JAMDA xxx (2018) 1e6

JAMDA
journal homepage: www.jamda.com

Review Article

Quick and Simple FRAIL Scale Predicts Incident Activities of Daily

Living (ADL) and Instrumental ADL (IADL) Disabilities: A Systematic
Review and Meta-analysis
Gotaro Kojima MD *
Department of Primary Care and Population Health, University College London, London, United Kingdom

a b s t r a c t

Keywords: Objectives: To quantitatively examine frailty defined by FRAIL scale as a predictor of incident disability
Frailty risks by conducting a systematic review and meta-analysis.
FRAIL scale Design: Systematic review and meta-analysis.
disability
Setting: A systematic review was conducted using 4 electronic databases (Embase, MEDLINE,
ADL
CINAHL, and PsycINFO) in April 2018 for prospective cohort studies of middle-aged or older people
IADL
examining associations between frailty and incident disability. Reference lists of the included
studies were hand-searched for additional studies. Authors of potentially eligible studies were
contacted for additional data if necessary. Methodological quality was assessed by the Newcastle-
Ottawa scale.
Participants: Community-dwelling middle-aged and older people.
Measurements: Incident risks of activities of daily living (ADL) or instrumental activities of daily living
(IADL) disability according the FRAIL scale-defined frailty.
Results: Seven studies provided odds ratios of incident disability risks according to frailty and were
included in the meta-analysis. A random effects meta-analysis showed that frailty and prefrailty were
significant predictors of ADL [pooled odds ratio (OR) ¼ 9.82, 95% confidence interval (CI) ¼ 4.71-20.46,
P < .001 for frailty (FRAIL scale ¼ 3-5) and pooled OR ¼ 2.08, 95% CI ¼ 1.77-2.45, P < .001 for prefrailty
(FRAIL scale ¼ 1-2) compared with robustness (FRAIL scale ¼ 0); pooled OR ¼ 4.44, 95% CI ¼ 3.26-6.04,
P < .001 for frailty compared with nonfrailty (FRAIL scale ¼ 0-2)] and IADL (pooled OR ¼ 2.50, 95% CI ¼
1.67-3.73, P < .001, for frailty and pooled OR ¼ 1.74, 95% CI ¼ 1.10-2.77, P ¼ .02, for prefrailty compared
with robustness). There was no evidence of publication bias.
Conclusions/Implications: The current study demonstrated that frailty status defined by the FRAIL scale
was a significant predictor of disability among community-dwelling middle-aged and older individuals.
In light of feasibility of the FRAIL scale, especially in a clinical setting, it may be a promising tool to
facilitate the translation of frailty research into clinical practice.
Ó 2018 AMDA e The Society for Post-Acute and Long-Term Care Medicine.

Frailty has been gaining scientific attention, and an exponential frailty has been shown to be consistently associated with
amount of research has been conducted over the last 2 decades,1e3 negative health outcomes, including falls,6,7 fractures,8,9 disabil-
especially since 2001 when the frailty phenotype was published by ities,10 hospitalization,11 institutionalization,12e14 dementia,15
Fried and colleagues from the Cardiovascular Health Study.4 poor quality of life,16,17 and premature death.18e20 To date, no
Although the Fried phenotype is still the most commonly consensus has been reached regarding a gold standard tool to
used criteria, there have also been numerous other tools proposed assess frailty.
to measure frailty.5 Irrespective of how frailty is defined, The FRAIL scale is a relatively new tool that was advocated by
the International Association of Nutrition and Aging Task Force
based on a systematic review of the literature as well as input from
The authors declare no conflicts of interest. a panel of geriatric experts.21 In their view, a frailty tool should be
* Address correspondence to Gotaro Kojima, MD, Department of Primary Care
quick, inexpensive, reliable, and easy to use in clinical settings
and Population Health, University College London (Royal Free Campus), Rowland
Hill Street, London NW3 2PF, United Kingdom. because the identification of frail older people at risk is the
E-mail address: [email protected]. important initial step, leading to appropriate preventive and/or

https://doi.org/10.1016/j.jamda.2018.07.019
1525-8610/Ó 2018 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
2 G. Kojima / JAMDA xxx (2018) 1e6

treatment interventions and ultimately to high-quality care for this Results

vulnerable population.21 The FRAIL scale is a simple tool consisting
of 5 yes-or-no questionsdFatigue, Resistance (inability to climb Selection Process
stairs), Ambulation (inability to walk a certain distance), Illnesses,
and Loss of weightdand does not require special equipment for Figure 1 shows a PRISMA flow diagram of the systematic search of
handgrip or such calculations as required for the frailty phenotype the literature. A total of 141 citations were identified by 4 electronic
(population-based lowest 20% of grip strength or gait speed) or the databases. After removing 57 duplicates, 76 by title and abstract
Frailty Index (summing and dividing the number of present and screening and 1 by full-text evaluation, 7 studies remained for
absent deficits, which is typically greater than 30).22 This simple methodologic quality assessment. All 7 studies were considered to
frailty tool can be administered by not only physicians but also have adequate study quality (Newcastle-Ottawa scale score range ¼ 5-
other healthcare professionals, and can be completed by phone, 7, mean ¼ 6.1) and therefore were included in this review.
mail, or email.
Frailty defined by the phenotype or other tools has been well Study Characteristics
validated and recognized as a risk factor of various adverse health
outcomes. Although the evidence regarding frailty based on the Table 1 summarizes characteristics and findings of the included
FRAIL scale is still rather limited compared with other tools, an studies. All 7 studies25e31 provided the data on ADL disability risks and
increasing amount of related research has been published in the 327e29 of them also provided the data on IADL. The studies were from
literature, and a recent meta-analysis demonstrated that frailty various countries, including Australia, China, the United Kingdom,
defined by the FRAIL scale is a significant predictor of mortality in United States, and Mexico. The sample size ranged from 779 to 8933.
community-dwelling middle-aged and older populations.18 To The shortest and longest follow-up periods were 2 and 15 years,
further strengthen validation of the FRAIL scale as a frailty tool, this respectively. All studies controlled at least for age and gender (age if 1
article will systematically review the literature and conduct a gender only cohort) except for 1 study,30 for which an unadjusted
meta-analysis on frailty based on the FRAIL scale and disability odds ratio (OR) was calculated. Most studies defined disability based
incidence among community-dwelling middle-aged and older on Katz ADL and Lawton IADL27e29,31 whereas some used different
individuals. definitions.25,26,30

ADL Disability Risk

Method
In 5 studies,25e28,30 frailty status was categorized into 3 groups,
Search Strategy and Selection Criteria robust, prefrail, and frail, defined by 0, 1-2, and 3-5 of FRAIL scale,
respectively. Incident ADL disability risks for frailty and prefrailty were
The systematic review was conducted along with a protocol significantly higher than robustness in a dose-response manner
developed in accordance with the PRISMA statements23 and [frailty: 4 studies, pooled OR ¼ 9.82, 95% confidence interval (CI) ¼
registered at PROSPERO (registration number CRD42018094603). 4.71-20.46, P < .001; prefrailty: 5 studies, pooled OR ¼ 2.08, 95% CI ¼
Four electronic databases (Embase, MEDLINE, CINAHL, and Psy- 1.77-2.45, P < .001]. Four studies provided data based on 2 frailty
cINFO) were searched in April 2018 for prospective cohort studies groups: nonfrail and frail defined by 0-2 and 3-5 of FRAIL scale,
of middle-aged or older people who were free of disability at respectively.25,26,29,31 Frailty was associated with a significantly higher
baseline examining incident activities of daily living (ADL) or incident ADL disability risk compared with nonfrailty (4 studies,
instrumental activities of daily living (IADL) disability according to pooled OR ¼ 4.44, 95% CI ¼ 3.26-6.04, P < .001) (Figure 2A).
frailty defined by the FRAIL scale. Publication years ranged from
2008, when the FRAIL scale was initially described,21,22 to April IADL Disability Risk
2018. Comprehensive search terms included both Medical Subjec-
tive Headings and text words related to the FRAIL scale and mor- Three studies examined risks of developing IADL disability.27e29
tality was used (available at PROSPERO). Reference lists of the Frailty and prefrailty were associated with significantly higher risk
included studies were hand-searched for additional studies. Au- of incident IADL disability compared with robustness (frailty: 2
thors of potentially eligible studies were contacted for additional studies, pooled OR ¼ 2.50, 95% CI ¼ 1.67-3.73, prefrailty: 2 studies,
data if necessary. Adequate methodological quality was defined as pooled OR ¼ 1.74, 95% CI ¼ 1.10-2.77, P ¼ .02). Incident IADL disability
meeting more than 5 of the 9 items of the Newcastle-Ottawa risk based on 2 frailty groups (nonfrail and frail) was provided by 1
scale.24 study (OR ¼ 4.90, 95% CI ¼ 3.67-6.54, P < .001)29 (Figure 2B).
Sensitivity or subgroup analyses were not pursued because of the
small number of studies included. There was no evidence of publica-
Statistical Analysis tion bias in the funnel plots.

If 2 or more studies provided the same effect measures of inci- Discussion

dent disability according to frailty defined by  3 components of the
FRAIL scale, a meta-analysis was attempted. When multiple studies The current study found 7 studies examining ADL and/or IADL
used the same cohort, only the study with the largest sample size disability risks according to frailty defined by the FRAIL scale among
was included in the meta-analysis. A random effects meta-analysis community-dwelling middle-aged and older people. The meta-
with the generic inverse variance method was performed because analyses showed that frailty and prefrailty defined by the FRAIL
of the anticipated high heterogeneity. Heterogeneity was assessed scale were associated with significantly higher risks of developing ADL
using the chi-square test and I2 statistic. Publication bias was and IADL disability.
assessed by visual inspection of funnel plots. Subgroup and sensi- A previous systematic review and meta-analysis including 20
tivity analyses were considered if possible. All analyses were per- original studies, most of which used the frailty phenotype criteria
formed using Review Manager 5 (Cochrane Collaboration, from the Cardiovascular Health Study, showed that frailty is associated
Denmark). with more than twice higher risks of incident ADL and IADL disability
 G. Kojima / JAMDA xxx (2018) 1e6 3

Fig. 1. PRISMA flow diagram.

than that of robustness (ADL: 8 studies, pooled OR ¼ 2.85, 95% CI ¼ studies, it was not possible to conduct additional analyses to explore
2.18-3.71, P < .00001; IADL: 3 studies, pooled OR ¼ 2.69, 95% CI ¼ 1.12- causes of high heterogeneity. Although it would have been ideal for 2
6.43, P ¼ .03).10 Although the ADL disability risks according to the researchers to conduct the systematic review independently to reduce
FRAIL scale seem to be higher (pooled OR ¼ 4.44 compared with errors and missing studies, only 1 researcher (G.K.) was available.
nonfrailty, pooled OR ¼ 9.82 compared with robustness), whether Strengths of this study include the robust and reproducible
there is a significant difference is not certain because of the wide 95% methodology according to the PRISMA statements. The literature
CI resulting from the small number of studies using the FRAIL scale. search of 4 electronic databases was extensive and comprehensive
Although frailty research has increasingly been conducted, rela- using a combination of the Medical Subjective Headings and text
tively limited evidence supports implementation into clinical prac- terms. Furthermore, methodological quality and publication bias were
tice.32,33 The first step is identification of the target: frail or prefrail also assessed. Additional data were requested and, although not all,
older individuals who require interventions to prevent further pro- some were provided by authors of the original studies and used for the
gression and adverse outcomes due to frailty.33 The frailty phenotype meta-analyses. Most of the odds ratios included in the meta-analyses
and the Frailty Index are the 2 most popular approaches to measure were adjusted for important covariates, such as age, gender, and so-
frailty.2,5 However, these 2 instruments may not necessarily be cioeconomic factors.
designed for use in busy clinical practice due to the dearth of time,
space, and equipment. For example, the frailty phenotype requires Conclusion
measurement of gait speed and grip strength and in general 30 or
more deficits need to be collected to calculate the Frailty Index. This systematic review and meta-analysis identified currently
Meanwhile, feasibility of the FRAIL scale is noteworthy. The 5 criteria available evidence of the FRAIL scale and disabilities among
included are brief, simple, and quick, as well as cost-effective as it does community-dwelling middle-aged and older individuals and
not require any special equipment or training. The FRAIL scale can be demonstrated that the FRAIL scale is a plausible and effective tool to
easily incorporated into comprehensive geriatric assessment in a busy measure frailty with regard to incident disability risks. In light of its
clinical setting to identify frail older individuals. feasibility, especially in a clinical setting, the FRAIL scale may be a
There are some limitations in this study. This study identified and promising tool to facilitate the translation of frailty research into
used only 7 studies for meta-analysis. Because of the small number of clinical practice.
4 G. Kojima / JAMDA xxx (2018) 1e6

Table 1
Characteristics of the Included Studies Examining FRAIL Scale and Incident Disability Risk Among Community-Dwelling Middle-Aged and Older People

Author/Year/Study Location Sample Female (%) Age Follow-up Adjustment Definition of Disability OR (95% CI) for Incident
Size (Range) Period Disability by FRAIL Scale

Susanto (2018) Australia 8933 100% d 15 y Age, body mass index, ADL ADL: 0: ref
ALSWH education, income “daily tasks” 1: aOR ¼ 1.97 (1.36-2.85)
management, physical 2: aOR ¼ 3.84 (2.49-5.91)
activity 3: aOR ¼ 11.28 (7.02-18.14)
4-5: aOR ¼ 31.15 (14.25-
68.08)
ADL: 0-2: ref
3-5: aOR ¼ 6.87 (4.84-9.77)
Papachristou (2017) UK 1615 0% d 3y Age Mobility limitation, ADL: 0: ref
BRHS (71-92) difficulty going up or 1-2: aOR ¼ 1.85 (1.19-2.89),
down stairs or walking P ¼ .01
400 yards 3-5: aOR ¼ 6.19 (3.29-
11.65), P < .001
ADL: 0-2: ref
3-5: aOR ¼ 4.07 (2.36-7.01),
P < .001
González (2016) Mexico 3270 53.4% 60 2y Age, gender, depressive ADL: Bathing, dressing, ADL: 0: ref
MHAS 3550 symptoms, chronic toileting, moving, eating, 1-2: aOR ¼ 1.89 (1.24-2.87),
diseases, cognition and being continent P ¼ .003
IADL: Preparing hot food, 3-5: aOR ¼ 5.30 (3.17-8.83),
buy food, taking P < .001
medications, and IADL: 0: ref
managing money 1-2: aOR ¼ 1.42 (1.09-1.85),
P ¼ .01
3-5: aOR ¼ 2.39 (1.63-3.94),
P < .001
Malmstrom (2014) US 779 d d 9y Age, gender ADL: ADL: 0: ref
AAH (49-65) Bathing, dressing, eating, 1-2: aOR ¼ 2.82 (1.7-48),
transferring bed or chair, P < .001
walking across a room, 3-5: aOR ¼ 14.93 (5.6-40.0),
getting outside, or using P < .001
toilet AUC ¼ 0.68 (0.62-0.75)
IADL: IADL: 0: ref
Preparing meals, shopping 1-2: aOR ¼ 2.29 (1.5-3.6),
for groceries, managing P < .001
money, making 3-5: aOR ¼ 3.08 (1.2-8.1),
telephone calls, doing P ¼ .02
light housework, doing AUC ¼ 0.62 (0.57-0.68)
heavy housework, getting
to places outside walking
distance, or managing
medications
Lopez (2012) Australia 8646 100% 77.8 (74-82) 6 y Age, body mass index, ADL: ADL: 0: ref
ALSWH education, living alone Katz ADL 1: aOR ¼ 1.86 (1.55-2.22)
IADL: 2: aOR ¼ 3.41 (2.82-4.13)
Lawton IADL 3: aOR ¼ 6.84 (5.46-8.56)
4-5: aOR ¼ 6.35 (4.35-9.27)
ADL: 0-2: ref
3-5: aOR ¼ 3.63 (3.05-4.32)
IADL: 0: ref
1: aOR ¼ 2.59 (2.18-3.07)
2: aOR ¼ 5.71 (4.55-7.16)
3: aOR ¼ 9.17 (6.66-12.62)
4-5: aOR ¼ 23.02 (9.34-
56.74)
IADL: 0-2: ref
3-5: aOR ¼ 4.90 (3.67-6.54)
Woo (2012) China 3153 50.3% 65 4y d ADL: ADL: 0: ref
Climbing stairs, performing 1-2: cOR ¼ 1.97 (1.60-2.44),
household activities such P < .001
as moving chairs or tables 3-5: unable to calculate due
and cleaning the floor to small number
using a vacuum cleaner or AUC ¼ 0.56 (0.53-0.59) for
mop men
AUC ¼ 0.53 (0.50-0.55) for
women
Hyde (2010) Australia 3616 0% 76.9 (70-88) 7 y Age, body mass index, Any inability in Katz ADL or ADL: 0-2: ref
HIMS medical comorbidity, Lawton IADL 3-5: aOR ¼ 3.95 (2.73-5.72),
smoking P < .001

AAH, African American Health; ALSWH, Australian Longitudinal Study on Women’s Health; aOR, adjusted odds ratio; BRHS, British Regional Heart Study; cOR, calculated odds
ratio; HIMS, Health in Men Study; MHAS, Mexican Health and Aging Study.
 G. Kojima / JAMDA xxx (2018) 1e6 5

Fig. 2. Forest plots of disability risks according to frailty status based on FRAIL scale: (A) activities of daily living disability; (B) instrumental activities of daily living.
6 G. Kojima / JAMDA xxx (2018) 1e6

References 17. Kojima G, Iliffe S, Morris RW, et al. Frailty predicts trajectories of quality of life
over time among British community-dwelling older people. Qual Life Res 2016;
25:1743e1750.
1. Morley JE, Vellas B, van Kan GA, et al. Frailty consensus: A call to action. J Am
18. Kojima G. Frailty defined by FRAIL scale as a predictor of mortality: A
Med Dir Assoc 2013;14:392e397.
systematic review and meta-analysis. J Am Med Dir Assoc 2018;19:
2. Clegg A, Young J, Iliffe S, et al. Frailty in elderly people. Lancet (London, En-
480e483.
gland) 2013;381:752e762.
19. Kojima G, Iliffe S, Walters K. Frailty index as a predictor of mortality: A sys-
3. Sloane PD, Cesari M. Research on frailty: Continued progress, continued chal-
tematic review and meta-analysis. Age Ageing 2018;47:193e200.
lenges. J Am Med Dir Assoc 2018;19:279e281.
20. Kojima G, Taniguchi Y, Kitamura A, Shinkai S. Are the Kihon Checklist and the
4. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: Evidence for a
Kaigo-Yobo Checklist compatible with the Frailty Index? J Am Med Dir Assoc
phenotype. J Gerontol A Biol Sci Med Sci 2001;56:M146eM156.
2018;19:797e800.e2.
5. Buta BJ, Walston JD, Godino JG, et al. Frailty assessment instruments: Sys-
21. Abellan van Kan G, Rolland Y, Bergman H, et al. The I.A.N.A Task Force on frailty
tematic characterization of the uses and contexts of highly-cited instruments.
assessment of older people in clinical practice. J Nutr Health Aging 2008;12:
Ageing Res Rev 2016;26:53e61.
29e37.
6. Kojima G. Frailty as a predictor of future falls among community-dwelling
22. Abellan van Kan G, Rolland YM, Morley JE, Vellas B. Frailty: Toward a clinical
older people: A systematic review and meta-analysis. J Am Med Dir Assoc
definition. J Am Med Dir Assoc 2008;9:71e72.
2015;16:1027e1033.
23. Moher D, Liberati A, Tetzlaff J, Altman DG. PRISMA Group. Preferred reporting
7. Kojima G, Kendrick D, Skelton DA, et al. Frailty predicts short-term incidence of
items for systematic reviews and meta-analyses: The PRISMA statement. Ann
future falls among British community-dwelling older people: A prospective
Intern Med 2009;151:264e269. W64.
cohort study nested within a randomised controlled trial. BMC Geriatr 2015;
24. Wells GA, Shea D, O’Connell D, et al. The Newcastle-Ottawa Scale (NOS) for
15:155.
assessing the quality of nonrandomised studies in meta-analyses. Available
8. Kojima G. Frailty as a predictor of fractures among community-dwelling
at:http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp. Accessed
older people: A systematic review and meta-analysis. Bone 2016;90:
August 20, 2015.
116e122.
25. Susanto M, Hubbard RE, Gardiner PA. Validity and responsiveness of the FRAIL
9. Kojima G. Frailty significantly increases the risk of fractures among middle-
Scale in middle-aged women. J Am Med Dir Assoc 2018;19:65e69.
aged and older people. Evid Based Nurs 2017;20:119e120.
26. Papachristou E, Wannamethee SG, Lennon LT, et al. Ability of self-reported
10. Kojima G. Frailty as a predictor of disabilities among community-dwelling
frailty components to predict incident disability, falls, and all-cause mortal-
older people: A systematic review and meta-analysis. Disabil Rehabil 2017;
ity: Results from a population-based study of older British men. J Am Med Dir
39:1897e1908.
Assoc 2017;18:152e157.
11. Kojima G. Frailty as a predictor of hospitalisation among community-dwelling
27. Diaz de Leon Gonzalez E, Gutierrez Hermosillo H, Martinez Beltran JA, et al.
older people: A systematic review and meta-analysis. J Epidemiol Commun
Validation of the FRAIL scale in Mexican elderly: Results from the Mexican
Health 2016;70:722e729.
Health and Aging Study. Aging Clin Exp Res 2016;28:901e908.
12. Kojima G. Frailty as a predictor of nursing home placement among community-
28. Malmstrom TK, Miller DK, Morley JE. A comparison of four frailty models. J Am
dwelling older adults: A systematic review and metaeanalysis. J Geriatr Phys
Geriatr Soc 2014;62:721e726.
Ther (2001) 2018;41:42e48.
29. Lopez D, Flicker L, Dobson A. Validation of the frail scale in a cohort of older
13. Kojima G. Prevalence of frailty in nursing homes: A systematic review and
Australian women. J Am Geriatr Soc 2012;60:171e173.
meta-analysis. J Am Med Dir Assoc 2015;16:940e945.
30. Woo J, Leung J, Morley JE. Comparison of frailty indicators based on clinical
14. Kojima G, Tanabe M. Frailty is highly prevalent and associated with vitamin D
phenotype and the multiple deficit approach in predicting mortality and
deficiency in male nursing home residents. J Am Geriatr Soc 2016;64:
physical limitation. J Am Geriatr Soc 2012;60:1478e1486.
e33ee35.
31. Hyde Z, Flicker L, Almeida OP, et al. Low free testosterone predicts frailty in
15. Kojima G, Taniguchi Y, Iliffe S, Walters K. Frailty as a predictor of Alzheimer
older men: The Health In Men study. J Clin Endocrinol Metab 2010;95:
disease, vascular dementia, and all dementia among community-dwelling
3165e3172.
older people: A systematic review and meta-analysis. J Am Med Dir Assoc
32. Cesari M, Prince M, Thiyagarajan JA, et al. Frailty: An emerging public health
2016;17:881e888.
priority. J Am Med Dir Assoc 2016;17:188e192.
16. Kojima G, Iliffe S, Jivraj S, Walters K. Association between frailty and quality of
33. Vellas B, Cestac P, Moley JE. Implementing frailty into clinical practice: We
life among community-dwelling older people: A systematic review and meta-
cannot wait. J Nutr Health Aging 2012;16:599e600.
analysis. J Epidemiol Commun Health 2016;70:716e721.