.1016/j.jcin.2020.04.004 1093/eurheartj/ehs215
American College of Cardiology Foundation/ 29. Vallabhajosyula S, Dunlay SM, Prasad A, et al. 34. Ibanez B, James S, Agewall S, et al. 2017 ESC
American Heart Association Task Force on Practice Cardiogenic shock and cardiac arrest complicating guidelines for the management of acute myocar-
Guidelines. J Am Coll Cardiol. 2013;61(4):e78– ST-segment elevation myocardial infarction in the dial infarction in patients presenting with
e140. https://www.jacc.org/doi/10.1016/j.jacc. United States, 2000-2017. Resuscitation. ST-segment elevation: the task force for the
2012.11.019 2020;155:55–64. https://doi.org/10.1016/j.resus- management of acute myocardial infarction in
25. Heidenreich PA, Bozkurt B, Aguilar D, et al. citation.2020.07.022 patients presenting with ST-segment elevation of
2022 AHA/ACC/HFSA guideline for the man- the European Society of Cardiology (ESC). Eur
30. Jentzer JC, van Diepen S, Henry TD. Under-
agement of heart failure: a report of the Heart J. 2018;39(2):119–177. https://doi.org/10.
standing how cardiac arrest complicates the
American College of Cardiology/American Heart 1093/eurheartj/ehx393
analysis of clinical trials of cardiogenic shock.
Association Joint Committee on Clinical Prac- Circ Cardiovasc Qual Outcomes. 2020;13(9):
tice Guidelines. J Am Coll Cardiol. 2022;79(17): e006692. https://doi.org/10.1161/CIRCOUT-
e263–e421. https://doi.org/10.1016/j.jacc. COMES.120.006692 KEY WORDS acute myocardial infarction,
2021.12.012 cardiogenic shock, heart failure
31. Kapur NKK, M K, Sinha SS. Criteria for defining
26. Heidenreich PA, Bozkurt B, Aguilar D, et al.
stages of cardiogenic shock severity. J Am Coll
2022 ACC/AHA/HFSA Guideline for the Manage-
Cardiol. 2022;80(3):185–198.
ment of Heart Failure: executive summary. J Am A PP END IX For supplemental tables and
Coll Cardiol. 2022;79(17):1757–1780. https:// 32. Thiele H, Zeymer U, Thelemann N, et al. figures, please see the online version of this
www.jacc.org/doi/10.1016/j.jacc.2021.12.011 Intraaortic balloon pump in cardiogenic shock paper.
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHOR. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

EDITORIAL COMMENT

Cardiogenic Shock
a Quarter Century Later
A Dire Outcome Barely Changed*
avík, MD
Vladimír Dz

A lmost a quarter of a century after the SHOCK

(Should We Emergently
Occluded Coronary Arteries for Cardiogenic
Shock) trial showed an absolute 13% reduction from
Revascularize
The authors provide an important contribution
with their careful description of the clinical, hemo-
dynamic, and metabolic parameters in STEMI and
NSTEMI shock patients. From the report, we find
63.1% to 50.3% in 6-month mortality of patients that patients with shock and NSTEMI are older.
with acute myocardial infarction (AMI) complicated Those who present after a cardiac arrest have a
by cardiogenic shock (CS) 1 outcomes of these criti- much worse outcome and require separate analysis
cally ill patients and therapies in CS have changed lit- in future randomized trials. Importantly, we find
tle.2,3 Furthermore, by far the majority of studies, that most patients who present with Society for
randomized controlled trials or observational studies, Cardiovascular Angiography and Intervention (SCAI)
have focused primarily on CS in the context ST- Stages B and C tend to progress to higher stages over
segment elevation myocardial infarction (STEMI), the course of hospitalization, highlighting the need
and even in those that included patients with non- not only for careful vigilance and management but
ST-segment elevation myocardial infarction also the need to further improve our initial treat-
(NSTEMI), little has been written about characteris- ment protocols. Interestingly they have not touched
tics and outcomes specific to this latter group of at all on the revascularization practices in these pa-
high-risk patients. tients, and we can only assume that these were
In this issue of JACC: Advances, Sinha et al 4 provide guideline based.
the most contemporary window into the treatment of They found the utilization of vasoactive and
patients with CS complicating AMI in the Cardiogenic inotropic drugs, as well as temporary mechanical
Shock Working Group (CSWG) registry, a collabora- circulatory support devices, especially in the first
tion of 17 community and university hospitals in the 24 hours, to be low with more than one-half of the
United States. The investigators enrolled 1,110 pa- patients receiving neither treatment modality within
tients between 2016 and 2020, focusing on the inter- this early period. At least in part, this could be due to
play between patient characteristics, shock severity, the relatively high systolic and mean arterial pressure
utilization of drugs and mechanical support devices, at baseline, suggesting that clinicians may have been
and patient outcomes. responding more to systemic pressures than to pul-
monary artery catheter measurements and metabolic
parameters of shock severity. Also in part, the evi-
dence to guide the timing and selection of pressor

*Editorials published in JACC: Advances reflect the views of the authors

agents, other than a subgroup analysis of randomized
and do not necessarily represent the views of JACC: Advances or the trial conducted over a decade ago showing superior-
American College of Cardiology. ity of norepinephrine over dopamine in CS patients,
From the Peter Munk Cardiac Centre, University Health Network, Tor- is generally lacking. 5
onto, Ontario, Canada. High quality evidence guiding the use of temporary
The author attests he is in compliance with human studies committees
mechanical circulatory support is at least as sparse.
and animal welfare regulations of the authors’ institutions and Food and
Drug Administration guidelines, including patient consent where While the intra-aortic balloon pump (IABP) has been a
appropriate. For more information, visit the Author Center. staple of CS treatment for close to half a century,6 a

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100357

2 Dzavík JACC: ADVANCES, VOL. 2, NO. 3, 2023

Practice Patterns in Cardiogenic Shock MAY 2023:100357

large randomized controlled trial showed unequivo- with CS secondary to STEMI or NSTEMI between 2005
cally that its routine use initiated after percutaneous and 2014, treated with primary PCI in the first
coronary intervention does not reduce mortality of 24 hours, the use of a percutaneous LVAD was asso-
patients with CS complicating either STEMI or ciated with a doubling of in-hospital mortality risk
NSTEMI.7 Yet, in the current study, the CSWG in- (OR: 2.21; 95% CI: 2.01-2.54).15 Of course administra-
vestigators report that the IABP was the most tive databases are likely to have unmeasured con-
commonly utilized MCS device at all-time points founders that may underestimate the risk of a given
during the hospitalization, a phenomenon even more group, thus adding imprecision to the assessment of
likely in most other geographies, where newer de- the efficacy and safety of any therapy. However,
vices such as the axial flow pump Impella (Abiomed these data do serve to point out the need for carefully
Inc) are simply unaffordable and unavailable as a designed randomized controlled trials to bring clarity
treatment option. to this important aspect of management of CS
The authors correctly point out that future trials patients.
must be conducted to determine the optimal Perhaps the issue is the timing of MCS support. In
setting and role of the IABP in the CS spectrum. the majority of patients enrolled in randomized trials,
Some avenues to explore might be in the timing of the pLVAD was deployed after the primary PCI pro-
IABP support. While SHOCK-IABP 2 found no cedure. The National Cardiogenic Shock Initiative
benefit from routine IABP use, the majority of de- (NCSI) is a collaboration of 35 U.S. sites that con-
vices were deployed after the PCI procedure, while ducted a prospective single-arm study in AMI CS pa-
10% of patients randomized to control crossed over tients utilizing a common algorithm of patient
to IABP support.7 There is at least a suggestion management, 16 including deployment of the Impella
from a retrospective single-center registry that CP device (most cases) prior to primary PCI. Inclusion
placement of the IABP within the first hour of onset criteria matched those of the SHOCK trial, 1 with the
of shock is associated a 50% lower in-hospital exception that patients already on IABP support were
mortality compared to later deployment.8 The size excluded. In-hospital mortality was 28%, consider-
of the IABP deployed may be important as well. A ably lower than observed in previously reported
50 cc IABP provides significantly greater unloading studies in similar shock populations. Furthermore, a
than a 40 cc device, 9 and in one single-center recent meta-analysis of 13 observational studies of
study its use resulted in very acceptable outcomes patients with AMI CS undergoing primary PCI and
in CS patients, when utilized alone, or as a bridge Impella support suggested an association with lower
to more advanced therapies. 10 mortality with pre-PCI deployment of the device. 17
Of course, as it currently stands, routine use of the Again, carefully designed and conducted random-
IABP in AMI CS patients is actually not recommended ized controlled trials are needed to confirm the effi-
in the 2021 European Society of Cardiology guide- cacy of this strategy. The currently enrolling DanGer
lines,11 while a 2021 American Heart Association pol- (Danish-German Cardiogenic Shock) trial, with a
icy statement goes further to recommend early planned study size of 360 participants, may well
initiation of advanced mechanical circulatory sup- provide the answer.18
port.12 The evidence to support this recommendation An intriguing twist to this is the concept of waiting
remains tenuous. Not only do the few very small for a period of time after placement of a pMCS device to
randomized trials comparing percutaneous LVADs vs actually perform the primary PCI. Based on a porcine
IABP not suggest even the smallest signal of survival study in which reperfusion after 30 minutes of
benefit, 13 observational U.S. registries actually reveal unloading with an Impella CP device significantly
worse outcomes with use of a percutaneous LVAD reduced infarct size by reducing proapoptotic
compared with the IABP. A propensity-matched signaling,19 the STEMI-DTU trial, a randomized
analysis of an administrative database from 14 controlled trial in patients with an anterior STEMI but
states showed a significantly higher risk of in-hospital no CS is currently in the enrollment phase, with a
mortality with the use of a percutaneous LVAD planned enrollment of 668 patients. 20 If there is a
compared with the IABP (OR: 1.63; 95% CI: 1.32- positive outcome for the pre-PCI Impella CP in the
2.02). 14 In an adjusted multivariable analysis of a very DanGer trial and in the STEMI-DTU (Primary Unloading
large National Inpatient Sample database of patients and Delayed Reperfusion in ST-Elevation Myocardial
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Dzavík 3
MAY 2023:100357 Practice Patterns in Cardiogenic Shock

Infarction) trial, STEMI-DTU, this strategy could FUNDING SUPPORT AND AUTHOR DISCLOSURES
eventually find its way to application in AMI CS.
Understandably, Sinha et al4 and the CSWG in- The author has reported that he has no relationships relevant to the
contents of this paper to disclose.
vestigators are vague in their recommendations with
respect to these therapies. However, with their careful
characterization of STEMI and NSTEMI CS patients,
documentation of SCAI stages and progression, as ADDRESS FOR CORRESPONDENCE: Dr Vladimír
well as drug and device utilization, they add an D
zavík, Interventional Cardiology Program, Peter
important chapter to our current knowledge of AMI Munk Cardiac Centre, University Health Network,
CS. Their work also suggests that we have a long way 6-246A EN, Toronto General Hospital, 200 Elizabeth
to go and much more work to do in order to improve Street, Toronto, Ontario M5G 2C4, Canada. E-mail:
the outcome of these, our most critically ill patients. [email protected].

REFERENCES

1. Hochman JS, Sleeper LA, Webb JG, et al. Early 9. Kapur NK, Paruchuri V, Majithia A, et al. coronary intervention for acute myocardial
revascularization in acute myocardial infarction Hemodynamic effects of standard versus infarction with cardiogenic shock. Euro-
complicated by cardiogenic shock. SHOCK In- larger-capacity intraaortic balloon counter- Intervention. 2021;16:e1254–e1261. https://doi.
vestigators. Should we emergently revascularize pulsation pumps. J Invasive Cardiol. 2015;27: org/10.4244/EIJ-D-19-00226
occluded coronaries for cardiogenic shock. N Engl 182–188.
16. Basir MB, Kapur NK, Patel K, et al. Improved
J Med. 1999;341:625–634.
10. Visveswaran GK, Cohen M, Seliem A, et al. outcomes associated with the use of shock pro-
2. Thiele H, Akin I, Sandri M, et al. PCI strategies in A single center tertiary care experience utilizing tocols: updates from the National Cardiogenic
patients with acute myocardial infarction and the large volume mega 50cc intra-aortic balloon Shock Initiative. Catheter Cardiovasc Interv.
cardiogenic shock. N Engl J Med. 2017;377:2419– counterpulsation in contemporary clinical practice. 2019;93:1173–1183. https://doi.org/10.1002/ccd.
2432. https://doi.org/10.1056/NEJMoa1710261 Catheter Cardiovasc Interv. 2017;90:E63–E72. 28307
https://doi.org/10.1002/ccd.26908
3. Ouweneel DM, Eriksen E, Sjauw KD, et al. 17. Iannaccone M, Franchin L, Hanson ID, et al.
Percutaneous mechanical circulatory support 11. Collet JP, Thiele H, Barbato E, et al. 2020 ESC Timing of impella placement in PCI for acute
versus intra-aortic balloon pump in cardiogenic guidelines for the management of acute coronary myocardial infarction complicated by cardiogenic
shock after acute myocardial infarction. J Am Coll syndromes in patients presenting without persis- shock: an updated meta-analysis. Int J Cardiol.
Cardiol. 2017;69:278–287. https://doi.org/10. tent ST-segment elevation. Eur Heart J. 2021;42: 2022;362:47–54. https://doi.org/10.1016/j.ijcard.
1016/j.jacc.2016.10.022 1289–1367. https://doi.org/10.1093/eurheartj/ 2022.05.011
ehaa575
4. Sinha SS, Pahuja M, Kataria R, et al. Treatment 18. Udesen NJ, Moller JE, Lindholm MG, et al.
intensity for the management of cardiogenic 12. Jacobs AK, Ali MJ, Best PJ, et al. Systems of care
Rationale and design of DanGer shock: Danish-
shock: comparison between STEMI and NSTEMI. for ST-segment-elevation myocardial infarction: a
German cardiogenic shock trial. Am Heart J.
JACC: Adv. 2023;2:100314. policy statement from the American Heart Associ-
2019;214:60–68. https://doi.org/10.1016/j.ahj.
ation. Circulation. 2021;144:e310–e327. https://doi.
5. De Backer D, Biston P, Devriendt J, et al. Com- 2019.04.019
org/10.1161/CIR.0000000000001025
parison of dopamine and norepinephrine in the
19. Esposito ML, Zhang Y, Qiao X, et al. Left
treatment of shock. N Engl J Med. 2010;362:779– 13. Ouazani Chahdi H, Berbach L, Boivin-Proulx LA,
ventricular unloading before reperfusion promotes
789. https://doi.org/10.1056/NEJMoa0907118 et al. Percutaneous mechanical circulatory support
functional recovery after acute myocardial infarc-
in post-myocardial infarction cardiogenic shock: a
6. Lawrence GH, Riggins RC, Johnston RR, tion. J Am Coll Cardiol. 2018;72:501–514. https://
systematic review and meta-analysis. Can J Car-
Balfour RI, Thompson GE. Intra-aortic balloon doi.org/10.1016/j.jacc.2018.05.034
diol. 2022;38:1525–1538. https://doi.org/10.1016/
counterpulsation for the treatment and prevention j.cjca.2022.05.018 20. Kapur NK, Kim RJ, Moses JW, et al. Primary
of cardiogenic shock. Am J Surg. 1974;128:188– left ventricular unloading with delayed reperfu-
14. Miller PE, Bromfield SG, Ma Q, et al. Clinical
194. sion in patients with anterior ST-elevation
outcomes and cost associated with an intravas-
7. Thiele H, Zeymer U, Neumann FJ, et al. cular microaxial left ventricular assist device vs myocardial infarction: rationale and design of the
Intraaortic balloon support for myocardial intra-aortic balloon pump in patients presenting STEMI-DTU randomized pivotal trial. Am Heart J.
infarction with cardiogenic shock. N Engl J Med. with acute myocardial infarction complicated by 2022;254:122–132. https://doi.org/10.1016/j.ahj.
2012;367:1287–1296. https://doi.org/10.1056/ cardiogenic shock. JAMA Intern Med. 2022;182: 2022.08.011
NEJMoa1208410 926–933. https://doi.org/10.1001/jamainternmed.
2022.2735
8. Gul B, Bellumkonda L. Usefulness of intra-
aortic balloon pump in patients with cardiogenic 15. Vallabhajosyula S, Prasad A, Sandhu GS, et al. KEY WORDS cardiogenic shock, inotropes,
shock. Am J Cardiol. 2019;123:750–756. https:// Ten-year trends, predictors and outcomes of me- NSTEMI, practice patterns, STEMI, temporary
doi.org/10.1016/j.amjcard.2018.11.041 chanical circulatory support in percutaneous circulatory mechanical support
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ORIGINAL RESEARCH

OUTCOMES AND QUALITY

Sex-Based Differences in Severe

Outcomes, Including Cardiovascular
Hospitalization, in Adults With
COVID-19 in Ontario, Canada
Bahar Behrouzi, MSC,a,b,c,d Atul Sivaswamy, MSC,c Anna Chu, MHSC,c Laura E. Ferreira-Legere, RN, MSCN,c
Husam Abdel-Qadir, MD, PHD,a,b,c,d,e Clare L. Atzema, MD, MSC,a,c,e,f Cynthia Jackevicius, PHARMD, MSC,a,c,g
Moira K. Kapral, MD, MSC,a,b,c,e Harindra C. Wijeysundera, MD, PHD,a,b,c,f Michael E. Farkouh, MD, MSC,h,i
Heather J. Ross, MD, MHSC,b,h Andrew C.T. Ha, MD, MSC,e,h Mina Tadrous, PHARMD, PHD,c,j,k Michael Paterson, MSC,a,c
avík, MD,h Jiming Fang, PHD,c Padma Kaul, PHD,l,m,n
Andrea S. Gershon, MD, MSC,a,c,e,f Vladimír Dz
Sean van Diepen, MD, MSC,l,m,n Shaun G. Goodman, MD, MSC,e,l,o Justin A. Ezekowitz, MBBCH, MSC,l,m,n
Kevin R. Bainey, MD, MSC,l,m,n Dennis T. Ko, MD, MSC,a,c,e,f Peter C. Austin, PHD,a,c Finlay A. McAlister, MD, MSC,l,m,n
Douglas S. Lee, MD, PHD,a,c,e,h Jacob A. Udell, MD, MPHa,c,d,e,h

ABSTRACT

BACKGROUND While men have experienced higher risks of SARS-CoV-2 infection compared to women, an analysis of
sex differences by age in severe outcomes during the acute phase of infection is lacking.

OBJECTIVES The purpose of this study was to assess heterogeneity in severe outcome risks by age and sex by
conducting a retrospective cohort study of community-dwelling adults in Ontario who tested positive for SARS-CoV-2
infection during the first 3 waves.

METHODS Adjusted odds ratios were estimated using multilevel multivariable logistic regression models including an
interaction term for age and sex. The primary outcome was a composite of severe outcomes (hospitalization for a car-
diovascular (CV) event, intensive care unit admission, mechanical ventilation, or death) within 30 days.

RESULTS Among 30,736, 199,132, and 186,131 adults who tested positive during the first 3 waves, 1,908 (6.2%), 5,437
(2.7%), and 5,653 (3.0%) experienced a severe outcome within 30 days. For all outcomes, the sex-specific risk depended
on age (all P for interaction <0.05). Men with SARS-CoV-2 infection experienced a higher risk of outcomes than infected
women of the same age, except for the risk of all-cause hospitalization being higher for young women than men (ages 18-
45 years) during waves 2 and 3. The sex disparity in CV hospitalization across all ages either persisted or increased with
each subsequent wave.

CONCLUSIONS To mitigate risks in subsequent waves, it is helpful to further understand the factors that contribute
to the generally higher risks faced by men across all ages, and the persistent or increasing sex disparity in the risk of CV
hospitalization. (JACC Adv 2023;2:100307) © 2023 The Authors. Published by Elsevier on behalf of the American
College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

From the aInstitute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada; bTemerty Faculty of
Medicine, University of Toronto, Toronto, Canada; cICES, Toronto, Canada; dCardiovascular Division, Department of Medicine,
Women’s College Hospital, Toronto, Canada; eDepartment of Medicine, University of Toronto, Toronto, Canada; fSunnybrook
Health Sciences Centre, University of Toronto, Toronto, Canada; gCollege of Pharmacy, Western University of Health Sciences,
Pomona, California, USA; hPeter Munk Cardiac Centre, University Health Network, Toronto, Canada; iHeart and Stroke Richard

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100307

2 Behrouzi et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex-Based Differences in Severe COVID-19 Outcomes MAY 2023:100307

C OVID-19 was first confirmed via a

ABBREVIATIONS and downstream sequelae, while also better informing
AND ACRONYMS travel-related case on January 27, patients of their risk and tailoring treatment or pre-
2020, in Ontario, the most populous ventive policy decisions. Here, we investigated the
aOR = adjusted odds ratio
province within Canada (with 14 million peo- incidence and predictors of severe SARS-CoV-2 out-
CAD = coronary artery disease
ple of diverse background), which has a uni- comes among community-dwelling adults with
CKD = chronic kidney disease
versal health care system. Studies around laboratory-confirmed SARS-CoV-2 infection in Ontario
CV = cardiovascular the world have pointed to male sex, old age, during the first 3 waves of the pandemic, while pre-
HF = heart failure and various clinical comorbidities as inde- specifying an interaction between age and sex.
HIV = human immunodeficiency pendent risk factors for increased SARS-
virus
CoV-2 infection, disease severity, and METHODS
MI = myocardial infarction death. 1-5
PHU = public health unit Interactions between sex and certain clin- STUDY POPULATION. We performed a population-
RT-PCR = reverse transcription ical comorbidities, such as chronic liver dis- based, retrospective, and SARS-CoV-2 test-positive
polymerase chain reaction
ease and obesity, were also initially found to cohort study using Ontario administrative health da-
SARS-CoV-2 = severe acute have a greater impact on the increased risk of tabases, which capture data pertaining to the vast
respiratory syndrome-
coronavirus-2
intubation and death among women.2,6 majority of the province’s population. Datasets were
However, a comprehensive analysis of deterministically linked using unique, encoded
possible sex differences by age remains to be seen for identifiers and analyzed at ICES, which is an inde-
severe outcomes, including cardiovascular complica- pendent, nonprofit research institute whose legal
tions during the acute phase of infection, particularly status under provincial health information privacy
from a large and diverse population-based cohort in law allows it to collect and analyze health care and
North America. 7-11 Further, it is unclear how such demographic data, without consent, for health sys-
possible sex differences in the risk of a severe tem evaluation and improvement.20 We identified
outcome following SARS-CoV-2 infection may have adults aged 18 years and older who were alive as of
changed at a population level throughout the waves January 1, 2020, and underwent testing for SARS-
with newer variants and vaccinations. CoV-2 through August 29, 2020, (wave 1; wild-type),
Cardiovascular (CV) involvement has been associ- August 30, 2020, to February 27, 2021, (wave 2;
ated with increasing COVID-19 risk, morbidity, and wild-type, Alpha (B.1.1.7), Beta (B.1.351), and Gamma
mortality during the first wave—becoming a hallmark (P.1) variants of concern), and February 28, 2021, to
extra-pulmonary sequela. 12 Heart failure, stroke, and July 17, 2021, (wave 3; wild-type, Alpha, Beta,
acute myocardial infarction as predominant etiol- Gamma, and Delta (B.1.617.2) variants of concern). 21
ogies of CV hospitalization following SARS-CoV-2 These time points were identified using percent
infection, have been reported in small studies. 13-15 positivity rates for testing per calendar week. We
Atrial flutter and fibrillation have been significantly required individuals to be eligible for the province’s
associated with increased mortality.16,17 Even so, sex universal Ontario Health Insurance Plan (OHIP) 1 year
differences in CV events following SARS-CoV-2 prior to the index event date as determined from the
infection are less reported, although it was recog- Ontario Registered Persons Database (RPDB), which
nized prior to the pandemic that CV disease patterns includes basic demographic information about every
not only differed between the sexes but also between person who has ever had an Ontario health insurance
age groups within the same sex. 18,19 number. Individuals who were not residents of the
As a result, examining the potential for sex dispar- province on the index date were excluded. For this
ities by age can aid efforts in understanding the bio- analysis, we focused on community-dwelling in-
logical processes involved in SARS-CoV-2 infection dividuals, residents living in long-term care facilities

Lewar Centre, University of Toronto, Toronto, Canada; jLeslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Canada;
k
Women’s College Research Institute, Toronto, Canada; lCanadian VIGOUR Centre, University of Alberta, Edmonton, Canada;
m
Department of Medicine, Faculty of Medicine & Dentistry, University of Alberta, Edmonton, Canada; nDepartment of Critical
Care Medicine and Division of Cardiology, Department of Medicine, University of Alberta, Edmonton, Canada; and the oDivision of
Cardiology, St. Michael’s Hospital, Toronto, Canada.
Muthiah Vaduganathan, MD, MPH, served as Guest Associate Editor for this paper. Michael Landzberg, MD, served as Guest
Editor-in-Chief for this paper.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received October 18, 2022; revised manuscript received December 30, 2022, accepted February 8, 2023.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Behrouzi et al 3
MAY 2023:100307 Sex-Based Differences in Severe COVID-19 Outcomes

were identified via the Ontario Drug Benefit (ODB) coronary artery disease (defined as a prior myocardial
database and excluded. The index date for study in- infarction and percutaneous or surgical coronary
clusion was the date of an individual’s first positive revascularization); hospitalization for heart failure
SARS-CoV-2 test as determined by reverse transcrip- (HF) or stroke, history of liver disease, chronic lung
tion polymerase chain reaction (RT-PCR) and recor- disease (including pneumonia, tuberculosis, asthma,
ded in the Ontario Laboratories Information System or chronic obstructive pulmonary disease), organ
(OLIS), Case and Contact Management System (CCM), transplantation, atrial fibrillation, chronic kidney
and COVID-19 Integrated Testing Data (C19INTGR) disease (CKD), or malignant cancer. Any prior history
databases. A description of the number of study par- and duration of hypertension, including diabetes or
ticipants and reasons for inclusion and exclusion are HIV were also identified. Frailty was defined using
summarized in Supplemental Figure 1. the Johns Hopkins Adjusted Clinical Groups (ACG )
version 10 frailty indicator. 24,25 Sex-specific and/or
DATA SOURCES. The cohort was created by linking
age-standardized regional rates of smoking, obesity,
multiple provincial and federal health care-related
and racial/ethnic diversity were calculated at the
databases (eg, hospital discharge abstracts, physi-
level of the public health unit using national census
cian claims, chronic disease registries, health survey,
and survey data available from Public Health
laboratory, and drug dispensing data), as well as
Ontario.26-29 Racial/ethnic community diversity was
the Immigration, Refugees, and Citizenship Canada
defined as the estimated regional visible minority
(IRCC) Permanent Resident database.22 Details on
proportion of individuals who self-identified as Black,
the data sources used are described in
South Asian, Chinese, Filipino, Latin American, Arab,
Supplemental Table 1.
Southeast Asian, West Asian, Korean, and Japanese
EXPOSURES AND STUDY COVARIATES. Sociodemo- according to national census data. 30 The receipt of
graphic data on age, sex, and area-level community influenza vaccination since the 2019/2020 season was
characteristics were derived from the Registered determined from the OHIP and Ontario Drug Benefit
Persons Database. Age was categorized into groups databases among eligible individuals. For the second
(18-45, 46-65, 66-75, 76-85, and 85 plus years) per the and third waves, receipt of none, 1, or 2 doses of
reporting styles of public health agencies to best align COVID-19 vaccination was determined from the
with potential policy implementation. Communities COVaxON database among eligible individuals.
were categorized by regional public health units,
geographic location, and size according to Statistics OUTCOMES. The primary endpoint was a severe

Canada’s 2016 census data. 23 Communities with outcome composite of hospitalization for a CV event,
<10,000 residents were classified as rural. Median intensive care unit admission, need for mechanical
neighborhood income was categorized by quintile ventilation, or all-cause death. Secondary outcomes
according to national census data. Individuals who were hospitalization for a CV event, all-cause hospi-
immigrated to Ontario as their first place of landing in talization, and all-cause death. CV events included
Canada between 1985 and 2017 were identified via the need for a pacemaker or implantable cardioverter
IRCC permanent resident database. defibrillator insertion, resuscitated cardiac arrest,
Clinical comorbidities were identified based on myocardial infarction (MI), HF, stroke, carditis (ie,
hospitalization and emergency department records myocarditis or pericarditis), and any tachy- or bra-
from the Canadian Institute for Health Information dyarrhythmia. All outcomes were ascertained within
Discharge Abstract Database (CIHI DAD) and National 30 days following the index SARS-CoV-2 positive test
Ambulatory Care Reporting System (NACRS), respec- date, corresponding to the acute phase of infection.
tively using International Classification of Diseases- Details on the definitions and codes used are
10th Revision-Canada (ICD-10-CA) coding, hospital described in Supplemental Table 1.
and physician procedure coding, and chronic STATISTICAL ANALYSIS. Baseline characteristics
disease diagnoses from the OHIP database among all eligible COVID-19 test-positive individuals
(Supplemental Table 1). In addition to the number of were compared between those who did and did not
prior hospitalization or emergency department epi- experience a severe outcome, with standardized
sodes in the prior year, we also included the following differences <0.1 indicating no relevant difference
characteristics within the previous 5 years using between the groups. Multilevel multivariable logistic
previously validated case definition algorithms for regression models, with clustering at the level of the
Canadian administrative databases: history of public health unit were used with adjustment for the
4 Behrouzi et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex-Based Differences in Severe COVID-19 Outcomes MAY 2023:100307

aforementioned covariates, to examine the associa- frequently men, with higher comorbidities and
tion between clinical and sociodemographic risk fac- frailty, and residing in regions with higher smoking
tors and risks of the primary and secondary outcomes rates.
of interest. Models were adjusted for time (days from The primary and secondary outcome rates,
start of the wave), sex, age group, rurality, income stratified by age and sex groups, among community-
quintile, immigrant status, number of hospitaliza- dwelling adults with laboratory-confirmed SARS-
tions/emergency department visits in the past year, CoV-2 infection are presented in the Central
frailty, history of hypertension, diabetes mellitus, Illustration and Supplemental Figure 2. Across all 3
coronary artery disease, HF, stroke, atrial fibrillation, waves, men experienced higher unadjusted rates of a
CKD, HIV, organ transplantation, cancer, liver dis- severe outcome compared with women, although the
ease, lung disease, influenza vaccination since 2019/ risk of a severe outcome increased progressively with
2020, COVID-19 vaccination (waves 2 and 3), and age for both. For all age groups, except for those 85
public health unit-level smoking rate, obesity rate, years and above, there was a reduction in unadjusted
and visible minority rate. Given the reports of sex rates across the waves.
differences in adverse outcomes among older in- Clinical risk factors and community-level socio-
dividuals in other regions, an interaction term be- demographic factors that were independently asso-
tween age and sex was included for each regression. ciated with a severe outcome are presented in Table 2.
To evaluate these risks during successive waves of After multivariable adjustment, independent pre-
the pandemic, we modeled each wave separately. A dictors of a severe outcome across all 3 waves
sensitivity analysis excluding persons who were included the number of hospitalization/emergency
hospitalized on the same day as SARS-CoV-2 testing department visits in the prior year, and a history of
was conducted to reduce reverse causation bias. The hypertension, diabetes mellitus, HF, atrial fibrilla-
interaction between age and sex was also assessed tion, CKD, cancer, or lung disease. Neighborhood in-
with age as a continuous variable using restricted come quintile, frailty, and a history of a transplant or
cubic splines with 5 knots. Due to lags in data liver disease were additional independent predictors
reporting and specimen labeling, it is appropriate to that were common to waves 2 and 3. The receipt of
define severe outcomes using the earliest of the COVID-19 vaccination was found to be significantly
specimen collection date or hospital admission protective in waves 2 and 3.
date. 31,32 All P values were 2-sided with P < 0.05 For all outcomes, across all waves, there was a
considered statistically significant. All statistical significant interaction between age and sex (all
analyses were conducted using SAS version 9.4 (SAS P-interaction <0.05; Figure 1), indicating that sex-
Institute). This study followed the Strengthening specific risk depended on age. Overall, men with
the Reporting of Observational Studies in Epidemi- SARS-CoV-2 infection experienced a higher risk of the
ology (STROBE) reporting guideline.33 The use of primary and secondary outcomes than SARS-CoV-2
data in this project is authorized under section 45 infected women of the same age, except for the risk
of Ontario’s Personal Health Information Protection of all-cause hospitalization being higher for young
Act and does not require review by a research women than men (ages 18-45 years) during waves 2
ethics board. and 3. The adjusted odds ratio (aOR) for men
compared to women across the full spectrum of age is
RESULTS also shown in Supplemental Figure 3 (all P for
interaction <0.05 except for all-cause death due to
Among the 30,736 eligible community-dwelling imprecise estimates from low event counts among
adults that tested positive for SARS-CoV-2 in younger persons).
Ontario in wave 1, 1,908 (6.2%) experienced a severe In younger individuals (age 18-65 years), the sex-
outcome (ie, the primary composite endpoint) within specific multivariable aOR for the primary outcome
30 days. In wave 2, among the 199,132 adults that approached approximately 2-fold higher in men
tested positive, 5,437 (2.7%) experienced a severe compared with women and decreased with each
outcome within 30 days. In wave 3, among the 186,131 subsequent wave. Among older age groups (66 years
adults that tested positive, 5,653 (3.0%) experienced a and up), the sex-specific risk increased from waves 1
severe outcome within 30 days. Baseline character- to 2 and attenuated in wave 3, with men experiencing
istics among individuals with and without a severe a 1.5 times higher risk than women (Central
outcome following laboratory-confirmed SARS-CoV-2 Illustration).
infection are presented in Table 1. Across all waves, The sex difference in risk of all-cause hospitaliza-
individuals with a severe outcome were older, more tion generally decreased with each wave (all P for
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Behrouzi et al 5
MAY 2023:100307 Sex-Based Differences in Severe COVID-19 Outcomes

interaction <0.001) (Figure 1). Across all waves, hos- discourse has been centered on the risks faced by older
pitalization was mainly driven by pneumonia, which adults, we have shown here that younger age groups
did not show a significant age-sex interaction except across sexes also experienced severe adverse out-
during wave 2 (Supplemental Table 2). comes, including death, as has been seen else-
The total incidence of CV hospitalization among all where.3,34 As a hard outcome, our findings for the sex-
persons who tested positive was 1.95%, 1.09%, and specific risk of all-cause death indicate a substantial
1.05% in waves 1 to 3, respectively (Supplemental mitigation of the sex gap in young adults (age 18-45
Table 2). For CV hospitalization, across all waves, years) by the third wave, from 4-fold higher in men to
middle-aged men (ages 46-65 years) were at the null. Barring changes to testing criteria, it is possible
highest risk compared to women of the same age, that nonpharmaceutical interventions and vaccina-
with up to a 3-fold higher aOR (Figure 1). Further, with tion may have had their greatest mitigating effect for
each subsequent wave, there was no change in the this age group. However, for adults aged 45 years and
increased risk borne by men versus women for all age up, the sex-specific risk was higher during the second
groups as the aORs generally increased or stayed the wave (corresponding to the fall/winter) and subse-
same. Regarding underlying CV event etiology, the quently returned to magnitudes seen in the first wave,
most common cause of CV hospitalization was which may reflect viral seasonality or other factors that
attributed to arrhythmias (tachy- or brady- warrant further investigation.
arrhythmia), followed by HF, stroke, and acute MI Secondly, although we observed some reductions
across all 3 waves (Figure 2, Supplemental Figure 4). in the adjusted sex-specific risks for the other out-
Carditis as the cause of CV hospitalization was comes with each wave, the sex-specific aORs for CV
infrequent. hospitalization increased further or remained high.
Regarding all-cause death, young men (age 18-45 This suggests that despite learning to better manage
years) faced the highest discrepancy in risk during COVID-19-associated morbidity and mortality with
wave 1, with a 4-fold higher risk compared to similar time, this did not extend to mitigating CV morbidity,
aged women. Although absolute counts were low for which may lead to persistent future gaps for men.
this outcome among younger age groups, the aORs Further, CV hospitalization is the only endpoint to
and 95% CIs were statistically significant compared to show greater disparities in sex-specific risk among
those for older age groups (45 years and older), for younger versus older age groups across all 3 waves.
whom the risk attenuated. However, this pattern This could be partly explained by the systematic
changed with subsequent waves when older men underappreciation or underdetection of adverse out-
were at a significantly higher risk, approaching up to a comes among younger women, which may be due to
3-fold difference in risk compared to women of the sex-based differences in CV symptom presentation,
same age. Across waves, aORs decreased the most for patterns of comorbidities, and health care use. 35-37
young men, whereas they either remained constant This underdetection has been shown to attenuate
or increased for older men. with increasing age due to better recognition and
Finally, results for the primary outcome were timely care. 38,39 Corroborating existing literature,
robust to challenge in a sensitivity analysis where we also found that CV hospitalization was largely
patients who were hospitalized on the day of testing attributed to arrhythmias followed by HF, stroke,
were excluded (Supplemental Table 3). and acute MI. 13-15 Whether this is specific to COVID-
19, or more plausibly, a generic response to the
DISCUSSION inflammatory milieu driven by viral respiratory ill-
nesses remains to be seen.12 Prior studies have
In this large, population-based study, we observed 2 noted similar etiologies and rates with influenza,
key findings to inform the ongoing pandemic. First, which may be confounded by systematic differences
the significant age-sex interaction for adverse out- in testing, care, and hospitalization patterns of pa-
comes underscores that men and women of the same tients with common viral respiratory in-
age have thus far not had the same risk of severe out- fections. 16,40,41 Regardless, these CV hospitalizations
comes following laboratory-confirmed SARS-CoV-2 during the acute phase of infection can have sub-
infection. Men have almost always experienced a stantial long-term consequences for the CV health
higher risk through the first 3 waves, despite changes of a large number of people, across all ages and
in the dominant strain, nonpharmaceutical in- sexes.42 Hence, these findings underscore the
terventions, treatment guidelines, testing eligibility importance of maintaining steadfast focus on stra-
criteria, and the increasing availability/use of vaccines tegies for the primary prevention of SARS-CoV-2
and COVID-19-specific therapies. Of note, although infection.
6 Behrouzi et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex-Based Differences in Severe COVID-19 Outcomes MAY 2023:100307

T A B L E 1 Baseline Characteristics of Patients (Age $18 Years) in Ontario With Laboratory-Confirmed SARS-CoV-2 Infection by Severe Outcome Status

Wave 1 Wave 2 Wave 3

Severe No Severe Severe No Severe Severe No Severe

Outcome Outcome Outcome Outcome Outcome Outcome
(n ¼ 1,908) (n ¼ 28,828) SD (n ¼ 5,437) (n ¼ 193,695) SD (n ¼ 5,653) (n ¼ 180,478) SD

Sociodemographic characteristics
Age at test date 71 (59-83) 46 (32-58) 1.42 75 (63-85) 42 (29-56) 1.70 66 (54-77) 40 (29-54) 1.35
18-45 y 120 (6.3) 14,352 (49.8) 1.11 288 (5.3) 107,669 (55.6) 1.31 700 (12.4) 107,135 (59.4) 1.12
46-65 y 621 (32.5) 10,823 (37.5) 0.11 1,264 (23.2) 65,272 (33.7) 0.23 2,026 (35.8) 57,006 (31.6) 0.09
66-75 y 405 (21.2) 1,952 (6.8) 0.43 1,201 (22.1) 12,309 (6.4) 0.46 1,336 (23.6) 10,861 (6.0) 0.51
76-85 y 392 (20.5) 993 (3.4) 0.55 1,369 (25.2) 5,514 (2.8) 0.68 975 (17.2) 4,109 (2.3) 0.52
>85 y 370 (19.4) 708 (2.5) 0.56 1,315 (24.2) 2,931 (1.5) 0.72 616 (10.9) 1,367 (0.8) 0.44
Sex
Female 780 (40.9) 15,477 (53.7) 0.26 2,170 (39.9) 99,402 (51.3) 0.23 2,299 (40.7) 90,283 (50.0) 0.19
Male 1,128 (59.1) 13,351 (46.3) 0.26 3,267 (60.1) 94,293 (48.7) 0.23 3,354 (59.3) 90,195 (50.0) 0.19
Rural 81 (4.2) 1,133 (3.9) 0.02 199 (3.7) 7,517 (3.9) 0.01 258 (4.6) 7,409 (4.1) 0.02
Income quintile
1 (lowest) 542 (28.4) 7,750 (26.9) 0.03 1,639 (30.1) 47,066 (24.3) 0.13 1,782 (31.5) 43,271 (24.0) 0.17
2 419 (22.0) 6,047 (21.0) 0.02 1,275 (23.5) 42,633 (22.0) 0.03 1,247 (22.1) 38,665 (21.4) 0.02
3 358 (18.8) 5,938 (20.6) 0.05 1,050 (19.3) 42,019 (21.7) 0.06 1,146 (20.3) 38,912 (21.6) 0.03
4 303 (15.9) 4,840 (16.8) 0.03 799 (14.7) 34,575 (17.9) 0.09 842 (14.9) 33,266 (18.4) 0.10
5 (highest) 286 (15.0) 4,253 (14.8) 0.01 674 (12.4) 27,402 (14.1) 0.05 636 (11.3) 26,364 (14.6) 0.10
Community size
>1.5M 1,231 (64.5) 18,272 (63.4) 0.02 3,431 (63.1) 127,869 (66.0) 0.06 3,545 (62.7) 115,742 (64.1) 0.03
500K-1.5M 258 (13.5) 3,867 (13.4) 0.003 758 (13.9) 24,675 (12.7) 0.04 841 (14.9) 24,981 (13.8) 0.03
100K-500K 257 (13.5) 3,708 (12.9) 0.02 814 (15.0) 26,977 (13.9) 0.03 771 (13.6) 26,667 (14.8) 0.03
10K-100K 80 (4.2) 1,826 (6.3) 0.10 231 (4.2) 6,716 (3.5) 0.04 240 (4.2) 5,658 (3.1) 0.06
<10K 82 (4.3) 1,155 (4.0) 0.02 203 (3.7) 7,458 (3.9) 0.006 256 (4.5) 7,430 (4.1) 0.02
Regionalb smoking rate 12 (9-14) 12 (9-14) 0.18 12 (9-15) 12 (9-14) 0.20 12 (9-15) 12 (9-14) 0.16
Regionalb obesity rate 19 (14-23) 17 (15-22) 0.02 19 (14-23) 17 (15-22) 0.02 19 (15-23) 19 (15-22) 0.03
Regionalb visible minority proportion 50 (19-52) 50 (19-52) 0.04 50 (19-52) 51 (19-52) 0.11 50 (19-52) 50 (19-52) 0.06
Regionalb visible minority proportion quartile
1 57 (3.0) 802 (2.8) 0.01 89 (1.6) 3,131 (1.6) 0.002 196 (3.5) 4,984 (2.8) 0.04
2 49 (2.6) 1,028 (3.6) 0.06 229 (4.2) 7,636 (3.9) 0.01 349 (6.2) 7,591 (4.2) 0.09
3 255 (13.4) 3,557 (12.3) 0.03 733 (13.5) 24,931 (12.9) 0.02 606 (10.7) 22,670 (12.6) 0.06
4 1,547 (81.1) 23,441 (81.3) 0.006 4,386 (80.7) 157,997 (81.6) 0.02 4,502 (79.6) 145,233 (80.5) 0.02
Immigrant 538 (28.2) 10,821 (37.5) 0.20 1,523 (28.0) 70,820 (36.6) 0.18 1,829 (32.4) 58,251 (32.3) 0.03
Continued on the next page

The sex difference we report is likely a culmination have shown that in the first 3 waves, men experienced
of genetic and hormonal factors, mechanistic biolog- more severe disease and fatality than women of the
ical differences following infection, and gender- same age even after adjusting for clinical comorbid-
specific behaviors. For example, men have higher ities and sociodemographic characteristics. 43
angiotensin-converting enzyme 2 (ACE2) receptor This study has several strengths. Using a rich
expression that can promote viral entry, more robust network of health care databases that capture
innate immune and proinflammatory responses that approximately 15 million people in Ontario (repre-
lead to severe disease, and higher proclivities for poor senting more than half of Canada’s total visible mi-
health behaviors (eg, smoking).10,43-45 Access to norities), we were able to include almost all patients
testing and health care is also associated with in Ontario with laboratory-confirmed SARS-CoV-2
gendered behaviors and societal and cultural norms. 46 infection during the first 3 waves. We had access to
In Ontario, more females than males were tested gold standard diagnostic data via RT-PCR confirma-
during the first wave, which suggests that the tion versus relying on clinical syndrome proxies to
observed differences in outcomes may not be identify patients with COVID-19. While we pre-
explained by systematic testing differences, or that specified a comprehensive list of outcomes, we
there may have been a selection bias toward males focused our analyses on a composite primary
that were further along in illness. 47 Importantly, we endpoint and CV hospitalization as their event
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Behrouzi et al 7
MAY 2023:100307 Sex-Based Differences in Severe COVID-19 Outcomes

T A B L E 1 Continued

Wave 1 Wave 2 Wave 3

Severe No Severe Severe No Severe Severe No Severe

Outcome Outcome Outcome Outcome Outcome Outcome
(n ¼ 1,908) (n ¼ 28,828) SD (n ¼ 5,437) (n ¼ 193,695) SD (n ¼ 5,653) (n ¼ 180,478) SD

Clinical characteristics
Hospitalized during index event 421 (22.1) 394 (1.4) 0.68 1,392 (25.6) 1,208 (0.6) 0.80 898 (15.9) 841 (0.5) 0.59
If hospitalized during index event, 260 (13.6) 248 (0.9) 0.51 583 (10.7) 472 (0.2) 0.47 389 (6.9) 364 (0.2) 0.37
main diagnosis was COVID-19
Charlson score
No hospitalization 953 (49.9) 23,377 (81.1) 0.69 2,319 (42.7) 160,880 (83.1) 0.92 3,143 (55.6) 149,400 (82.8) 0.62
0 231 (12.1) 3,542 (12.3) 0.005 671 (12.3) 23,164 (12.0) 0.01 682 (12.1) 23,247 (12.9) 0.03
1 206 (10.8) 760 (2.6) 0.33 583 (10.7) 3,922 (2.0) 0.36 454 (8.0) 3,375 (1.9) 0.29
2 170 (8.9) 538 (1.9) 0.32 593 (10.9) 2,771 (1.4) 0.40 421 (7.4) 2,192 (1.2) 0.31
>3 348 (18.2) 611 (2.1) 0.55 1,271 (23.4) 2,958 (1.5) 0.70 953 (16.9) 2,264 (1.3) 0.57
Frailty (Johns Hopkins Aggregated Clinical Groups) 434 (22.7) 1,011 (3.5) 0.59 1,445 (26.6) 4,144 (22.2) 0.74 749 (13.2) 2,453 (1.4) 0.47
Number of hospitalizations or
ED visits in the past year
0 819 (42.9) 17,511 (60.7) 0.36 2,397 (44.1) 137,963 (71.2) 0.57 2,931 (51.8) 131,529 (72.9) 0.45
1-2 732 (38.4) 9,429 (32.7) 0.12 2,044 (37.6) 47,588 (24.6) 0.28 1,910 (33.8) 41,305 (22.9) 0.24
>3 357 (18.7) 1,888 (6.5) 0.37 996 (18.3) 8,144 (4.2) 0.46 812 (14.4) 7,644 (4.2) 0.35
Hypertensionc 1,403 (73.5) 7,662 (26.6) 1.06 4,106 (75.5) 43,026 (22.2) 1.26 3,593 (63.6) 34,553 (19.1) 1.01
Duration of hypertensiond 16 (9-24) 11 (5-18) 0.39 17 (10-25) 11 (5-18) 0.59 16 (9-23) 11 (5-18) 0.46
Diabetes mellitusc 712 (37.3) 3,609 (12.5) 0.60 2,205 (40.6) 22,034 (11.4) 0.71 2,080 (36.8) 17,974 (10.0) 0.67
CADc 228 (11.9) 664 (2.3) 0.38 841 (15.5) 3,918 (2.0) 0.49 678 (12.0) 3,298 (1.8) 0.41
HFc 223 (11.7) 234 (0.8) 0.46 781 (14.4) 1,060 (0.5) 0.55 539 (9.5) 753 (0.4) 0.43
Strokec 57 (3.0) 168 (0.6) 0.18 220 (4.0) 720 (0.4) 0.25 141 (2.5) 506 (0.3) 0.19
CKDc 405 (21.2) 807 (2.8) 0.59 1,337 (24.6) 4,413 (2.3) 0.69 1,023 (18.1) 3,569 (2.0) 0.56
Atrial fibrillationc 343 (18.0) 625 (2.2) 0.54 1,129 (20.8) 3,115 (1.6) 0.64 833 (14.7) 2,384 (1.3) 0.51
HIVc 8 (0.4) 76 (0.3) 0.03 13 (0.2) 416 (0.2) 0.005 19 (0.3) 436 (0.2) 0.02
Transplantationc 14 (0.7) 33 (0.1) 0.10 70 (1.3) 195 (0.1) 0.14 79 (1.4) 148 (0.1) 0.15
Cancerc 150 (7.9) 631 (2.2) 0.26 424 (7.8) 3,322 (1.7) 0.29 351 (6.2) 2,668 (1.5) 0.25
Liver diseasec 29 (1.5) 89 (0.3) 0.13 127 (2.3) 448 (0.2) 0.19 107 (1.9) 469 (0.3) 0.16
Lung diseasec 1,207 (63.3) 5,696 (19.8) 0.98 3,086 (56.8) 32,738 (16.9) 0.91 2,834 (50.1) 30,934 (17.1) 0.75
COPDc 256 (13.4) 562 (1.9) 0.44 736 (13.5) 2,663 (1.4) 0.48 503 (8.9) 2,160 (1.2) 0.36
Asthmac 277 (14.5) 2,879 (10.0) 0.14 812 (14.9) 20,260 (10.5) 0.14 743 (13.1) 19,834 (11.0) 0.07
Respiratory tuberculosisc #5a #18a (0.0) >0.01a 7 (0.1) 61 (0.03) 0.03 8 (0.1) 39 (0.0002) 0.04
Pneumoniac 1,062 (55.7) 3,156 (10.9) 1.08 2,624 (48.3) 13,913 (7.2) 1.03 2,392 (42.3) 12,519 (6.9) 0.90
Influenza vaccination in the past year 756 (39.6) 5,908 (20.5) 0.43 1,871 (34.4) 35,648 (18.4) 0.37 1,531 (27.1) 28,479 (15.8) 0.02
after 2019/2020 season
COVID-19 vaccination status
Two NA NA - #5a #82a <0.01a 71 (1.3) 1,808 (1.0) 0.02
One NA NA - #31a #700a >0.01a 757 (13.4) 16,828 (9.3) 0.13
None 1,908 (100) 28,828 (100) - 5,405 (99.4) 192,917 (99.6) 0.03 4,825 (85.4) 161,842 (89.7) 0.13

Values are median (IQR) or n (%). A severe outcome was defined as a composite of CV hospitalization, ICU admission, MV, or death. There were n ¼ 30,736 during the first wave, n ¼ 199,132 during the second
wave, and n ¼ 186,131 during the third wave. aNot reported to protect the confidentiality of individuals due to small cells. bSee the Supplemental Appendix for details regarding calculating regional variables.
c
History of comorbidity. dOnly identified for a subset of patients in the ICES HTN registry as of September 2020.
CAD ¼ coronary artery disease; CKD ¼ chronic kidney disease; COPD ¼ chronic obstructive pulmonary disease; CV ¼ cardiovascular; ED ¼ emergency department; HF ¼ heart failure; HIV ¼ human
immunodeficiency virus; ICU ¼ intensive care unit; IQR ¼ interquartile range; K ¼ thousand; M ¼ million; MV ¼ mechanical ventilation; NA ¼ not applicable as SARS-CoV-2 vaccines not yet available;
SD ¼ standardized difference.

ascertainment is less susceptible to variation in sub- variables selected based on established knowledge,
jective clinical decision-making and practice pat- including 8 variables from validated disease registries
terns. This is also one of the largest contemporary that leverage diagnostic codes, prescription records,
analyses to specifically report on CV hospitalization and laboratory test results. We also challenged the
in the acute phase of laboratory-confirmed SARS- robustness of results in a sensitivity analysis. As a
CoV-2 infection (#30 days). 48,49 Our modeling result, this is one of the largest population-based
approach included specification of 30 predefined analyses in a North American context of the
8 Behrouzi et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex-Based Differences in Severe COVID-19 Outcomes MAY 2023:100307

C E NTR AL IL L USTR AT I O N Absolute Risk and aORs, Including Age-Sex Interaction, for the Severe Outcome
Composite of Cardiovascular Hospitalization, Intensive Care Unit Admission, Mechanical Ventilation, or Death,
Across 3 Waves

Behrouzi B, et al. JACC Adv. 2023;2(3):100307.

Absolute risk (ie, incidence proportion) on the primary y-axis is expressed in percentages of persons of each sex experiencing the outcome within 30 days, from persons
of the same sex with laboratory-confirmed COVID-19 in the noted age group. Denominators are not shown due to small cells. aORs and 95% CI are shown to 1 decimal
place, with women as the referent group for each age group; determined from a multivariable logistic regression model incorporating public health unit-specific
random effects. Models were adjusted for time (days from start of the wave), sex, age group, age-sex interaction, rurality, income quintile, immigrant status, number
of hospitalizations/emergency department visits in the past year, frailty, history of hypertension, diabetes mellitus, coronary artery disease, heart failure, stroke,
atrial fibrillation, chronic kidney disease, HIV, organ transplantation, cancer, liver disease, lung disease, influenza vaccination since 2019/2020, COVID-19 vaccination
(waves 2 and 3), and public health unit-level smoking rate, obesity rate and visible minority rate. aOR ¼ adjusted odds ratio.

cumulative effect of clinical, sociodemographic, and in the first wave. In restricting our analysis to people
socioeconomic risk factors stratified by age and sex tested for infection, there is the potential for collider
on severe outcomes following SARS-CoV-2 infection bias to distort associations between the adverse out-
in the first 3 waves. comes of interest and risk factors that may have
STUDY LIMITATIONS. Access to and eligibility affected the likelihood of an individual being
criteria for testing varied over time. Some patients tested. 50 As a result, we may be underestimating the
with severe COVID-19 may have died prior to testing odds by which clinical comorbidities may contribute
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Behrouzi et al 9
MAY 2023:100307 Sex-Based Differences in Severe COVID-19 Outcomes

T A B L E 2 Predictors of a Severe Outcome Within 30 Days of Laboratory-Confirmed SARS-CoV-2 Infection Among Community-Dwelling Adults in Ontario

Wave 1 Wave 2 Wave 3

Characteristic aOR (95% CI) P Value Characteristic aOR (95% CI) P Value Characteristic aOR (95% CI) P Value

Lung diseasea 3.47 (3.11-3.88) <0.0001 Lung diseasea 2.73 (2.56-2.91) <0.0001 Lung diseasea 2.65 (2.50-2.82) <0.0001
CKDa 1.74 (1.47-2.06) <0.0001 CKDa 1.60 (1.46-1.75) <0.0001 CKDa 1.65 (1.49-1.82) <0.0001
a a
Atrial fibrillation 1.66 (1.38-2.00) <0.0001 Atrial fibrillation 1.62 (1.47-1.78) <0.0001 Atrial fibrillationa 1.96 (1.76-2.18) <0.0001
Hypertensiona 1.60 (1.40-1.83) <0.0001 Hypertensiona 1.45 (1.34-1.57) <0.0001 Hypertensiona 1.55 (1.44-1.67) <0.0001
HFa 1.58 (1.24-2.01) 0.0002 HFa 1.69 (1.49-1.91) <0.0001 HFa 1.73 (1.49-2.00) <0.0001
Diabetes mellitusa 1.43 (1.26-1.61) <0.0001 Diabetes mellitusa 1.39 (1.30-1.49) <0.0001 Diabetes mellitusa 1.52 (1.42-1.62) <0.0001
Cancera 1.43 (1.15-1.79) 0.002 Cancera 1.42 (1.25-1.61) <0.0001 Cancera 1.42 (1.24-1.63) <0.0001
Number of 0.007 Number of <0.0001 Number of <0.0001
hospitalizations or ED hospitalizations or ED hospitalizations or ED
visits in previous year visits in previous year visits in previous year
0 Ref 0 Ref 0 Ref
1-2 0.88 (0.78-0.99) 1-2 1.30 (1.22-1.40) 1-2 1.29 (1.21-1.37)
3þ 1.13 (0.95-1.34) 3þ 1.62 (1.46-1.79) 3þ 1.63 (1.47-1.80)
Strokea 0.70 (0.49-0.99) 0.047 Strokea 1.15 (0.96-1.38) 0.132 Strokea 1.28 (1.02-1.59) 0.033
Frailty (Johns Hopkins 1.14 (0.96-1.34) 0.139 Frailty (Johns Hopkins 1.53 (1.40-1.67) <0.0001 Frailty (Johns Hopkins’ 1.39 (1.25-1.56) <0.0001
Aggregated Clinical Aggregated Clinical Aggregated Clinical
Groups) Groups) Groups)
Transplantationa 1.61 (0.77-3.37) 0.207 Transplantationa 3.00 (2.15-4.19) <0.0001 Transplantationa 4.74 (3.40-6.62) <0.0001
Liver diseasea 1.18 (0.72-1.96) 0.511 Liver diseasea 1.93 (1.51-2.48) <0.0001 Liver diseasea 1.69 (1.30-2.19) <0.0001
Income quintile 0.071 Income quintile <0.0001 Income quintile <0.0001
1 (lowest) 1.26 (1.05-1.51) 1 (lowest) 1.19 (1.07-1.32) 1 (lowest) 1.31 (1.18-1.46)
2 1.27 (1.06-1.52) 2 1.09 (0.97-1.21) 2 1.15 (1.03-1.28)
3 1.11 (0.92-1.34) 3 1.03 (0.93-1.16) 3 1.10 (0.99-1.22)
4 1.19 (0.98-1.44) 4 0.93 (0.83-1.05) 4 1.02 (0.91-1.14)
5 (highest) Ref 5 (highest) Ref 5 (highest) Ref
Immigrant 1.13 (0.995-1.29) 0.059 Immigrant 1.02 (0.95-1.10) 0.539 Immigrant 1.08 (1.01-1.15) 0.026
Vaccination status 0.001 Vaccination status <0.0001
1 dose 0.46 (0.30-0.71) 1 dose 0.39 (0.36-0.43)
2 doses 0.47 (0.08-2.70) 2 doses 0.32 (0.24-0.42)
None Ref None Ref
Rural 1.06 (0.77-1.44) 0.731 Rural 0.86 (0.72-1.02) 0.085 Rural 0.99 (0.84-1.16) 0.855
CADa 0.92 (0.76-1.12) 0.403 CADa 1.01 (0.91-1.11) 0.897 CADa 1.00 (0.90-1.11) 0.986
HIVa 1.50 (0.64-3.49) 0.348 HIVa 0.83 (0.44-1.54) 0.545 HIVa 1.06 (0.64-1.74) 0.827
Influenza vaccination in 0.99 (0.88-1.12) 0.901 Influenza vaccination in 0.96 (0.90-1.02) 0.206 Influenza vaccination in 0.98 (0.92-1.05) 0.600
the past year after the past year after the past year after
2019/2020 season 2019/2020 season 2019/2020 season
Regionalb smoking rate 1.01 (0.97-1.05) 0.737 Regionalb smoking rate 0.99 (0.97-1.02) 0.495 Regionalb smoking rate 1.00 (0.98-1.03) 0.754
Regionalb obesity rate 1.01 (0.98-1.03) 0.630 Regionalb obesity rate 0.995 (0.98-1.01) 0.362 Regionalb obesity rate 0.997 (0.99-1.01) 0.661
Regionalb visible 1.01 (0.997-1.02) 0.179 Regionalb visible 0.996 (0.99-1.00) 0.059 Regionalb visible 0.997 (0.99-1.002) 0.283
minority proportion minority proportion minority proportion

A severe outcome was defined as a composite of CV hospitalization, ICU admission, MV, or death from any cause. Odds ratios were determined from multivariable logistic regression models that incorporated
PHU-specific random effects. The models were adjusted for time (days from start of wave using restricted cubic splines), sex, age group, age-sex interaction, rurality, income quintile, immigrant status,
number of hospitalizations or ED visits in the past year, frailty, history of hypertension, diabetes mellitus, CAD, HF, stroke, atrial fibrillation, CKD, HIV, organ transplantation, cancer, liver disease, lung disease,
influenza vaccination since 2019/2020, COVID-19 vaccination (waves 2 and 3 only), PHU-level smoking rate (sex-specific and age-standardized), PHU-level obesity rate (sex-specific and age-standardized),
and PHU-level visible minority rate (sex-specific). There was a significant interaction in the overall model between age group and sex for each wave: For age-sex aORs, see Figure 1 and Supplemental Table 4.
a b
Bold values emphasize protective predictors. 95% CIs crossing unity are set in italic for emphasis. History of comorbidity. See Supplemental Appendix for details regarding calculating regional variables.
CAD ¼ coronary artery disease; CI ¼ confidence interval; CKD ¼ chronic kidney disease; CV ¼ cardiovascular; ED ¼ emergency department; HF ¼ heart failure; HIV ¼ human immunodeficiency virus;
ICU ¼ intensive care unit; MV ¼ mechanical ventilation; OR ¼ odds ratio; PHU ¼ public health unit.

to severe outcomes. To best mitigate the effects of confounding. Although we assessed sex as captured
ascertainment bias, we adjusted for time in our model in administrative data, we were not able to utilize
and used a test-positive design for simplicity. gender-based analysis to evaluate how women, men,
Further, although our exposures and covariates were and gender-diverse people may have been impacted.
comprehensive, they were captured as part of We were also circumspect to include hypertension
routinely collected data and may be subject to vary- duration in our multivariable analyses due to lags in
ing degrees of misclassification bias and residual registry updates. Finally, we also did not have access
10 Behrouzi et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex-Based Differences in Severe COVID-19 Outcomes MAY 2023:100307

F I G U R E 1 Multivariable Adjusted Odds Ratios for Sex Differences in Severe Outcomes From COVID-19

A B

C D

Multivariable adjusted odds ratios (aORs) for sex differences in (A) the severe outcome composite of cardiovascular (CV) hospitalization, intensive care unit admission,
mechanical ventilation, or death; (B) all-cause hospitalization; (C) CV hospitalization; and (D) all-cause death. Adjusted odds ratios showing the gap in risk for men
compared to women of the same age were determined from a multivariable logistic regression model that incorporated public health unit (PHU)-specific random
effects. Models were adjusted for time (days from start of wave), sex, age group, age-sex interaction, rurality, income quintile, immigrant status, number of hos-
pitalizations/emergency department visits in the past year, frailty, history of hypertension, diabetes mellitus, coronary artery disease, heart failure, stroke, atrial
fibrillation, chronic kidney disease, HIV, organ transplantation, cancer, liver disease, lung disease, influenza vaccination since 2019/2020, COVID-19 vaccination (waves
2 and 3), PHU-level smoking rate (sex-specific and age-standardized), PHU-level obesity rate (sex-specific and age-standardized), and PHU-level visible minority rate
(sex-specific). There was a significant interaction between age group and sex for all models of all outcomes. aORs shown are for men within each age group; women
within each age group are the reference. The gridline corresponding to aOR ¼ 1.00, indicating no difference in risk between men and women of the same age, is
emphasized. Refer to Supplemental Table 4 for tabulated point estimates and 95% CIs.

to individual-level data on weight, smoking status, allow for more equitable care, in terms of sex and age,
income, and race/ethnicity; hence, we relied on among patients.51,52 These findings may also more
community-level variables as proxies, which may be specifically identify target populations for further
over- or underestimating impact at the individ- refining COVID-19 treatment algorithms, as well as
ual level. the future study of COVID-19 long-term effects and
sequelae, especially those of a CV nature.42
CONCLUSIONS
ACKNOWLEDGMENTS This study was supported by
Following laboratory-confirmed SARS-CoV-2 infec- ICES, which is funded in part by an annual grant from
tion, men exhibit poorer outcomes than women— the Ontario Ministry of Health (MOH) and the Minis-
particularly CV—in all age groups, although the try of Long-Term Care (MLTC). This work was also
magnitude varied by age. Given that adopting a sex- supported by the Ontario Health Data Platform
and age-informed perspective has improved patient (OHDP), a Province of Ontario initiative to support
care in other realms, it is crucial to account for this Ontario’s ongoing response to COVID-19 and its
heterogeneity in risk during early decision-making to related impacts. This document used data adapted
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Behrouzi et al 11
MAY 2023:100307 Sex-Based Differences in Severe COVID-19 Outcomes

F I G U R E 2 Total 30-Day Incidences of CV Event Hospitalizations and of Individual CV Event Types per 10,000 Persons Who Tested
Positive for SARS-CoV-2 During the First 3 Waves

Denominator is the total count of all persons who tested positive in each wave (wave 1, n ¼ 30,736; wave 2, n ¼ 199,132; and wave
3, n ¼ 186,131). Sum of incidences of individual CV event types do not equal to incidence of CV hospitalizations (blue bars) due to differences
in accounting for multiple instances of the same event type per person. AMI ¼ acute myocardial infarction; CV ¼ cardiovascular; HF ¼ heart
failure.

from the Statistics Canada Postal Code OM Conversion Public Health Unit data used for this manuscript were
File, which is based on data from Canada Post Cor- adapted with the permission of Public Health Ontario
poration, and/or data adapted from the Ontario Min- (PHO). PHO assumes no responsibility for the content
istry of Health Postal Code Conversion File, which of any publication resulting from translation/
contains data copied under license from Canada changes/adaptation of PHO documents by third
Post Corporation and Statistics Canada. Parts of this parties. The authors acknowledge that the clinical
material are based on data and information compiled registry data used in this publication is from partici-
and provided by the MOH and MLTC, Ontario Health, pating hospitals through CorHealth Ontario, which
Canadian Institute for Health Information (CIHI), serve as an advisory body to the MOH, is funded by
CorHealth Ontario, Immigration, Refugees and Citi- the MOH, and is dedicated to improving the quality,
zenship Canada (IRCC) (current to 2018-05-25), efficiency, access and equity in the delivery of the
Ontario Registrar General (ORG) information on continuum of adult cardiac, vascular and stroke ser-
deaths (the original source of which is Service- vices in Ontario, Canada. This study made use of the
Ontario), and ICES. The analyses, results, conclu- Johns Hopkins ACG System (Version 10) to identify
sions, opinions, and statements expressed herein are frailty conditions. The CENSUS (2016) and CCHS
those of the author(s) and do not reflect those of the (2007/08 onwards) datasets were adapted from Sta-
funding or data sources; no endorsement is intended tistics Canada. This does not constitute an endorse-
or should be inferred. No endorsement by the OHDP, ment by Statistics Canada of these products.
its partners, or the Province of Ontario is intended or
should be inferred. The views expressed therein are
FUNDING SUPPORT AND AUTHOR DISCLOSURES
those of the authors and do not necessarily reflect
those of ORG or the Ministry of Public and Business All authors have completed and submitted the ICMJE Form for
Service Delivery. The authors thank IQVIA Solutions Disclosure of Potential Conflicts of Interest. Ms Behrouzi has received
Canada Inc for use of their Drug Information File. The research grant support to her institutions from Boehringer Ingelheim,
12 Behrouzi et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex-Based Differences in Severe COVID-19 Outcomes MAY 2023:100307

Lilly, and Sanofi-Aventis, outside the submitted work. Dr Farkouh has and Department of Medicine, Women’s College Hospital. All other
received research grants from Amgen, Novartis, and Novo Nordisk, authors have reported that they have no relationships relevant to the
outside the submitted work. Dr Goodman reports research grant contents of this paper to disclose.
support (eg, steering committee or data and safety monitoring com-
mittee) and/or speaker/consulting honoraria (eg, advisory boards)
ADDRESS FOR CORRESPONDENCE: Jacob A. Udell,
from: Amgen, Anthos Therapeutics, AstraZeneca, Bayer, Boehringer
Ingelheim, Bristol Myers Squibb, CSL Behring, Daiichi-Sankyo/ Cardiovascular Division, Peter Munk Cardiac Centre,
American Regent, Eli Lilly, Esperion, Ferring Pharmaceuticals, HLS Toronto General Hospital and Women’s College
Therapeutics, JAMP Pharma, Merck, Novartis, Novo Nordisk A/C,
Hospital, University of Toronto, 76 Grenville Street,
Pendopharm/Pharmascience, Pfizer, Regeneron, Sanofi, Servier, Tol-
Toronto, Ontario M5S 1B2, Canada. E-mail: jay.udell@
mar Pharmaceuticals, Valeo Pharma; and salary support/honoraria
from the Heart and Stroke Foundation of Ontario/University of Tor- utoronto.ca. Twitter: @JayUdell.
onto (Polo) Chair, Canadian Heart Failure Society, Canadian Heart
Research Centre and MD Primer, Canadian VIGOUR Centre, Cleveland
Clinic Coordinating Centre for Clinical Research, Duke Clinical PERSPECTIVES
Research Institute, New York University Clinical Coordinating Centre,
PERFUSE Research Institute, TIMI Study Group (Brigham Health). Dr
Udell has received research grant support to his institutions from
COMPETENCY IN MEDICAL KNOWLEDGE: The
AstraZeneca, Novartis, and Sanofi; reported service as a consultant sex gap in the risk of CV hospitalization during the
for Amgen, Boehringer Ingelheim, Janssen, Merck, Novartis, Nova- acute phase of SARS-CoV-2 infection has either per-
vax, and Sanofi; and received honoraria from Boehringer Ingelheim
sisted or increased with each new wave, unlike with
and Janssen outside the submitted work. No other disclosures were
reported. This study received funding from a Canadian Institutes for other severe clinical outcomes.
Health Research (CIHR) Strategy for Patient-Oriented Research
Innovative Clinical Trial Multi-year grant (MYG-151211), a Ted Rogers COMPETENCY IN PATIENT CARE: In the first
Centre for Heart Research Innovation Fund - COVID-19 Award, and in month after a positive SARS-CoV-2 test, patient care
part by the Peter Munk Cardiac Centre Innovation Fund. Ms Behrouzi
should be tailored to better prevent CV events,
is supported by a CIHR Canada Graduate Scholarship – Doctoral (CGS
D) award, a University of Toronto MD/PhD studentship, and a Ted
particularly among younger men.
Rogers Centre for Heart Research Doctoral Award. Dr Atzema is sup-
ported by Mid-Career Investigator Awards from the Heart and Stroke
TRANSLATIONAL OUTLOOK 1: Future studies
Foundation, Sunnybrook Health Sciences Centre, and ICES. Dr Kapral should assess whether other jurisdictions also face a
holds the Lillian Love Chair in Women’s Health from the University similarly persistent or increasing sex gap in the risk of
Health Network/University of Toronto. Dr Kaul holds a CIHR Sex and
CV hospitalization with each subsequent wave of the
Gender Science Chair and a Heart and Stroke Foundation Chair in
Cardiovascular Research. Dr Goodman was supported by the Heart COVID-19 pandemic.
and Stroke Foundation of Ontario/University of Toronto Polo Chair.
Dr Austin is supported by a Mid-Career Investigator Award from the TRANSLATIONAL OUTLOOK 2: Further efforts are
Heart and Stroke Foundation. Dr McAlister holds the Alberta Health
needed to examine the clinical, biological, psychoso-
Services Chair in Cardiovascular Outcomes Research. Dr Lee is the
cial, and health system factors contributing to the
Ted Rogers Chair in Heart Function Outcomes, University Health
Network, University of Toronto. Dr Udell is supported by a Govern- growing disparity between the sexes in the risk of CV
ment of Ontario Early Researcher Award (ER15-11-037), a Clinician hospitalizations following COVID-19.
Scientist Merit Award, Department of Medicine, University of Tor-
onto, Peter Munk Cardiac Centre, Women’s College Research Institute

REFERENCES

1. Udell JA, Behrouzi B, Sivaswamy A, et al. Clinical cohort study. BMJ. 2020;371:m3731. https://doi. 8. Vahidy FS, Pan AP, Ahnstedt H, et al. Sex dif-
risk, sociodemographic factors, and SARS-CoV-2 org/10.1136/bmj.m3731 ferences in susceptibility, severity, and outcomes
infection over time in Ontario, Canada. Sci Rep. of coronavirus disease 2019: cross-sectional
5. Abate BB, Kassie AM, Kassaw MW, Aragie TG,
2022;12(1):10534. https://doi.org/10.1038/ analysis from a diverse US metropolitan area.
Masresha SA. Sex difference in coronavirus disease
s41598-022-13598-z PLoS One. 2021;16(1):e0245556. https://doi.org/
(COVID-19): a systematic review and meta-anal-
10.1371/journal.pone.0245556
2. Jun T, Nirenberg S, Weinberger T, et al. Analysis ysis. BMJ Open. 2020;10(10):e040129. https://
of sex-specific risk factors and clinical outcomes in doi.org/10.1136/bmjopen-2020-040129 9. Ge E, Li Y, Wu S, Candido E, Wei X. Association
COVID-19. Commun Med. 2021;1(1):3. https://doi. 6. Peters SAE, MacMahon S, Woodward M. of pre-existing comorbidities with mortality and
org/10.1038/s43856-021-00006-2 Obesity as a risk factor for COVID-19 mortality disease severity among 167,500 individuals with
3. Cunningham JW, Vaduganathan M, Claggett BL, in women and men in the UK biobank: COVID-19 in Canada: a population-based cohort
et al. Clinical outcomes in young US adults hos- comparisons with influenza/pneumonia and study. PLoS One. 2021;16(10):e0258154. https://
pitalized with COVID-19. JAMA Intern Med. coronary heart disease. Diabetes Obes Metab. doi.org/10.1371/journal.pone.0258154
2021;181(3):379. https://doi.org/10.1001/jamai- 2021;23(1):258–262. https://doi.org/10.1111/dom.
10. Scully EP, Schumock G, Fu M, et al. Sex and
nternmed.2020.5313 14199
gender differences in testing, hospital admission,
4. Clift AK, Coupland CAC, Keogh RH, et al. Living 7. Tereshchenko LG, Bishop A, Fisher-Campbell N, clinical presentation, and drivers of severe out-
risk prediction algorithm (QCOVID) for risk of et al. Risk of cardiovascular events after COVID-19. comes from COVID-19. Open Forum Infect Dis.
hospital admission and mortality from coronavirus Am J Cardiol. 2022;179:102–109. https://doi.org/ 2021;8(9):ofab448. https://doi.org/10.1093/ofid/
19 in adults: national derivation and validation 10.1016/j.amjcard.2022.06.023 ofab448
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Behrouzi et al 13
MAY 2023:100307 Sex-Based Differences in Severe COVID-19 Outcomes

11. Quaresima V, Scarpazza C, Sottini A, et al. Sex Mak. 2016;16(1):135. https://doi.org/10.1186/ MI registry. Am J Med. 2020;133(5):605–612.e1.
differences in a cohort of COVID-19 Italian pa- s12911-016-0375-3 https://doi.org/10.1016/j.amjmed.2019.10.020
tients hospitalized during the first and second
23. Statistics Canada. Health regions: boundaries 35. Udell JA, Fonarow GC, Maddox TM, et al.
pandemic waves. Biol Sex Differ. 2021;12(1):45.
and correspondence with census geography. 2021. Sustained sex-based treatment differences in
https://doi.org/10.1186/s13293-021-00386-z
Accessed April 1, 2023. https://www150.statcan. acute coronary syndrome care: insights from the
12. Behrouzi B, Araujo Campoverde MV, Liang K, gc.ca/n1/en/catalogue/82-402-X American Heart Association get with the guide-
et al. Influenza vaccination to reduce cardiovas- 24. The Johns Hopkins University Bloomberg lines coronary artery disease registry. Clin Cardiol.
cular morbidity and mortality in patients with School of Public Health. The Johns Hopkins 2018;41(6):758–768. https://doi.org/10.1002/clc.
COVID-19. J Am Coll Cardiol. 2020;76(15):1777– Adjusted Clinical Group (ACG) system. 2013. 22938
1794. https://doi.org/10.1016/j.jacc.2020.08.028 Accessed April 1, 2023. https://www.hopkinsacg. 36. Khan NA, Daskalopoulou SS, Karp I, et al. Sex
13. Modin D, Claggett B, Sindet-Pedersen C, et al. org/ differences in acute coronary syndrome symptom
Acute COVID-19 and the incidence of ischemic 25. Austin PC, van Walraven C, Wodchis WP, presentation in young patients. JAMA Intern Med.
stroke and acute myocardial infarction. Circulation. Newman A, Anderson GM. Using the Johns 2013;173(20):1863–1871. https://doi.org/10.1001/
2020;142(21):2080–2082. https://doi.org/10. Hopkins Aggregated Diagnosis Groups (ADGs) jamainternmed.2013.10149
1161/CIRCULATIONAHA.120.050809 to predict mortality in a general adult population
37. Kaul P, Chang W-C, Westerhout CM,
cohort in Ontario, Canada. Med Care. 2011;49
14. Rey JR, Caro-Codón J, Rosillo SO, et al. Heart Graham MM, Armstrong PW. Differences in
(10):932–939. https://doi.org/10.1097/MLR.
failure in COVID-19 patients: prevalence, incidence admission rates and outcomes between men and
0b013e318215d5e2
and prognostic implications. Eur J Heart Fail. women presenting to emergency departments
2020;22(12):2205–2215. https://doi.org/10.1002/ 26. Ontario Agency for Health Protection and with coronary syndromes. Can Med Assoc J.
ejhf.1990 Promotion (Public Health Ontario). Snapshots: 2007;177(10):1193–1199. https://doi.org/10.1503/
self-reported smoking snapshot (2015-2016). cmaj.060711
15. Qureshi AI, Baskett WI, Huang W, et al. Acute
2018. Accessed April 1, 2023. https://www.
ischemic stroke and COVID-19. Stroke. 2021;52(3): 38. Vaccarino V, Rathore SS, Wenger NK, et al. Sex
publichealthontario.ca/en/data-and-analysis/
905–912. https://doi.org/10.1161/STROKEAHA. and racial differences in the management of acute
substance-use/smoking-status
120.031786 myocardial infarction, 1994 through 2002. N Engl
27. Ontario Agency for Health Protection and J Med. 2005;353(7):671–682. https://doi.org/10.
16. Musikantow DR, Turagam MK, Sartori S, et al. Promotion (Public Health Ontario). Snapshots: 1056/NEJMsa032214
Atrial fibrillation in patients hospitalized with self-reported nutrition and healthy weights (2015-
COVID-19: incidence, predictors, outcomes and 39. Vaccarino V, Horwitz RI, Meehan TP,
2016). 2018. Accessed April 1, 2023. https://www.
comparison to influenza. J Am Coll Cardiol EP. Petrillo MK, Radford MJ, Krumholz HM. Sex dif-
publichealthontario.ca/en/data-and-analysis/
2021;7(9):1120–1130. https://doi.org/10.1016/j. ferences in mortality after myocardial infarction.
health-behaviours/nutrition-and-healthy-weights
jacep.2021.02.009 Arch Intern Med. 1998;158(18):2054. https://doi.
28. Ontario Agency for Health Protection and org/10.1001/archinte.158.18.2054
17. Colon CM, Barrios JG, Chiles JW, et al. Atrial Promotion (Public Health Ontario). Snapshots: so-
40. Kwong JC, Schwartz KL, Campitelli MA, et al.
arrhythmias in COVID-19 patients. J Am Coll Car- cial determinants of health snapshot (2016). 2019.
Acute myocardial infarction after laboratory-
diol EP. 2020;6(9):1189–1190. https://doi.org/10. Accessed April 1, 2023. https://www.
confirmed influenza infection. N Engl J Med.
1016/j.jacep.2020.05.015 publichealthontario.ca/en/data-and-analysis/health-
2018;378(4):345–353. https://doi.org/10.1056/
equity/sdoh
18. Minhas AS, Shade JK, Cho S-M, et al. The role NEJMoa1702090
of sex and inflammation in cardiovascular out- 29. Statistics Canada. Ethnic and cultural origins of
41. Chang T-Y, Chao T-F, Liu C-J, et al. The asso-
comes and mortality in COVID-19. Int J Cardiol. Canadians: portrait of a rich heritage. Census in
ciation between influenza infection, vaccination,
2021;337:127–131. https://doi.org/10.1016/j.ijcard. brief. 2017. Accessed April 1, 2023. https://
and atrial fibrillation: a nationwide case-control
2021.05.011 www12.statcan.gc.ca/census-recensement/2016/
study. Hear Rhythm. 2016;13(6):1189–1194.
as-sa/98-200-x/2016016/98-200-x2016016-
19. Benjamin EJ, Virani SS, Callaway CW, et al. https://doi.org/10.1016/j.hrthm.2016.01.026
eng.cfm
Heart disease and stroke statistics—2018 update: a
30. Statistics Canada. Visible minority. Dictionary, 42. Xie Y, Xu E, Bowe B, Al-Aly Z. Long-term
report from the American Heart Association. Cir-
census of population, 2016. 2017. Accessed April 1, cardiovascular outcomes of COVID-19. Nat Med.
culation. 2018;137(12). https://doi.org/10.1161/
2023. https://www12.statcan.gc.ca/census- 2022;28(3):583–590. https://doi.org/10.1038/
CIR.0000000000000558
recensement/2016/ref/dict/pop127-eng.cfm s41591-022-01689-3
20. Tu JV, Chu A, Donovan LR, et al. The cardio-
31. Buchan SA, Chung H, Brown KA, et al. Effec- 43. Haitao T, Vermunt JV, Abeykoon J, et al.
vascular health in ambulatory care research team
tiveness of COVID-19 vaccines against Omicron or COVID-19 and sex differences. Mayo Clin Proc.
(CANHEART): using big data to measure and
Delta symptomatic infection and severe outcomes. 2020;95(10):2189–2203. https://doi.org/10.1016/
improve cardiovascular health and healthcare
JAMA Netw Open. 2022;5(9):e2232760. https:// j.mayocp.2020.07.024
services. Circ Cardiovasc Qual Outcomes.
2015;8(2):204–212. https://doi.org/10.1161/CIR- doi.org/10.1001/jamanetworkopen.2022.32760 44. Qi S, Ngwa C, Morales Scheihing DA, et al. Sex
COUTCOMES.114.001416 32. Ray JG, Schull MJ, Vermeulen MJ, Park AL. differences in the immune response to acute
Association between ABO and Rh blood groups COVID-19 respiratory tract infection. Biol Sex
21. Ontario Agency for Health Protection and Pro-
and SARS-CoV-2 infection or severe COVID-19 Differ. 2021;12(1):66. https://doi.org/10.1186/
motion (Public Health Ontario). Confirmed SARS-
illness. Ann Intern Med. 2021;174(3):308–315. s13293-021-00410-2
CoV-2 variant of concern cases from November 29,
2020 to November 12, 2021. 2022. Accessed April 1, https://doi.org/10.7326/M20-4511 45. Reitsma MB, Fullman N, Ng M, et al. Smoking
2023. https://www.publichealthontario.ca/-/ 33. Chen N, Zhou M, Dong X, et al. Epidemiolog- prevalence and attributable disease burden in 195
media/Documents/nCoV/epi/2022/03/covid-19- ical and clinical characteristics of 99 cases of 2019 countries and territories, 1990–2015: a systematic
confirmed-voc-cases-epi-summary.pdf?rev¼2 novel coronavirus pneumonia in Wuhan, China: a analysis from the Global Burden of Disease Study
df81ad72dd84796b79451613db57f77&sc_lang¼en descriptive study. Lancet. 2020;395(10223):507– 2015. Lancet. 2017;389(10082):1885–1906.
513. https://doi.org/10.1016/S0140-6736(20) https://doi.org/10.1016/S0140-6736(17)30819-X
22. Chiu M, Lebenbaum M, Lam K, et al. Describing
30211-7
the linkages of the immigration, refugees and 46. Mauvais-Jarvis F, Bairey Merz N, Barnes PJ,
citizenship Canada permanent resident data and 34. Yang J, Biery DW, Singh A, et al. Risk factors et al. Sex and gender: modifiers of health, disease,
vital statistics death registry to Ontario’s Admin- and outcomes of very young adults who experi- and medicine. Lancet. 2020;396(10250):565–582.
istrative Health Database. BMC Med Inform Decis ence myocardial infarction: the partners YOUNG- https://doi.org/10.1016/S0140-6736(20)31561-0
14 Behrouzi et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex-Based Differences in Severe COVID-19 Outcomes MAY 2023:100307

47. Stall NM, Wu W, Lapointe-Shaw L, et al. Sex- and risk of death in patients admitted to hospital 52. Inouye SK. Creating an anti-ageist health-
and age-specific differences in COVID-19 testing, with covid-19 and seasonal influenza: cohort care system to improve care for our current
cases, and outcomes: a population-wide study in study. BMJ. 2020;371:m4677. https://doi.org/10. and future selves. Nat Aging. 2021;1(2):
Ontario, Canada. J Am Geriatr Soc. 2020;68(10): 1136/bmj.m4677 150–152. https://doi.org/10.1038/s43587-020-
2188–2191. https://doi.org/10.1111/jgs.16761 00004-4
50. Griffith GJ, Morris TT, Tudball MJ, et al.
48. Katsoularis I, Fonseca-Rodríguez O, Collider bias undermines our understanding of
Farrington P, Lindmark K, Fors Connolly A-M. Risk COVID-19 disease risk and severity. Nat Commun.
KEY WORDS cardiovascular disease, health
of acute myocardial infarction and ischaemic 2020;11(1):5749. https://doi.org/10.1038/s41467-
administrative data, PCR, population-based,
stroke following COVID-19 in Sweden: a self- 020-19478-2
SARS-CoV-2, vaccination
controlled case series and matched cohort study.
51. Bartz D, Chitnis T, Kaiser UB, et al. Clinical
Lancet. 2021;398(10300):599–607. https://doi.
advances in sex- and gender-informed medicine to
org/10.1016/S0140-6736(21)00896-5
improve the health of all. JAMA Intern Med. A PPE NDI X For supplemental tables and
49. Xie Y, Bowe B, Maddukuri G, Al-Aly Z. 2020;180(4):574. https://doi.org/10.1001/jamai- figures, please see the online version of this
Comparative evaluation of clinical manifestations nternmed.2019.7194 paper.
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY LICENSE (http://creativecommons.org/licenses/by/4.0/).

EDITORIAL COMMENT

Sex and Age

2 Predictors of Cardiovascular Events With COVID-19 Infection*

Gina Lundberg, MD, Olga Toleva, MD, Alexis Cutchins, MD

T he pandemic from COVID-19 from SARS-

CoV-2 infection has been with us for 3 years
in North America. Many healthcare workers
have experienced burnout and exhaustion as a
analysis was extremely
known confounding factors.
careful to

We commend the authors’ work evaluating sex

and age differences in CV events, including all-
account for

result. Almost one-half of all public health workers cause hospitalizations and all-cause deaths, in the
have quit their jobs or retired in the last 5 years, acute setting of SARS-CoV-2 infection. Sex differ-
many due to COVID-19.1 Many healthcare workers ences in severity and outcomes in COVID-19 in-
are experiencing “COVID fatigue” at this point. Yet fections have been evident since the first outbreak
if we are to truly put COVID-19 in the rear-view in North America. 3 Male sex has been indepen-
mirror, we must continue to study COVID-19 infec- dently associated with hospitalization, need for
tions to determine best practices for preventing intensive care, ventilatory support, and death. 4
and treating infections to reduce cardiovascular COVID-19 severity and mortality have also varied
(CV) events. between sexes based on age and other chronic
The paper Behrouzi et al2 in this issue of JACC: conditions. CV risk factors including hypertension,
Advances reveals data from a massive data base in a diabetes, heart failure, coronary artery disease as
country with socialized medicine across the first 3 well as chronic lung and kidney disease have been
waves of the pandemic. This study is a seminal associated with worse outcomes with COVID-19 as
example of a well-conducted and very large well. Multiple studies have shown these findings in
population-based epidemiological study using Europe, Japan, and globally. 5
strong captured administrative data. Although such The main finding in this study is that males were
data may lack clinical risk adjusters such as oxygen consistently, through all age strata, at higher risk of
saturation, symptomatology, respiratory rate, it is adverse outcomes including CV events. The authors
still highly valuable for answering the research demonstrate increased sex-specific risk for males
question. As highlighted rightfully by the authors, over females during all 3 waves of the pandemic,
they included almost all patients in Ontario with noting this was not due to disparities in testing. In
laboratory confirmed COVID-19 infection, had access fact, more females than males were tested during the
to gold standard test results, and focused analyses first wave. Across all 3 waves, males experienced
on highly identifiable outcomes of interest that are more severe disease and fatality than females after
nonsusceptible to ascertainment biases. Statistical adjusting for comorbidities and socioeconomic
factors.
Males remain more adversely affected overall in
*Editorials published in JACC: Advances reflect the views of the authors
the acute phase of infection for all populations
and do not necessarily represent the views of JACC: Advances or the except one (females ages 18-45 years old during the
American College of Cardiology. second and third waves of the pandemic). Inter-
From the Division of Cardiology, Department of Medicine, Emory estingly, the population most affected by long
University School of Medicine, Atlanta, Georgia, USA. COVID and lingering symptoms is younger females,
The authors attest they are in compliance with human studies commit-
although males are affected at a lower extent.6
tees and animal welfare regulations of the authors’ institutions and Food
and Drug Administration guidelines, including patient consent where Subramanian et al 6 describe an average age of
appropriate. For more information, visit the Author Center. 43.8 years old and 55.3% females in their study of

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100325

2 Lundberg et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Sex and Age: 2 Predictors of Cardiovascular Events With COVID-19 Infection MAY 2023:100325

long COVID. This increase in hospitalizations in heart failure, stroke, and acute myocardial
younger females during the second and third waves infarction.
is potentially the onset of symptoms of long COVID Finally, the most important and reassuring
that may warrant hospitalization. The time delay finding to us is that all-cause mortality decreased
from the positive COVID-19 test and the hospitali- substantially by the third wave, and the male
zation episode would be key to answering this dominance in death profile tended to disappear. In
question but is lacking in the analysis. While a each new wave, sex disparities reduced for all out-
diagnosis of long COVID cannot be entertained until comes except for CV hospitalization. In the current
>4 weeks after infection, patients can develop the era of herd immunity and multiple vaccines, it is
start of these prolonged symptoms immediately af- impossible to generalize the study findings and
ter infection and may even be hospitalized initially predict the future. Nevertheless, this study showed
during that 30-day post infection period. the value of vaccination to reduce adverse out-
When comparing hospitalizations and admissions comes and bringing males’ outcomes similar to fe-
for CV events across the 3 waves, it is interesting to males for COVID-related outcomes. In this stage of
see a consistent difference between males and fe- COVID fatigue and vaccine resistance, non-
males. There is a substantial drop in hospitaliza- vaccinated individuals of all ages should be
tions for both males and females between the first informed that males have a poorer prognosis and
wave and the second and third waves across the 46 higher risk with COVID-19 infection than females.
to 85-year-old age groups, which is shown in the Physicians and healthcare workers should remain
Central Illustration of Behrouzi et al2 . The youngest active and engaged in educating patients on the
and the oldest patients appear to be outliers. This life-saving value of vaccines.
may result from the vaccine being less effect in The actionable results from this study are to
those over 80 years of age and more effective in the encourage men of all ages, aging women, frail pa-
younger group and underscore the importance of tients, and those with recent hospitalizations,
age and pre-existing conditions in relation to the hypertension, diabetes, heart failure, atrial fibrilla-
severity of the infection. 7,8 The findings support the tion, chronic kidney disease, cancer, and lung dis-
idea that the older age groups (>80 years) may not ease to remain current with vaccinations and follow
7,9
have had as robust of a response to the vaccine. preventive measures to avoid COVID-19 infection.
Of note, one study looking at sex differences in And these patients should seek medical care at
efficacy of the COVID-19 vaccine showed higher ef- the first signs of arrhythmias, heart failure, stroke,
ficacy in males over females which may speculate or heart attack. Despite tremendous public health
that the sex-specific findings of Behrouzi et al 2 may efforts, COVID-19 is still virulent and remains a
have been even more significant had a vaccine not significant health threat with CV events that still
been developed.9 lead to hospitalization, morbidity, and even
Importantly, this study confirms that across the mortality.
first 3 waves of COVID-19 infections, the most se-
vere outcomes were seen among older patients, FUNDING SUPPORT AND AUTHOR DISCLOSURES
more male patients, patients with higher numbers
The authors have reported that they have no relationships relevant to
of comorbidities and among frail patients. Males the contents of this paper to disclose.
experienced higher rates of severe outcomes
compared with females with severe outcomes
increasing with age for both sexes. The incidence of ADDRESS FOR CORRESPONDENCE: Prof Gina Lund-
CV hospitalization across the first 3 waves revealed berg, Division of Cardiology, Department of Medicine,
males ages 46 to 65 years were at the highest risk, Emory University School of Medicine, 137 Johnson
3-fold higher than seen in females, with the most Ferry Road, Suite 1200, Marietta, Georgia 30068, USA.
common CV hospitalizations due to arrhythmias, E-mail: [email protected].
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Lundberg et al 3
MAY 2023:100325 Sex and Age: 2 Predictors of Cardiovascular Events With COVID-19 Infection

REFERENCES

1. Leider J. The exodus of state and local public and mortality. J Womens Health (Larchmt). 8. Collier DA, Ferreira I, Kotagiri P, et al. Age-
health employees: separations started before and 2021;30(5):646–653. related immune response heterogeneity to
continued throughout COVID-19. Health Aff. SARS-CoV-2 vaccine BNT162b2. Nature. 2021;
5. Matsumoto S, Noda S, Torii S, et al. Sex differences
2023;42(3):338–348. 596(7872):417–422.
in clinical outcomes among patients with COVID-19
2. Behrouzi B, Sivaswamy A, Chu A, et al. Sex-based and cardiovascular disease - insights from the 9. Bignucolo A, Scarabel L, Mezzalira S,
differences in severe outcomes, including cardio- CLAVIS-COVID registry. Circ Rep. 2022;4(7):315–321. Polesel J, Cecchin E, Toffoli G. Sex disparities
vascular hospitalization, in adults with COVID-19 in in efficacy in COVID-19 vaccines: a systematic
6. Subramanian A, Nirantharakumar K, Hughes S,
Ontario, Canada. JACC: Adv. 2023;2:100307. review and meta-analysis. Vaccines. 2021;9(8):
et al. Symptoms and risk factors for long COVID in
3. Peckham H, de Gruijter NM, Raine C, et al. Male non-hospitalized adults. Nat Med. 2022;28(8): 825.
sex identified by global COVID-19 meta-analysis as 1706–1714.
a risk factor for death and ITU admission. Nat
7. Bates TA, Leier HC, Lyski ZL, et al. Age-depen-
Commun. 2020;11(1):6317.
dent neutralization of SARS-CoV-2 and P.1 variant KEY WORDS cardiovascular events,
4. Gomez JMD, Du-Fay-de-Lavallaz JM, Fugar S, by vaccine immune serum samples. JAMA. cardiovascular risk factors, COVID-19, sex-specific
et al. Sex differences in COVID-19 hospitalization 2021;326(9):868–869. outcomes
JACC: ADVANCES VOL. 2, NO. 3, 2023

ª 2023 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY LICENSE (http://creativecommons.org/licenses/by/4.0/).

ORIGINAL RESEARCH

OUTCOMES AND QUALITY

Prognostic Value of Self-Reported

Subjective Exercise Capacity in
Patients With Acute Dyspnea
Maria Belkin, MD,a,b,c,* Desiree Wussler, MD,a,b,c,* Eleni Michou, MD,a,c Ivo Strebel, MS,a Nikola Kozhuharov, MD,a,c,d
Zaid Sabti, MD,a,c Albina Nowak, MD,e,f Samyut Shrestha, MD,a,c Pedro Lopez-Ayala, MD,a,c
Alexandra Prepoudis, MD,a,c Sabrina Stefanelli, MD,a,c Ibrahim Schäfer, MD,a,b,c Constantin Mork, MD,a,c
Miriam Albus, MD,a,b,c Isabelle Danier, MD,a,c Cornelia Simmen, MD,a,c Tobias Zimmermann, MD,a,c
Matthias Diebold, MD,a,c Tobias Breidthardt, MD,a,b,c Christian Mueller, MDa,c

ABSTRACT

BACKGROUND Self-reported exercise capacity is a well-established prognostic measure in stable ambulatory patients
with cardiac and pulmonary disease.

OBJECTIVES The authors aimed to directly compare the prognostic accuracy of quantified self-reported exercise ca-
pacity using the Duke Activity Status Index (DASI) with the established objective disease-severity marker B-type natri-
uretic peptide (BNP) in patients presenting with acute dyspnea to the emergency department.

METHODS The DASI was obtained in a prospective multicenter diagnostic study recruiting unselected patients
presenting with acute dyspnea to the emergency department. The prognostic accuracy of DASI and BNP for 90-day and
720-day all-cause mortality was evaluated using C-index.

RESULTS Among 1,019 patients eligible for this analysis, 75 (7%) and 297 (29%) patients died within 90 and 720 days
after presentation, respectively. Unadjusted hazard ratios (HRs) and multivariable adjusted hazard ratios (aHRs) for
90- and 720-day mortality increased continuously from the fourth (best self-reported exercise capacity) to the first DASI
quartile (worst self-reported exercise capacity). For 720-day mortality the HR of the first quartile vs the fourth was 9.1
(95% CI, 5.5-14.9) vs (aHR: 6.1, 95% CI: 3.7-10.1), of the second quartile 6.4 (95% CI: 3.9-10.6) vs (aHR: 4.4, 95% CI:
2.6-7.3), while of the third quartile the HR was 3.2 (95% CI: 1.9-5.5) vs (aHR: 2.4, 95% CI: 1.4-4.0). The prognostic
accuracy of the DASI score was high, and higher than that of BNP concentrations (720-day mortality C-index: 0.67 vs
0.62; P ¼ 0.024).

CONCLUSIONS Quantification of self-reported subjective exercise capacity using the DASI provides high
prognostic accuracy and may aid physicians in risk stratification. (Basics in Acute Shortness of Breath EvaLuation
[BASEL V] Study [BASEL V]; NCT01831115) (JACC Adv 2023;2:100342) © 2023 The Authors. Published by Elsevier
on behalf of the American College of Cardiology Foundation. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

From the aCardiovascular Research Institute Basel (CRIB) and Department of Cardiology, University Hospital Basel, University of
Basel, Basel, Switzerland; bDepartment of Internal Medicine, University Hospital Basel, Basel, Switzerland; cGREAT network,
Rome, Italy; dDepartment of Cardiology, Liverpool Heart and Chest Hospital, Liverpool, United Kingdom eDepartment of Endo-
crinology and Clinical Nutrition, University Hospital Zurich, Zurich, Switzerland; and the fDivision of Internal Medicine, University
Psychiatry Clinic Zurich, Zurich, Switzerland. *Drs Belkin and Wussler have contributed equally and should be considered first
author.

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2023.100342

2 Belkin et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Self-Reported Exercise Capacity in Acute Dyspnea MAY 2023:100342

O
ABBREVIATIONS bjectively quantified exercise ca- included irrespective of renal function, whereas pa-
AND ACRONYMS pacity is a well-established prog- tients with terminal kidney failure on chronic dialysis
nostic factor in stable ambulatory were excluded. For this analysis, patients were
aHR = adjusted hazard ratio
patients with cardiovascular and/or pulmo- eligible if they had completed the DASI questionnaire
AUROC = area under the
nary diseases.1-3 For example, peak oxygen within the first days after presentation.
receiver operating
characteristic curve uptake measured by cardiopulmonary exer- The investigation conforms with the principles
BNP = B-type natriuretic cise testing and the 6-minute walk test aid outlined in the Declaration of Helsinki and the study
peptide physicians in the risk stratification of ambu- was approved by the local ethics committee. The au-
BUN = blood urea nitrogen latory patients with cardiac or pulmonary thors designed the study, gathered, and analyzed the
COPD = chronic obstructive disease.3-5 Quantitative assessment of self- data according to the STARD (Standards for Reporting
pulmonary disease reported subjective exercise capacity has Diagnostic accuracy studies) guidelines for studies of
DASI = Duke Activity Status been evaluated as a simple alternative tool diagnostic accuracy.
Index
to potentially even better reflect the impact QUANTIFIED SELF-REPORTED EXERCISE CAPACITY:
ED = emergency department of exercise capacity on health-related qual- DASI. After patients had provided written informed
NT-proBNP = N-terminal pro- ity of life.6-8 Likely the best validated one consent, they received paper forms with 12 questions
B-type natriuretic peptide
is the Duke Activity Status Index (DASI), a regarding exercise capacity during daily life repre-
simple self-assessment tool with 12 questions for senting the DASI and were asked to complete the
estimating exercise capacity. 9 DASI-scores, with forms within the first days of hospitalization.
higher values indicating higher exercise capacity, Initially, the original 12-item DASI version was
correlate very well with peak oxygen uptake and translated into German and used (Figure 1).9 After
are validated measures of functional status, particu- 2 patients were severely offended by the question
larly in outpatients with heart failure and in the pre- regarding sexual activities, in consultation with the
operative setting. 9,10 ethics committee we decided to omit this question
It is unknown, whether beyond providing impor- from all further questionnaires. The maximum score
tant insights regarding functional health-related of this modified version of the DASI, which was later
quality of life, quantitative assessment of self- applied for all patients in this analysis, was 52.95 vs
reported subjective exercise capacity using the DASI 58.2 in the original version.
may also help physicians in the risk stratification of
ADJUDICATION OF FINAL DIAGNOSIS. The final
patients presenting to the emergency department
diagnosis of the main disorder responsible for acute
(ED) with acute dyspnea due to either acute heart
dyspnea was centrally adjudicated by 2 independent
failure (AHF) or pulmonary disease. Given the lack of
cardiologists/internists who had access to all pa-
established disease-independent tools for accurately
tients’ medical records including clinical history,
risk stratifying patients with acute dyspnea, the DASI
physical examination, 12-lead electrocardiogram,
may well have clinical utility. Therefore, we aimed to
laboratory findings, chest x-ray, echocardiography,
test this hypothesis by directly comparing the prog-
lung function testing, computed tomography, the
nostic accuracy of quantified self-reported exercise
response to therapy, and also autopsy data for pa-
capacity using the DASI with the established objective
tients who had died in hospital. All laboratory find-
disease-severity marker B-type natriuretic peptide
ings obtained through the clinician’s routine
(BNP) in a large prospective multicenter study.
diagnostic workup were available for this study.
METHODS These findings included one of the natriuretic pep-
tides (BNP or N-Terminal pro-BNP [NT-proBNP]) that
STUDY DESIGN AND POPULATION. BASEL V (Basics current guidelines recommended for diagnosing AHF
in Acute Shortness of Breath EvaLuation) was a pro- with a Class I recommendation. 1,2 In situations of
spective, multicenter, diagnostic, and prognostic disagreement about the final diagnosis, cases were
study enrolling adult patients presenting with acute reviewed and adjudicated by a third cardiologist.
dyspnea to the ED in 2 University Hospitals in FOLLOW-UP. Patients were contacted 3, 12, and
Switzerland (Basel & Zurich). 11-15 Patients were 24 months after discharge by telephone or in written

The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received September 26, 2022; revised manuscript received January 19, 2023, accepted February 9, 2023.
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Belkin et al 3
MAY 2023:100342 Self-Reported Exercise Capacity in Acute Dyspnea

F I G U R E 1 Original Duke Activity Status Index

12 original DASI questions and corresponding points which result in the DASI-score ranging from 0 (lowest) to 58.2 (highest exercise capacity).
DASI ¼ Duke Activity Status Index.

form. Specific questions regarding dyspnea, possible testing was not performed. Statistical analysis was
rehospitalizations, and relevant diagnostics (eg, operated using SPSS for Windows 26.0 (SPSS Inc) and
electrocardiogram, echocardiography .) performed R statistical Software Version 3.4.3 (MathSoft).
after discharge were assessed. Furthermore, patients DASI-SCORE FOR RISK STRATIFICATION. Unad-
were asked to fill out another DASI questionnaire justed and multivariable adjusted Cox regression
which was sent in paper form. Information regarding analyses were performed to quantify the independent
death during follow-up was obtained from the hos- effect of the DASI on 90- and 720-day all-cause mor-
pital medical records, the general practitioner, and tality. The DASI was considered as a continuous score
the national mortality registry. as well as stratified into quartiles with the fourth
GENERAL STATISTICAL METHODS. Continuous var- DASI-quartile including patients with the highest
iables are presented as median (IQR) and categorical exercise capacity and serving as reference group.
variables are expressed by numbers (percentages). When treated as a continuous variable DASI score was
Linear regression and Cochran–Armitage tests were reversed, thus a higher score indicated lower exercise
used to calculate P values for trend across the quar- capacity. The Cox regression model of 90-day mor-
tiles for continuous and categorical characteristics, tality was adjusted for previously published cova-
respectively.16,17 All hypothesis testing was 2-tailed riates, further referred to as the “compact model”,
and P values of <0.05 were considered to indicate including age, the natural logarithm blood urea ni-
statistical significance. Adjustment for multiple trogen, hemoglobin level, the natural logarithm of
4 Belkin et al JACC: ADVANCES, VOL. 2, NO. 3, 2023

Self-Reported Exercise Capacity in Acute Dyspnea MAY 2023:100342

NT-proBNP concentration at presentation, beta- analysis (“rmda”) calculating the clinical net benefit
blocker intake on entry, and besides, sex was (true-positive counts minus false-positive counts
added. 18 Higher event rate allowed to additionally weighted by the respective threshold probability). For
adjust the Cox regression model of 720-day mortality, a specific threshold probability, a larger net benefit
based on relevant clinical findings and medical indicates a greater number of true-positive pre-
knowledge, to: systolic blood pressure and peripheral dictions without increase of false positives. Possible
oxygen saturation at presentation, leg edema on time dependencies were studied using time-
entry, history of hypertension, diabetes mellitus, dependent area under the receiver operating charac-
coronary artery disease, atrial fibrillation, previous teristic curve which accommodates censored
heart failure and obstructive pulmonary disease, data (“timeROC”).28
serum creatinine and sodium level, intake of angio- SUBGROUP ANALYSES. Finally, a subgroup analysis
tensin converting enzyme inhibitors/angiotensin re- was performed in patients with an adjudicated final
ceptor blockers, and diuretics.18,19 Missing predictor diagnosis of AHF. Interaction tests were conducted
values were imputed using the Markov Chain Monte between gender and the DASI score to evaluate a
Carlo method. The number of imputed datasets was potential sex-specific effect on mortality in general
20. The Cox regression analysis was performed on all and in inpatients with adjudicated AHF in particular.
20 data sets and the results were pooled using
Rubin’s rules.20 Findings were confirmed in a sensi-
tivity analysis using the original data. Schoenfeld’s RESULTS
global test was used to test the proportional hazards
assumption of the Cox regression models. Cox Among 2,153 patients enrolled in BASEL V, 1,019 pa-
regression models were internally bootstrap vali- tients with a median age of 74 years comprising 43%
dated. Kaplan–Meier curves of 90- and 720-day all- women were eligible for this analysis, as they
cause mortality were plotted, and comparisons were completed the DASI questionnaire shortly after pre-
performed by log-rank tests. Censored patients were sentation (median, 2 days; IQR: 1-5 days)
displayed in tables including number at risk with the (Supplemental Figure 1). Overall, baseline patient
use of “survminer” package R statistical software. characteristics were comparable among patients
completing the DASI questionnaire vs those who did
PROGNOSTIC ACCURACY. Secondary analysis included
not (Supplemental Table 1). The median-modified
the comparison of the prognostic accuracy between
DASI score was 24.95 (IQR: 15.45-39.45). The most
the continuous DASI score and BNP, an established
common adjudicated final diagnosis as the cause of
objective disease-severity marker in patients pre-
acute dyspnea was AHF in 529 (52%) patients. Table 1
senting with acute dyspnea to the ED (primary anal-
displays the baseline characteristics of patients
ysis), as well as between the DASI score and
included in this analysis and Supplemental Table 2
NT-proBNP, which in addition to quantifying hemo-
shows an overview of the covariates with missing
dynamic cardiac stress similar to BNP seems also to be
values and summary statistics.
associated with renal dysfunction as another prog-
13,21-25
nostic variable. The prognostic accuracy of the DASI SCORE FOR RISK STRATIFICATION. Within 90
DASI score was also compared to a previously and 720 days of follow-up, 75 (7%) and 297 (29%)
described risk score derived from the compact mor- patients died, respectively. The unadjusted hazard
tality model, further referred to as the “Voors score”, ratios (HRs) and the adjusted hazard ratios (aHR) for
ranging from 0 to 5.18 Calculations were performed in 90- and 720-day mortality increased continuously
the original, not imputed data. Moreover, the internal from the fourth quartile (best self-reported exercise
bootstrap validation of the fully adjusted model (19 capacity reference, DASI score: 40.45-52.95, n ¼ 251)
variables plus DASI score) was performed in a single to the third quartile (DASI score: 24.95 < 40.45,
imputed dataset. All prognostic accuracies were re- n ¼ 268), to the second quartile (DASI score
ported as optimism corrected C-indices. Correlated 15.45 < 24.95, n ¼ 248) up to the first quartile (worst
C-indices were compared using the “compareC” self-reported exercise capacity, DASI score: 0-15.25,
package R statistical software.26 Calibration curves n ¼ 252) (Figure 2, Central Illustration, Tables 2 and 3).
showed observed vs predicted probability for 720-day For 720-day mortality the HR of the first vs the fourth
all-cause mortality of BNP level on admission, the quartile was 9.1 (95% CI: 5.5-14.9) and the aHR 6.1
Voors score, the DASI quartiles, and the DASI score (95% CI: 3.7-10.1), the HR of the second quartile was
are part of the supplemental material (“rms”).27 6.4 (95% CI: 3.9-10.6) and the aHR: 4.4 (95% CI: 2.6-
Clinical usefulness was assessed by a decision curve 7.3), while the HR of the third was 3.2 (95% CI: 1.9-5.5)
JACC: ADVANCES, VOL. 2, NO. 3, 2023 Belkin et al 5
MAY 2023:100342 Self-Reported Exercise Capacity in Acute Dyspnea

T A B L E 1 Baseline Characteristics in the Overall Cohort Grouped According to DASI Quartiles

All Patients First Quartile Second Quartile Third Quartile Fourth Quartile P Value for
(N ¼ 1,019) (n ¼ 252) (n ¼ 248) (n ¼ 268) (n ¼ 251) Trend

DASI-score 24.95 (15.45-39.45) 7.25 (4.5-12.7) 18.95 (15.45-18.95) 31.45 (26.95-34.95) 52.95 (45.45-52.95) <0.001
Age, y 74 (61-82) 75 (65-82) 76 (65-83) 76 (63-82) 66 (52-77) <0.001
Female 436 (43) 117 (46) 110 (44) 114 (43) 95 (38) 0.049
Body mass index, kg/m2 26 (22-30) 26 (22-31) 25 (22-30) 26 (22-30) 26 (23-30) 0.708
History
Hypertension 689 (68) 196 (78) 177 (71) 186 (69) 130 (52) <0.001
Diabetes 239 (24) 77 (31) 56 (23) 67 (25) 39 (16) <0.001
Ever a smoker 684 (68) 178 (72) 168 (69) 173 (65) 165 (67) 0.135
Coronary artery disease 355 (35) 97 (39)