Personalizing non-surgical care in patients with

osteoarthritis of the hip or knee

Di-Janne Barten
Personalizing non-surgical care in patients with osteoarthritis of the hip or knee
Utrecht University, Utrecht, The Netherlands

ISBN: 978-94-6299-918-3
Printing: Ridderprint BV | [Link]

The research presented in this thesis was conducted at the Netherlands Institute for Health
Services Research (NIVEL), Utrecht, The Netherlands.

The printing of this thesis was financially supported by the Netherlands Institute for Health
Services Research (NIVEL) and the Scientific College Physical Therapy (WCF) of the Royal Dutch
Society for Physical Therapy (KNGF).

© 2018. This manuscript version is made available under the CC-BY-NC-ND 4.0 license
[Link]
 Personalizing non-surgical care in patients with
osteoarthritis of the hip or knee

Gepersonaliseerde, niet-chirurgische zorg bij patiënten met artrose van de heup of knie
(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de
rector magnificus, [Link]. G.J. van der Zwaan, ingevolge het besluit van het
college voor promoties in het openbaar te verdedigen op
donderdag 24 mei 2018 des ochtends te 10.30 uur

door

Jacoba Anna Barten

geboren op 13 mei 1985 te Wageningen

General Introduction | 3
Promotoren: Prof. dr. C. Veenhof
Prof. dr. D.H. de Bakker†
Prof. dr. J. Dekker

Copromotor: Dr. I.C.S. Swinkels

Contents

Chapter 1 General introduction 9

Chapter 2 Treatment of hip/knee osteoarthritis in Dutch general practice

and physical therapy practice: an observational study 25

Chapter 3 Factors associated with referral to secondary care in patients

with osteoarthritis of the hip or knee after implementation
of a stepped-care strategy 41

Chapter 4 One size fits all in physiotherapy management of knee osteoarthritis?

A cross-sectional, clinical vignette study 61

Chapter 5 Patient-specific self-assessment instruments for measuring

physical function: a systematic review of measurement properties 83

Chapter 6 Validity and responsiveness of the Dutch McMaster Toronto Arthritis

patient preference questionnaire (MACTAR) in patients with
osteoarthritis of the hip or knee 109

Chapter 7 General discussion 131

Summary 145
Samenvatting 153
Dankwoord 163
About the author 169
List of publications 173
 1
General introduction
Osteoarthritis is one of the most common musculoskeletal disorders in developed countries.
Approximately 10-12% of the adult population have symptomatic OA1. OA can affect several
joints, but the knee and the hip joints, are the most commonly affected joints in patients with
OA2. In the Netherlands, the prevalence of hip/knee OA is estimated on 40.7/1000 males and
73.2/1000 females. Incidence rates for hip/knee OA are 4.0/1000 males and 6.5/1000 females3.
These numbers are likely to be underestimated, since many patients do not seek help for their
complaints. Nevertheless, the economic burden of OA is already high, representing 21.5% of
the annual costs concerning musculoskeletal disorders and 1.2% of Dutch national health care
costs4. Due to the aging population and current Western lifestyle patterns5,6, the number of
people living with hip or knee OA is anticipated to increase largely in the next decades7. In the
Netherlands, an increase of almost 40% is expected over the period 2011-20304. Considering
patients with hip and knee OA are expected to live for prolonged periods with severe
complaints due to OA8, the burden of OA will become a major problem for our health care
systems as well.

Pathogenesis of OA

Osteoarthritis is a heterogeneous pathology characterized by a complex and multifactorial

nature9. Its’ pathogenesis is still not completely understood. However, over the last years, it
has become evident that OA is characterized by a failure of the entire synovial joint organ 2.
Besides progressive loss of articular cartilage, this failure could be expressed by muscle
atrophy, ligament dysfunction, meniscal damage, osteophytes formation, changes in the
subchondral bone (sclerosis) and a patchy chronic synovitis2,10 (figure 1). Structural tissue
changes in OA are only to a limited extent related to clinical characteristics presented by
patients with OA10,11.

FIGURE 1. Schematic drawing of an osteoarthritic joint (from Bijlsma et al. (2011))10, printed with permission)

General Introduction | 11
Wide range of functional (dis)ability

As previously mentioned, hip/knee OA is a heterogeneous disease, resulting in varying

disabilities in functioning. A patient’s functioning could be systematically described by the
International Classification of Functioning, Disability, and Health (ICF) ([Link]). The ICF
model distinguishes ‘impairments in body functions’, ‘limitations in activities’, and ‘restrictions
in participation’ in relation to a medical disorder and influenced by personal and environmental
factors. Figure 2 demonstrates the functioning of patients with hip/knee OA classified by the
ICF.

Osteoarthritis of the hip or knee

Body Functions and Activities Participation

Structures

proprioception, pain, transferring oneself, remunerative or non-

mobility, stability, muscle walking, standing, sitting, remunerative employment,
power, muscle endurance, moving around, washing community life, recreation
alignment, muscular oneself, dressing and
atrophy, hypertonia toileting

Environmental factors Personal factors

e.g. home adaptations and aids for daily life, co-morbidity, lifestyle, character,
work or sport, facilities, relatives, friends, care experiences, self- efficacy, age, sex,
providers, colleagues ethnicity, profession, social background and
disease perception

FIGURE 2. Functioning of patients with hip/knee osteoarthritis, classified by he International Classification

of Functioning, Disabilities and Health (Peter et al. (2011))14

The major impairment in hip/knee OA is (chronic) pain, which typically emerges with weight-
loading activities and diminishes with rest. In advanced OA, rest pain and night pain can also
occur12. Other impairments are short-lasting inactivity-related joint stiffness, cracking of joints
(crepitus), loss of joint mobility, muscle atrophy, joint inflammation, and joint instability10,13.
These impairments are accompanied by limitations in daily functioning and restrictions in a
patient’s participation in daily life. Commonly mentioned limitations in activities are
transferring oneself, moving around and dressing oneself. Potential restrictions in participation
are, for instance, absence from work and inability to perform a sport. Contextual factors
considerably predispose to the impact of impairments, limitations and restrictions on a
patient’s life2. Personal factors like self-efficacy and pain catastrophizing may attribute to

12
varying levels of physical disability. The same applies to environmental factors like social
support and financial status.

Increasing demand for personalized management of OA

Personalized treatment is recognized as one of the solutions to face the complex and
continuously expanding health care demand of patients with (multiple) chronic conditions15.
Those populations often show high heterogeneity16,17, high co-morbidity rates17, low responses
to commonly used interventions16, and high risk of economic burden16.
Due to its’ heterogeneity, and the expected increasing burden of disease, OA would be
particularly suitable for personalized treatment2,7,10,16. However, current clinical guidelines for
treatment of hip/knee OA hardly present personalized recommendations. To date, clinical
guidelines generally recommend clinicians to treat OA by pharmacological, non-
pharmacological and surgical interventions to diminish impairments in body functions and
reduce limitations in activities and restrictions in participation18,20. Little attention is paid, for
instance, on differences in onset of OA, course of OA, and patient’s priorities over time.

Therefore, in this thesis we will explore some opportunities to personalize a patient’s

treatment. Based on drivers for personalized treatment described by Karsdal et al.16, we will
consider two facets which potentially attribute to personalized treatment: timing of care and
focus of care.

Timing of care

Current clinical guidelines provide a clear insight into effective non-pharmacological,

pharmacological, and surgical treatment in hip/knee OA. Surgical treatment, by way of Total
Joint Replacement (TJR), seems an effective and cost-effective final treatment option in end-
stage hip/knee OA21,22. The absolute number of TJRs increases annually, up to the amount of
51.680 total hip replacements (+149% compared to 2005) and 57.893 total knee replacements
(+297%) in the Netherlands in 203023. Despite the increasing prevalence of OA24, it has been
stated that TJR is overutilized25. Appropriate timing of a TJR is important, since the lifespan of
a prosthesis is limited, and patient outcomes after a revision are worse compared to a primary
prosthesis26,27. For that reason, clinical guidelines recommend to start with (combinations of)
non-surgical interventions and to reserve TJR for patients who do not respond sufficiently to
non-surgical treatment options28-31. However, several studies have indicated that non-surgical
interventions in patients with hip/knee OA are underutilized26,28,32-34. The question arises why
surgical treatment in secondary care is relatively often preferred over non-surgical treatment
in primary care.

General Introduction | 13
First, the severity of clinical symptoms plays a role. Several research has shown that patients
with more pain and more limitations in activities are more likely to receive a TJR35-38. The same
applies to decreased knee-extension (in patients with knee OA)37, more severe radiological
deviations39, and absence of comorbidities38. Moreover, a patient’s willingness and lack of
knowledge are associated with greater likelihood of undergoing TJR38,40. By way of illustration,
in a study by Hofstede et al., almost 30% of the patients who are surgically treated agreed with
the statement ‘too much loss of cartilage to use non-surgical treatments’,40 despite the fact
that cartilage damage is only to a limited extent related to surgical necessity2,10. It is remarkable
that contextual factors, like a patient’s financial status or a patient’s psychosocial well-being
are rarely involved in research regarding the determination of a patient’s individual treatment
over time. Besides patient-related factors, clinicians’ or organizations’ characteristics are
currently seldom involved in determining factors associated with the timing and setting of care
in patients with hip/knee OA. However, it is reasonable to assume that these characteristics
influence a patient’s journey in OA care as well. For example, Cottrell et al. showed that
insufficient expertise was one of the most mentioned barriers by general practitioners (GPs)
for using appropriate non-surgical treatment (e.g. exercise therapy) at the right time in patients
with knee OA41.

One way to facilitate the appropriate use of non-surgical care in patients with hip/knee OA is
to phase non-surgical interventions over time. Phasing treatment is expected to reduce
inappropriate use of advanced care30,42. The recently developed, multidisciplinary, patient-
centered Stepped Care Strategy (SCS) phases non-surgical care in patients with hip/knee OA
and facilitates clinicians to personalize a patient’s treatment over time43. An important principle
of the SCS is that patients exclusively receive modalities of the subsequent steps if all
modalities of the previous steps have been offered. The SCS comprises three consecutive
steps. Step 1 comprises optimizing a patients’ self-care in primary care by education, lifestyle
advices, and the prescription of acetaminophen. In case of persisting complaints after adequate
application of all step-1 interventions, (topical) NSAIDs are added or a referral for allied health
care (exercise therapy, dietary therapy) will be considered (step 2). Finally, step 3 comprises
the application of TENS and intra-articular corticosteroid injections or a referral for
interprofessional evaluation in secondary care (Figure 3)43. It has not been investigated yet
which factors related to the patient respectively the clinician or practice influence the patient
journey in OA care after implementation of the SCS. This insight could be helpful to determine
the added value of the SCS regarding the timing of care in patients with hip/knee OA.

To date, the implementation of the SCS is limited to only one region in the Netherlands. The
vast majority of clinicians in OA care are not facilitated in phasing their treatment over time.
Insight into current practice in relation to the SCS will provide guidance for nationwide
implementation of the SCS and, subsequently, timely treatment in individual patients with
hip/knee OA.

14
 FIGURE 3. Stepped Care Strategy ‘Beating osteoARThritis’ for non-surgical care in hip/knee osteoarthritis
(Smink et al. (2011))43, printed with permission

Focus of care

Due to the heterogeneity in patients with especially knee OA, it has been hypothesized that
this population actually consists of different subgroups44,45. Recently, Knoop et al. and van der
Esch et al. have succeeded in distinguishing five different phenotypes in patients with knee OA
in two large cohorts of patients with knee OA46,47. The identified phenotypes are: ‘minimal
joint disease phenotype’, ‘strong muscle strength phenotype’, ‘severe radiographic OA
phenotype’, ‘obese phenotype’, and ‘depressive mood phenotype’46,47. Each phenotype
comprises one distinguishing feature (Figure 4). For example, patients fitting in the ‘strong
muscle phenotype’ show high quadriceps muscle strength compared to the remaining
phenotypes. It is conceivable that the strong muscle phenotype calls for a different focus of
treatment than, for instance, the obese phenotype. Therefore, several authors have
considered that the focus of treatment may need to be tailored to specific phenotypes of knee
OA to improve treatment results44-46,48. The question arises to what extent clinicians, especially
physical therapists, currently differentiate their focus of care between phenotypes of knee OA.

General Introduction | 15
 FIGURE 4. Differences in clinical characteristics of five phenotypes of knee osteoarthritis
(based on Knoop (2011)46 and van der Esch (2015) 47)

Measurement tools may facilitate clinicians to focus their treatment to subgroups or individuals
with hip/knee OA. To enable diagnostic procedures and evaluation over time focused on the
individual patient, patients’ priorities should be taken into account49. To date, mainly generic
and disease-specific tools are used in patients with hip/knee OA. These tools comprise fixed
items, which allow comparisons across populations and settings50. However, outcomes of
generic and disease-specific tools are often difficult to interpret at individual patient level since
these tools do not consider patients’ preferences and variability in performance on particular
activities49. Therefore, the use of patient-specific tools has become more and more popular in
clinical practice as patients are concerned with the content of the measurement tool49,51. An
overview of patient-specific measurement tools which could be used in patients with hip/knee
OA is still lacking. Furthermore, there is insufficient insight into the psychometric properties
of commonly used patient-specific measurement tools like the Patient Specific Functional Scale
and, especially used in patients with hip/knee OA, the McMaster Toronto Arthritis Patient
Preference Questionnaire (MACTAR)52. Providing an overview of commonly used patient-
specific measurement tools, including a psychometric evaluation, may facilitate clinicians to
focus care to the individual patient with hip/knee OA.

16
Scope of this thesis

Personalized treatment could be one of the solutions to face the complex and continuously
expanding health care demand of patients with chronic conditions like hip/knee OA. To date,
personalized treatment is in its’ infancy. In this thesis we will explore two drivers which
potentially attribute to personalized treatment in hip/knee OA: timing of care and focus of
care.

Part I: Timing of care

The first part of this thesis focuses on current timing of non-surgical care in patients with
hip/knee OA in Dutch general practice and physical therapy practice. Based on routinely
registered data in NIVEL Primary Care Database, we will get insight into current treatment of
GPs and physical therapists and to what extent this treatment is in accordance with the
Stepped Care Strategy (Chapter 2). The subsequent study focuses on the patient journey in
hip/knee OA and provides insight into factors associated with the setting of care (primary
care, secondary care) in which patients are treated (Chapter 3). Next to individual
determinants, characteristics of their GPs and general practices are involved in the
determination of a patients’ journey over time.

Research questions are:

 What is the content of current GP care in patients with hip/knee OA, including the
compliance to the Stepped Care Strategy?
 To what extent differs the content of physical therapy practice in GP-referred patients
compared to self-referred patients with hip/knee OA?
 Which individual factors, GP-related factors and general practice related factors are
associated with 1) treatment in primary care, 2) continuation of non-surgical care after
a referral, and (3) surgical treatment in patients visiting their GP due to hip/knee OA?

Part II: Focus of care

The topic of the second part of this thesis is the focus of care in patients with hip/knee OA.
In Chapter 4, clinical vignettes are used to determine to what extent physical therapists tailor
their treatment to different phenotypes of knee OA. In Chapter 5 and 6, personalized
measurement of functioning in hip/knee OA is the central issue. A systematic review is
performed to provide a comprehensive overview of measurement properties of patient-
specific tools measuring physical function (Chapter 5). Chapter 6 shows de results of the
validation of the Dutch McMaster Toronto Arthritis Patient Preference Questionnaire

General Introduction | 17
(MACTAR). The MACTAR is a patient-specific tool, which could be used in patients with
hip/knee OA.

Research questions are:

 To what extent is physical therapists’ treatment tailored to the ‘minimal joint disease’
phenotype, the ‘strong muscle strength’ phenotype, the ‘severe radiographic OA’
phenotype, the ‘obese’ phenotype, and the ‘depressive mood’ phenotype in patients
with knee OA?
 Which patient-specific self-assessment instruments are available to measure physical
function in patients with musculoskeletal disorders?
 What are the descriptive properties as well as the psychometric qualities of patient-
specific instruments measuring physical function in patients with musculoskeletal
disorders?
 What is the content validity, the construct validity and the responsiveness of the Dutch
McMaster Toronto Arthritis Patient Preference Questionnaire in patients with OA of
the hip or knee?

Chapter 7 provides an overall discussion of the findings of this thesis, including

recommendations for clinical practice, policy and future research. Lastly, this thesis will be
summarized in English and in Dutch.

The outline of the thesis is graphically presented in Figure 5.

18
FIGURE 5. Outline thesis ‘Personalizing non-surgical care in patients in osteoarthritis of the hip or knee
References
9. Dell'Isola A, Allan R, Smith SL, Marreiros SS,
Steultjens M. Identification of clinical
1. Hunter DJ, Schofield D, Callander E. The phenotypes in knee osteoarthritis: a systematic
individual and socioeconomic impact of review of the literature. BMC Musculoskelet
osteoarthritis. Nat Rev Rheumatol 2014 Disord 2016 Oct 12;17(1):425.
Jul;10(7):437-441.
10. Bijlsma JW, Berenbaum F, Lafeber FP.
2. Hunter DJ. Osteoarthritis. Best Pract Res Clin Osteoarthritis: an update with relevance for
Rheumatol 2011 Dec;25(6):801-814. clinical practice. Lancet 2011 Jun
18;377(9783):2115-2126.
3. Nielen M, Flinterman L, Kroneman M, Verheij
R. Incidentie en prevalentie van 11. Hunter DJ, McDougall JJ, Keefe FJ. The
gezondheidsproblemen in de Nederlandse symptoms of osteoarthritis and the genesis of
huisartsenpraktijk in 2016. Uit: NIVEL pain. Rheum Dis Clin North Am 2008
Zorgregistraties eerste lijn [internet]. 2017; Aug;34(3):623-643.
Available at: [Link]/node/4309.
Accessed 12/27, 2017. 12. Lane NE, Brandt K, Hawker G, Peeva E,
Schreyer E, Tsuji W, et al. OARSI-FDA
4. [Link]. Kosten van zorg initiative: defining the disease state of
voor artrose. 2013; Available at: osteoarthritis. Osteoarthritis Cartilage 2011
[Link] May;19(5):478-482.
rwerp/artrose/kosten/kosten#node-kosten-
van-zorg-voor-artrose. Accessed 12/27, 2017. 13. Felson DT, Lawrence RC, Dieppe PA, Hirsch
R, Helmick CG, Jordan JM, et al.
5. Crowninshield RD, Rosenberg AG, Sporer SM. Osteoarthritis: new insights. Part 1: the disease
Changing demographics of patients with total and its risk factors. Ann Intern Med 2000 Oct
joint replacement. Clin Orthop Relat Res 2006 17;133(8):635-646.
Feb;443:266-272.
14. Peter WF, Jansen MJ, Hurkmans EJ, Bloo H,
6. Ostendorf M, Johnell O, Malchau H, Dhert WJ, Dekker J, Dilling RG, et al. Physiotherapy in hip
Schrijvers AJ, Verbout AJ. The epidemiology of and knee osteoarthritis: development of a
total hip replacement in The Netherlands and practice guideline concerning initial
Sweden: present status and future needs. Acta assessment, treatment and evaluation. Acta
Orthop Scand 2002 Jun;73(3):282-286. Reumatol Port 2011 Jul-Sep;36(3):268-281.

7. Cross M, Smith E, Hoy D, Nolte S, Ackerman 15. Wijma AJ, Bletterman AN, Clark JR, Vervoort
I, Fransen M, et al. The global burden of hip and SCJM, Beetsma A, Keizer D, et al. Patient-
knee osteoarthritis: estimates from the global centeredness in physiotherapy: What does it
burden of disease 2010 study. Ann Rheum Dis entail? A systematic review of qualitative
2014 Jul;73(7):1323-1330. studies. Physiother Theory Pract 2017
Nov;33(11):825-840.
8. Global Burden of Disease Study 2013
Collaborators. Global, regional, and national 16. Karsdal MA, Christiansen C, Ladel C,
incidence, prevalence, and years lived with Henriksen K, Kraus VB, Bay-Jensen AC.
disability for 301 acute and chronic diseases Osteoarthritis--a case for personalized health
and injuries in 188 countries, 1990-2013: a care? Osteoarthritis Cartilage 2014
systematic analysis for the Global Burden of Jan;22(1):7-16.
Disease Study 2013. Lancet 2015 Aug 17. Rothwell PM. External validity of randomised
22;386(9995):743-800. controlled trials: "to whom do the results of

20
 this trial apply?". Lancet 2005 Jan 1- 25. Ghomrawi HM, Schackman BR, Mushlin AI.
7;365(9453):82-93. Appropriateness criteria and elective
procedures--total joint arthroplasty. N Engl J
18. Zhang W, Moskowitz RW, Nuki G, Abramson Med 2012 Dec 27;367(26):2467-2469.
S, Altman RD, Arden N, et al. OARSI
recommendations for the management of hip 26. Hofstede SN, Vliet Vlieland TP, van den Ende
and knee osteoarthritis, part I: critical appraisal CH, Nelissen RG, Marang-van de Mheen PJ,
of existing treatment guidelines and systematic van Bodegom-Vos L. Variation in use of non-
review of current research evidence. surgical treatments among osteoarthritis
Osteoarthritis Cartilage 2007 Sep;15(9):981- patients in orthopaedic practice in the
1000. Netherlands. BMJ Open 2015 Sep
9;5(9):e009117-2015-009117.
19. McAlindon TE, Bannuru RR, Sullivan MC,
Arden NK, Berenbaum F, Bierma-Zeinstra SM, 27. Bennell KL, Hunter DJ, Hinman RS.
et al. OARSI guidelines for the non-surgical Management of osteoarthritis of the knee. BMJ
management of knee osteoarthritis. 2012 Jul 30;345:e4934.
Osteoarthritis Cartilage 2014 Mar;22(3):363-
388. 28. Snijders GF, den Broeder AA, van Riel PL,
Straten VH, de Man FH, van den Hoogen FH,
20. Nelson AE, Allen KD, Golightly YM, Goode et al. Evidence-based tailored conservative
AP, Jordan JM. A systematic review of treatment of knee and hip osteoarthritis:
recommendations and guidelines for the between knowing and doing. Scand J
management of osteoarthritis: The chronic Rheumatol 2011 May;40(3):225-231.
osteoarthritis management initiative of the U.S.
bone and joint initiative. Semin Arthritis 29. Shrier I, Feldman DE, Gaudet MC, Rossignol M,
Rheum 2014 Jun;43(6):701-712. Zukor D, Tanzer M, et al. Conservative non-
pharmacological treatment options are not
21. Chang RW, Pellisier JM, Hazen GB. A cost- frequently used in the management of hip
effectiveness analysis of total hip arthroplasty osteoarthritis. J Sci Med Sport 2006 May;9(1-
for osteoarthritis of the hip. JAMA 1996 Mar 2):81-86.
20;275(11):858-865.
30. Hunter DJ, Neogi T, Hochberg MC. Quality of
22. Hawker GA, Badley EM, Croxford R, Coyte osteoarthritis management and the need for
PC, Glazier RH, Guan J, et al. A population- reform in the US. Arthritis Care Res
based nested case-control study of the costs of (Hoboken) 2011 Jan;63(1):31-38.
hip and knee replacement surgery. Med Care
2009 Jul;47(7):732-741. 31. Jordan KM, Sawyer S, Coakley P, Smith HE,
Cooper C, Arden NK. The use of conventional
23. Otten R, van Roermund PM, Picavet HS. and complementary treatments for knee
Trends in the number of knee and hip osteoarthritis in the community.
arthroplasties: considerably more knee and hip Rheumatology (Oxford) 2004 Mar;43(3):381-
prostheses due to osteoarthritis in 2030. Ned 384.
Tijdschr Geneeskd 2010;154:A1534.
32. Hoogeboom TJ, Snijders GF, Cats HA, de Bie
24. Cram P, Lu X, Kates SL, Singh JA, Li Y, Wolf RA, Bierma-Zeinstra SM, van den Hoogen FH,
BR. Total knee arthroplasty volume, utilization, et al. Prevalence and predictors of health care
and outcomes among Medicare beneficiaries, use in patients with early hip or knee
1991-2010. JAMA 2012 Sep 26;308(12):1227- osteoarthritis: two-year follow-up data from
1236. the CHECK cohort. Osteoarthritis Cartilage
2012 Jun;20(6):525-531.

General Introduction |21

33. Smink AJ, Dekker J, Vliet Vlieland TP, Swierstra 41. Cottrell E, Foster NE, Porcheret M, Rathod T,
BA, Kortland JH, Bijlsma JW, et al. Health care Roddy E. GPs' attitudes, beliefs and behaviours
use of patients with osteoarthritis of the hip or regarding exercise for chronic knee pain: a
knee after implementation of a stepped-care questionnaire survey. BMJ Open 2017 Jun
strategy: an observational study. Arthritis Care 17;7(6):e014999-2016-014999.
Res (Hoboken) 2014 Jun;66(6):817-827.
42. Williams B, Poulter NR, Brown MJ, Davis M,
34. Brand CA, Harrison C, Tropea J, Hinman RS, McInnes GT, Potter JF, et al. Guidelines for
Britt H, Bennell K. Management of management of hypertension: report of the
osteoarthritis in general practice in Australia. fourth working party of the British
Arthritis Care Res (Hoboken) 2014 Hypertension Society, 2004-BHS IV. J Hum
Apr;66(4):551-558. Hypertens 2004 Mar;18(3):139-185.

35. Birrell F, Afzal C, Nahit E, Lunt M, Macfarlane 43. Smink AJ, van den Ende CH, Vliet Vlieland TP,
GJ, Cooper C, et al. Predictors of hip joint Swierstra BA, Kortland JH, Bijlsma JW, et al.
replacement in new attenders in primary care "Beating osteoARThritis": development of a
with hip pain. Br J Gen Pract 2003 stepped care strategy to optimize utilization
Jan;53(486):26-30. and timing of non-surgical treatment
modalities for patients with hip or knee
36. McHugh GA, Campbell M, Luker KA. GP osteoarthritis. Clin Rheumatol 2011
referral of patients with osteoarthritis for Dec;30(12):1623-1629.
consideration of total joint replacement: a
longitudinal study. Br J Gen Pract 2011 44. Driban JB, Sitler MR, Barbe MF,
Aug;61(589):e459-68. Balasubramanian E. Is osteoarthritis a
heterogeneous disease that can be stratified
37. Zeni JA,Jr, Axe MJ, Snyder-Mackler L. Clinical into subsets? Clin Rheumatol 2010
predictors of elective total joint replacement Feb;29(2):123-131.
in persons with end-stage knee osteoarthritis.
BMC Musculoskelet Disord 2010 May 6;11:86- 45. Felson DT. Identifying different osteoarthritis
2474-11-86. phenotypes through epidemiology.
Osteoarthritis Cartilage 2010 May;18(5):601-
38. Hawker GA, Guan J, Croxford R, Coyte PC, 604.
Glazier RH, Harvey BJ, et al. A prospective
population-based study of the predictors of 46. Knoop J, van der Leeden M, Thorstensson CA,
undergoing total joint arthroplasty. Arthritis Roorda LD, Lems WF, Knol DL, et al.
Rheum 2006 Oct;54(10):3212-3220. Identification of phenotypes with different
clinical outcomes in knee osteoarthritis: data
39. Lievense AM, Bierma-Zeinstra SM, Verhagen from the Osteoarthritis Initiative. Arthritis
AP, Verhaar JA, Koes BW. Prognostic factors Care Res (Hoboken) 2011 Nov;63(11):1535-
of progress of hip osteoarthritis: a systematic 1542.
review. Arthritis Rheum 2002 Oct
15;47(5):556-562. 47. van der Esch M, Knoop J, van der Leeden M,
Roorda LD, Lems WF, Knol DL, et al. Clinical
40. Hofstede SN, Marang-van de Mheen PJ, Vliet phenotypes in patients with knee
Vlieland TP, van den Ende CH, Nelissen RG, osteoarthritis: a study in the Amsterdam
van Bodegom-Vos L. Barriers and Facilitators osteoarthritis cohort. Osteoarthritis Cartilage
Associated with Non-Surgical Treatment Use 2015 Apr;23(4):544-549.
for Osteoarthritis Patients in Orthopaedic
Practice. PLoS One 2016 Jan 48. Hinman RS, Crossley KM. Patellofemoral joint
22;11(1):e0147406. osteoarthritis: an important subgroup of knee

22
 osteoarthritis. Rheumatology (Oxford) 2007
Jul;46(7):1057-1062. 51. Wright JG, Young NL. The patient-specific
index: asking patients what they want. J Bone
49. Seror R, Tubach F, Baron G, Guillemin F, Joint Surg Am 1997 Jul;79(7):974-983.
Ravaud P. Measure of function in rheumatoid
arthritis: individualised or classical scales? Ann 52. Verhoeven AC, Boers M, van der Liden S.
Rheum Dis 2010 Jan;69(1):97-101. Validity of the MACTAR questionnaire as a
functional index in a rheumatoid arthritis
50. Jolles BM, Buchbinder R, Beaton DE. A study clinical trial. The McMaster Toronto Arthritis.
compared nine patient-specific indices for J Rheumatol 2000 Dec;27(12):2801-2809.
musculoskeletal disorders. J Clin Epidemiol
2005 Aug;58(8):791-801.

General Introduction |23

 Treatment of hip/knee
2
osteoarthritis in Dutch general
practice and physical therapy
practice: an observational study

Published in BMC Fam Pract. 2015 Jun 27;16:75

[Link]

Di-Janne Barten
Ilse Swinkels
Sara Dorsman
Joost Dekker
Cindy Veenhof
Dinny de Bakker
Abstract

Background A multidisciplinary, guideline-based Stepped-Care-Strategy (SCS),

has recently been developed to improve the management of hip and knee
osteoarthritis (OA). To date, it is unknown to what extent current Dutch OA care
is consistent with the SCS, both with respect to the content of care as well as the
sequence of care. Furthermore, there is a lack of clarity regarding the role of
different health care providers in the performance of OA care according to the
SCS. Therefore, the main purpose of this study is to describe the content of
primary care in patients with hip/knee OA, including the compliance to the SCS
and taking into account the introduction of patient self-referral to physical therapy.

Methods Data were used from NIVEL Primary Care Database. In total, 12.118
patients with hip/knee OA who visited their GP or physical therapist were selected.
Descriptive statistics were used to compare the content of care in GP-referred and
self-referred patients to physical therapy.

Results Content of care performed by GPs mostly concerned consultations,

followed by NSAID prescriptions and referrals to secondary care. Both
prescriptions of acetaminophen and referrals to physical therapy respectively
dietary therapy were rarely mentioned. Nevertheless, still 65% of the patients in
physical therapy practice were referred by their GP. Compared to GP-referred
patients, self-referred patients more often presented recurrent complaints and
were treated less often by activity-related exercise therapy. Education was rarely
registered as singular intervention, neither in GP-referred nor in self-referred
patients.

Conclusion In accordance with the SCS, less advanced interventions are more
often applied than more advanced interventions. To optimize the adherence to the
SCS, GPs could reconsider the frequent use of NSAIDs instead of analgesics and
the low referral rate to allied health care. Self-referral to physical therapy partially
distorts both the low referral rate in general practice and the low application rate
of education as singular intervention in physical therapy practice. Further research
is recommended to evaluate the effects of task-shifting in OA care, taking into
account the content of the SCS.

Current OA Care | 27
Background

Osteoarthritis (OA) is one of the most common disorders of the musculoskeletal system1. As
a consequence of the aging process, a large increase of the OA population is expected over
the next decades2. Considering OA as the major cause of musculoskeletal pain and disability
in the elderly, a large increase of demand for care could be expected as well3. To cope with
this demand, it is important to manage OA in an effective and efficient way. Over the last
decades, more than 50 modalities of non-pharmacological, pharmacological and surgical
interventions for hip and knee OA have been described in medical literature and integrated in
(inter)national, monodisciplinary and interdisciplinary clinical guidelines3-5. Recently, Smink et
al. developed a multidisciplinary, guideline-based Stepped-Care-Strategy (SCS), known as
BART, i.e. Beating Osteoarthritis, to improve the management of hip and knee OA6. In addition
to current clinical guidelines that recommend appropriate non-surgical treatment modalities,
the SCS focuses on the optimal order in which to employ them. It recommends offering all
modalities in the previous steps before turning to more advanced modalities in the subsequent
steps. According to the SCS, treatment of hip/knee OA starts in primary care with stimulating
patients’ self-care by emphasizing the usefulness of an adequate dose of acetaminophen and by
educating patients about OA and their lifestyle (step 1). Additionally, the use of
glucosaminesulphate could be considered for a trial period of three months. In case of
persisting complaints, which is identified during an evaluation visit at the general practitioner
(GP), (topical) non-steriodal anti-inflammatory drugs (NSAIDs) or tramadol are applied,
supplemented by prescribing exercise therapy and, in case of overweight, dietary therapy to
diminish the impairments and limitations due to OA (step 2). A referral to secondary care,
TENS and intra-articular corticosteroid injections could be applied as final non-surgical
interventions (step 3).

To date, it is unknown to what extent current Dutch OA care is consistent with the SCS,
both with respect to the content of care as well as the sequence of care. Furthermore, there
is a lack of clarity regarding the role of different health care providers in the performance of
OA care according to the SCS. The SCS describes several interventions, but do not apportion
these interventions to a specific discipline. It stands to reason that step-1 interventions mostly
are performed by a GP. In case of unsatisfactory results, the GP refers patients to allied health
care providers (step-2) or to an orthopaedic surgeon (step-3). However, the introduction of
patient self-referral for physical therapy in 20067, possibly has interrupted this natural sequence
of care. It is expected that an increasing number of patients will leave out their GP and directly
approach a physical therapist in case of experiencing musculoskeletal complaints8. In
consequence, the question arises to what extent patients using self-referral for physical
therapy still receive interventions described in step-1 of the SCS.

28
Therefore, the two main objectives of the present study are:
1. To describe the content of current GP care in patients with hip/knee OA, including the
compliance to the SCS.
2. To describe the content of care in physical therapy practice in GP-referred versus self-
referred patients.

Methods

Registration network
‘NIVEL Primary Care Database’ (NPCD) was used to achieve the research objectives9. This
database contains data of several, separated primary care health care providers, including GPs
and physical therapists (box 1). Participating GPs continuously record data on all patient
contacts, including diagnoses, interventions, prescriptions and referrals10. GP-data are
collected since 1992. For this study, data were used from 84 practices participating in NPCD.
These practices provide a representative sample regarding gender and age in comparison with
Dutch National Statistics. Participating physical therapists collect longitudinal data on patient
characteristics, referrals, diagnoses, interventions and evaluations11. This part of the NPCD
was constructed in 2001 and contains data of about 100 physical therapists, divided over 35
outpatient practices. The geographical distribution and the degree of urbanisation of the
participating practices are in line with all Dutch physical therapy practices. In contrast to the
representativeness of participating practices, participating physical therapists are more often
male and are older compared to non-participating Dutch physical therapists.

BOX 1. NIVEL Primary Care Database

NIVEL Primary Care Database (In Dutch: NIVEL Zorgregistraties eerste lijn) uses routinely recorded data from health care
providers to monitor health and utilisation of health services in a representative sample of the Dutch population. It
includes data on health problems and treatment. The aim of NIVEL Primary Care Database is to monitor developments in
health and the use of primary health services in the Netherlands.

Participants of NIVEL Primary Care Database are:

General practitioners, physical therapists, exercise therapists, dieticians, primary care psychologists, GP out-of-hours
services, health centres.
Gathered data are combined and supplemented with data of pharmaceutical care and secondary care collected by other
organisations.

Privacy
NIVEL handles the data in accordance with the Dutch Data Protection Act. Researchers have no access to identifiable
patient information, such as name, address or citizen service number. Research results cannot be traced back to
individual persons, health care providers or health care organisations. Participating health care providers may withdraw
from NIVEL Primary Care Database at any time, and without stating reasons.

Governance
Steering committees with representatives from national associations of health care providers decide about the use of the
data.

Current OA Care | 29
Participants
During 2006 to 2011, all patients with OA of the hip and/or knee who visited their GP and/or
a physical therapist were selected from the NPCD. Hip/knee OA was operationalized by the
‘International Classification of Primary Care’ (ICPC)12 codes L89 (hip OA) or L90 (knee OA).
In physical therapy practice, in case of lacking ICPC codes, patients with hip/knee OA were
identified by national diagnosis codes for allied health care, which are mandated by insurers.

Medical record data

General practice – In general practice, first, patient characteristics (gender, age, location of OA
(hip/knee/both hip and knee)) were collected. Second, GPs’ interventions were gathered,
including (telephone) consults, home visits, prescriptions, and referrals. Prescriptions were
registered according to the Anatomical Therapeutic Chemical (ATC) classification system13.
With respect to the referrals, only referrals to physical therapists, dieticians, and orthopaedic
surgeons were collected since these health care providers take part in the SCS.

Physical therapy – As in general practice, gender, age, and the location of OA of the participants
were collected. Per treatment episode due to hip/knee OA, the applied interventions
(information & advise/manual techniques/physical agent modalities/exercise therapy) and the
amount of care (duration and number of sessions) were collected. Furthermore, to evaluate
the effectiveness of the physical therapy episode, it was examined to what extent the
formulated treatment goals were achieved (<25% / 25-50% / 50-75% / >75%) at the end of a
treatment episode. Finally, the recurrence rate (recurrent complaint yes/no) and the type of
access (referred by GP/referred by medical specialist/direct access) were gathered.

Data analyses
Descriptive statistics were used to describe demographics of the OA population. The content
of current care in general practice was described by considering the use of non-surgical
treatment modalities proposed by the SCS. Operationalization of these treatment modalities
in the NPCD was illustrated in Table 1. Due to the nature of the NPCD, several translations
were necessary to enable interpretation of current registered OA care in terms of the SCS.
Firstly, since education and lifestyle advises both were not registered in the NPCD, it was
assumed that GPs educated their patients when ‘consults’ or ‘visits’ were registered in the
medical record. Research by Noordman et al. showed that patients’ lifestyle is increasingly
discussed during consultations in general practice, especially when it is relevant to patients’
complaints14. Secondly, in the NPCD, prescriptions and referrals were not necessarily directly
linked to a specific diagnosis but to treatment episodes in which prescriptions or referrals
were performed. Therefore, in case of prescriptions, we first selected the four most common
drugs (4-digit ATC) which were applied especially to a diagnosis of hip/knee OA and
subsequently counted the application of these prescriptions (NSAIDs, opioids, other analgesics
and corticosteroids) in treatment episodes due to hip/knee OA. When appropriate, secondary

30
analyses were performed to analyse the application of these prescriptions in more detail (7-
digit ATC). Analyses of referrals occurred similarly; referrals to exercise therapy, dietary
therapy and orthopaedic surgeons were selected.

To determine the compliance of current GP care to the SCS, we assessed the proportion of
patients who had been offered at least one treatment modality of step 1, respectively one or
more treatment modalities of step-1 in addition to the application of a step-2 intervention,
and the proportion of patients who used a step-1 and/or step-2 intervention in addition to a
step-3 intervention.

To compare the content of care of self-referred patients and GP-referred patients visiting a
physical therapist, two sample t-tests and chi-square tests were used, when appropriate. P-
values of <.05 were considered statistically significant.

All statistical analyses were performed using Stata 12 (StataCorp LP, College Station, TX, USA)
software.

TABLE 1. Operationalization content of current care in general practice in patients with hip/knee osteoarthritis
Treatment modality Positively assessed if NPCD contains:
Education or lifestyle advise ≥1 consult or visit at the GP due to hip/knee OA
Prescription of acetaminophen ≥1 prescription of other analgesics and antipyretics*
Step 1 SCS
Prescription of Not separately assessed but included in anti-inflammatory and
glucosaminesulphate anti-rheumatic products, non-steroids **
≥1 prescription anti-inflammatory and anti-rheumatic products,
Prescription of (topical) NSAIDs
non-steroids **
Step 2 SCS Prescription of tramadol ≥1 prescription of opioids †
Referral for exercise therapy ≥1 referral to physical therapy due to hip/knee OA
Referral for dietary therapy ≥1 referral to dietary therapy due to hip/knee OA
Referral to secondary care ≥1 referral to an orthopaedic surgeon due to hip/knee OA
TENS Not assessed
Step 3 SCS
Prescription intra-articular
≥1 Cyriax injection due to hip/knee OA
injections
Remaining ≥1 prescription of corticosteroids for systemic use ‡ (without the
Prescription oral corticosteroid
interventions application of a Cyriax injection)
Abbreviations: NPCD = NIVEL Primary Care Database, OA = osteoarthritis, SCS = Stepped-care strategy6, NSAID = non-
steroidal anti-inflammatory drug *Anatomical Therapeutic Chemical ((ATC) code N02B13; ** ATC code M01A ; † ATC code
N02A ; ‡ ATC code H02A

Current OA Care | 31
Results

Patient characteristics
In total, 12118 patients with hip/knee OA were included from the NCPD; 11248 patients were
extracted from general practice data and 870 patients were identified from physical therapy
data. The majority of patients with hip/knee OA is female and suffers from knee OA. More
patient characteristics are presented in Table 2.

TABLE 2. Characteristics of patients with hip/knee osteoarthritis in general practice and physical therapy practice (2006-2011)
General practice (n=11248) Physical therapy practice (n=870)
Gender, female (n (%)) 7552 (67) 581 (67)
Age, years (mean ± sd) 68.7 ± 12.4 66.7 ± 13.2
Location of OA (n (%))
Hip 4437 (39) 293 (34)
Knee 6462 (57) 577 (66)
Combination of hip and knee OA 349 (3) Not applicable
Abbreviations: OA= osteoarthritis, sd = standard deviation

Content of GP care
Figure 1 summarizes the content of current care in general practice, considering the different
steps of the SCS. In total, 84% of the population was treated by at least one of the step-1
modalities, 21% was treated by any step-2 modality, and 18% received any step-3 intervention.
Three percent of the patients received analgesics. NSAIDs were more frequently prescribed:
more than two out of three patients treated by a step-2 intervention received NSAIDs. In 40%
of the cases, the prescription of NSAIDs concerned Diclofenac or Diclophenac combinations.
Ibuprofen, Meloxicam and Naproxen were prescribed in respectively 12%, 12% and 11% of
the cases. In terms of numbers, referrals to orthopaedic surgeons were more often registered
in the medical records than referrals to physical therapists (exercise therapy) and dieticians.

32
 Content of current care in patients with hip/knee osteoarthritis in
Dutch general practice (2006-2011) (n=11 248)

STEP 1 STEP 2 STEP 3

100%
84% 83%
80%

60%

40%
21% 18%
20% 14% 13%
3% 5% 4% 5%
0,1%
0%

Prescription Tramadol
Any STEP 2 intervention

Referral for exercise therapy

Any STEP 3 intervention

*Education / lifestyle advice

Referral to secondary care

Prescription acetaminophen

Prescription (topical) NSAIDs

Any STEP 1 intervention

Intra-articular injections
Referral for dietary therapy

* Since education and lifestyle advises both were not registered in the NIVEL Primary Care Database,
it was assumed that GPs educated their patients when ‘consults’ or ‘visits’ were registered in the medical record14.

FIGURE 1. Content of current care in patients with hip/knee osteoarthritis in Dutch general practice (n=11248)

The extent to which GPs currently act in agreement with the SCS is shown in Table 3. It shows
that 85% of the population who was treated by a step-2 modality also received any step-1
intervention. Seven percent exclusively received acetaminophen and six percent of the patients
treated by a step-2 modality additionally received both step-1 interventions. Furthermore, in
addition to the application of a step-3 intervention, 90% was treated by a step-1 modality as
well. Twenty-seven percent of the step-3 population additionally received any step-2
intervention, mostly by NSAIDs (18%) and/or tramadol (7%). Two percent received both a
prescription of NSAIDs or Tramadol and was referred to an allied health care provider. None
of the patients receiving a step-3 intervention was offered all modalities described in step 1
and 2 (Table 3).

Current OA Care | 33
 TABLE 3. Compliance to the Stepped-Care-Strategy in patients with hip/knee OA in Dutch general practice (n=11 248)
n (%)
Step 1: Number of patients who received ≥1 of the advised step-1 modalities 9396 (84)
Education or lifestyle advise* 9332 (99)
Prescription of acetaminophen []H 342 (4)
Both modalities 278 (3)
Step 2: Number of patients who received ≥1 of the advised step-2 modalities 2311 (21)
Any step 1 modality 1961 (85)
Prior application of 1. Education or lifestyle advise* 1947 (85)
step-1 modalities: 2. Prescription of acetaminophen 153 (7)
Both education (1.) & prescription (2.) 139 (6)
Step 3: Number of patients who received ≥1 of the advised step-3 modalities 1988 (18)
Any step 1 modality 1794 (90)
Prior application of 1. Education or lifestyle advise* 1791 (90)
step-1 modalities: 2. Prescription of acetaminophen 68 (3)
Both education (1.) & prescription (2.) 65 (3)
Any step 2 modality 534 (27)
3. Prescription of (topical) NSAID 365 (18)
Prior application of 4. Prescription of tramadol 132 (7)
step-2 modalities: 5. Referral for physical therapy 143 (7)
6. Referral for dietary therapy 5 (<1)
Both prescription (3. or 4.) & referral (5. or 6.) 45 (2)
Prior application of
Both education (1.) & prescription (2. & (3. or 4.)) & referral (5. or 6.) 0 (0)
step-1&2 modalities:
Abbreviations: OA = osteoarthritis, n=Number, NSAIDs=Non-Steroidal Anti-Inflammatory Drug
* Since education and lifestyle advises both were not registered in the NIVEL Primary Care Database, it was assumed that
GPs educated their patients when ‘consults’ or ‘visits’ were registered in the medical record14.

Content of physical therapy

Almost two out of three of the patients suffering from hip/knee OA were referred by a GP
(65%), 15% were referred by a medical specialist, and 20% of the patients with hip/knee OA
visited a physical therapist on their own initiative. In a very small minority of patients (n=10),
physical therapists exclusively provided their patients information and advises. Due to this low
number of patients, an extending evaluation of the implementation of (one element of) step-1
of the SCS in physical therapy practice was hindered. Although education rarely was applied
as singular intervention, in 36% of the GP-referred population and 37% of the self-referred
population, ‘information & advise’ was represented as a part of physical therapists’ treatment
(p=.80) (Table 4).

One difference between GP-referred and self-referred patients concerned the recurrence
rate. Self-referred patients more often presented a recurrent complaint in comparison to GP-
referred patients (p<.01). A trend was indicated with respect to less treatment sessions in the
self-referred population compared to the GP-referred population. In both groups, exercise
therapy was the most applied intervention, followed by manual techniques, and information &
advise (Table 4). The focus of exercise therapy was not equal in both groups; although exercise
therapy focussed on improving impairments of body functions was applied similarly in GP-
referred and self-referred patients, GP-referred patients more often received exercise therapy
focussed on improving limitations in activities compared to self-referred patients (p<.01).

34
 TABLE 4. Treatment characteristics in patients with hip/knee osteoarthritis in Dutch physical therapy practice (n=870)
Total
population
(n=870) Referred by Self-referred p-
GP (n=523) (n=160) value
Disease characteristics
Recurrent complaint, yes (n (%)) a 297 (37) 171 (35) 70 (46) .01
Used interventions in ≥50% of the treatment sessions (n (%)) b *
Information & advice 237 (37) 152 (36) 45 (37) .80
Manual techniques 301 (47) 201 (58) 62 (51) .47
Physical agent modalities 45 (7) 30 (7) 7 (6) .61
Exercise therapy – functions 456 (72) 301 (71) 86 (71) .99
Exercise therapy – activities 225 (35) 164 (39) 31 (26) <.01
Amount of care c
Number of treatment sessions (mean ± sd) 10.0 ± 12.3 10.9 ± 13.5 8.6 ± 11.7 .07
Duration of treatment, weeks (mean ± sd) 9.1 ± 13.4 9.6 ± 12.8 9.0 ± 15.8 .69
Result* d
Treatment goals, ≥75% reached (n (%)) 240 (71) 152 (72) 44 (70) .78
Abbreviations: GP = general practitioner, sd = standard deviation
* Exclusively reported in finished treatment episodes (n=788); Number of missing values in total population: a9%, b20%,
c
15%, d57%

Discussion

In this study, we described the content of care registered in electronic records of 12.118
patients with hip/knee OA visiting their GP and/or physical therapist during 2006 to 2011,
including the compliance to the SCS.

A remarkable result of our study comprised a lower prescription rate of pain medication
(NSAIDs and acetaminophen) in patients with hip/knee OA in comparison to previous
studies15, Belo J, Berger M, Koes B, Bierma Zeinstra SM (unpublished work. The lower use of acetaminophen and NSAIDs
might be explained by the increasing availability of those (low-dosed) drugs over the counter.
As a consequence, the total use of NSAIDs and analgesics in the OA population is probably
underestimated in this study.

In line with the SCS, in Dutch general practice less advanced treatment modalities are generally
more often applied than more advanced treatment modalities. However, only a small minority
of patients is treated by a combination of different interventions belonging to one step before
turning to the next step, within the time frame of our study. Most deviations from the SCS
concern GPs’ prescriptions and their referral policy. With respect to GPs’ prescribed pain
medication, our results show that NSAIDs (especially Diclophenac (combinations), Ibuprofen,
Meloxicam and Naproxen) and tramadol (step-2 interventions) are more often prescribed
than analgesics (step-1 intervention). This prescription behaviour has previously been indicated
in an observational study by Cardol et al.15. Moreover, a more recent study investigating GPs’
attitudes regarding SCS recommendations, showed that 21% of the GPs (strongly) agree with

Current OA Care | 35
the statement ‘NSAIDs should be the first choice of pain medication in patients with OA’16.
Given the recognized increased risk of several adverse outcomes in older adults due to the
frequent use of NSAIDs and to improve guideline adherence, GPs could be advised to optimize
the analgesics policy prior to consider NSAIDs prescription in patients with hip/knee OA17.
Besides the prescription policy, deviations from the SCS are found regarding GPs’ referrals as
well. In the NPCD, GPs registered fewer referrals to allied health care providers (exercise
therapy, dietary therapy (step-2) than to orthopaedic surgeons (step-3). Partially, this could
be explained by the moderate quality of the referral-registration in the medical records and
the introduction of patient self-referral for allied health care. However, previous work, which
has been published prior to the introduction of direct access of allied health care, also showed
a lower referral rate for physical therapy compared to orthopaedic surgery15. Therefore, the
question arises whether GPs could improve care by first ensuring optimal non-surgical care in
primary care setting has been delivered, before referring to secondary care18. Fortunately,
recent (unpublished) research in a population in which the SCS has been implemented showed
that patients who are referred to secondary care are significantly more extensively treated by
non-surgical interventions in primary care compared to patients who were not referred to
secondary care (Barten JA, Smink AJ, Swinkels ICS, et al.)

The introduction of direct access to allied health care for example aimed to achieve a
rearrangement of health care organization. Translated to OA care, it could have been
expected that non-pharmacological step-1 interventions had been integrated in physical
therapists’ treatment in case of patient self-referral. However, we did not indicate a difference
with respect to the application of ‘information and advice’ between GP-referred and self-
referred patients in physical therapy practice. Besides, only a handful of patients exclusively
received education. The rearrangement of care, hence, seems to be in its infancy. It should be
remarked that almost half of the patients using self-referral presented recurrent complaints
(46%). These patients might have been treated by a step-1 intervention by a physical therapist
or their GP, prior to the timeframe of this study. Further research is recommended to able
an evaluation of the effects of task-shifting in OA-care.
As already mentioned, self-referred patients with hip/knee OA often present recurrent
complaints in physical therapy practice. In accordance with studies in the general population
and in patients with low back pain, the recurrence rate in self-referred patients significantly
exceeds the recurrence rate in GP-referred patients19,20. Patients with recurrent complaints
might be more aware of direct accessibility and, therefore, are more likely to omit their GP in
case of recognizable musculoskeletal complaints. This rationale is confirmed by research of
Leemrijse et al.19, indicating that the use of direct access was significantly higher in patients
who received earlier treatment by a physical therapist.
Another difference between self-referred and GP-referred patients concerned the less
frequent application of activities-related exercise therapy in self-referred patients. Commonly,
treatment starts with improving impairments of body functions and gradually shifts to
diminishing limitations in activities of daily life. At the same time, the role of the physical
therapist changes from ‘hands-on therapist’ to ‘coach’ and the frequency of treatment sessions

36
decreases. Possibly, this gradual phase out is less often used in patients who refers themselves.
Physical therapists might focus on improving impairments, leaving the translation to activities
of daily life to patients themselves. This situation stands to reason since a sizeable proportion
of the self-referred patients has already gained some experience in the translation to daily life:
recurrence rates are high. Furthermore, the lower amount of care in self-referred patients
seems to support this rationale.

This study has some limitations. Firstly, in the NPCD, treatment episodes in general practice
are constructed retrospectively. As a consequence, applied interventions (consults,
prescriptions and referrals) were related to a treatment episode due to OA, unless they were
aimed at treating any comorbidity. Secondly, both the increasing use of direct-access and the
moderate registration of referrals in the medical record could have induced an
underestimation of referrals to other health professionals, including physical therapy and
dietary therapy. Since exercise therapy and encouraging weight loss are key recommendations
in clinical guidelines for the treatment of lower limb OA5, a higher referral rate than the
indicated 5% respectively <1% could have been expected. Thirdly, we did not take into account
the hierarchical structure of the data with patients nested in health professionals, nested in
primary care practices both in general practice as well as in physical therapy practice.
However, previous work showed that variances in health care use in patients with hip/knee
OA were mainly located at patients’ level21. Fourthly, we were not able to evaluate thoroughly
the sequence of the applied interventions in general practice, but evaluated which
interventions from each step were applied in patients with hip/knee OA. Furthermore, we did
not take into account whether a patient’s treatment was evaluated during an evaluation visit
before turning to the next step, which is described as an integral part of the SCS6. Finally, data
were extracted from two voluntary-based, separate registrations, both part of the NPCD.
Selection bias could be excluded, as the number of patients objecting to participate in the
NPCD is negligible and participating practices reflects the reality of Dutch general practices.
As the NPCD comprises several, separate registrations, patients referred for physical therapy
were not necessarily represented in the GP data and vice versa, disabling an evaluation of the
compliance to the SCS in a singular patient by combining electronic data derived from several
health professionals. At this moment, the NPCD is prepared to enable integration of data from
several health professionals belonging to a singular patient. This opens the way to evaluate the
compliance to the SCS more thoroughly, including the effects of using direct accessibility of
allied health care on both patient-outcomes and the process of care.

Conclusion

In accordance with the SCS, less advanced treatment modalities are more often applied in
general practice than more advanced treatment modalities. However, completion of each SCS-
step is achieved rarely. To optimize the adherence to the SCS, GPs could reconsider their

Current OA Care | 37
analgesics policy prior to NSAID prescription and the low referral rate to exercise therapy
and/or dietary therapy compared to orthopaedic surgeons. Self-referral to physical therapy
partially distorts both the low referral rate in general practice and the low application rate of
education as singular intervention in physical therapy practice. Compared to GP-referred
patients, self-referred patients seems to be less intensively treated, possibly as a result of a
more impairment-minded treatment strategy. This chosen strategy could be related to the
higher recurrent rate in self-referred patients. Further research is recommended to evaluate
more thoroughly the effects of task-shifting in OA care, taking into account the content and
sequence of the SCS.

Acknowledgement

NIVEL Primary Care is funded by the Dutch Ministry of Health, Welfare and Sports. By this
funding, data presented on [Link]/zorgregistraties are collected, analysed and prepared
for presentation. The Ministry did not play any role in the content of neither data-collection
nor data-analyses. We would like to thank research-assistants concerned in NIVEL Primary
Care Database for their cooperation in data-collection and data-analyses.

38
References of 30 years of reform thought. Eur J Public
Health 2006, 16(suppl 1):34-35.

1. Litwic A, Edwards M, Dennison E, Cooper C: 8. Swinkels I, Kooijman M, Spreeuwenberg P,

Epidemiology and burden of osteoarthritis. Br Bossen D, Leemrijse C, van Dijk C, Verheij R,
Med Bull 2013, 105:185-199. de Bakker D, Veenhof C: An overview of 5
years of patient self-referral for physical
2. Hootman J, Helmick C: Projections of US therapy in the Netherlands. Phys Ther 2014.
prevalence of arthritis and associated activity
limitations. Arthritis Rheum 2006, 54(1):226- 9. Netherlands institute for health services
229. research: NIVEL Primary Care Database. In:
NIVEL. 2014.
3. Fernandes L, Hagen KB, Bijlsma JW,
Andreassen O, Christensen P, Conaghan PG, 10. Ursum J, Verheij R: Verantwoording
Doherty M, Geenen R, Hammond A, Kjeken I huisartsen. In: NIVEL Primary care database (In
et al: EULAR recommendations for the non- Dutch: NIVEL Zorgregistraties eerste lijn).
pharmacological core management of hip and 2014.
knee osteoarthritis. Ann Rheum Dis 2013,
72(7):1125-1135. 11. Leemrijse C, Swinkels I: Verantwoording
fysiotherapeuten. In: NIVEL Primary care
4. Hochberg M, Altman R, Toupin April K, database (In Dutch: NIVEL Zorgregistraties
Benkhalti M, Guyatt G, McGowan J, Towheed eerste lijn). 2013.
T, Welch V, Wells G, Tugwell P: American
College of Rheumatology 2012 12. World Organization of Family Doctors: ICPC-
recommendations for the use of 2-R. International Classification of Primary
nonpharmacologic and pharmacologic Care, Revised Second Edition edn. Oxford:
therapies in osteoarthritis of the hand, hip, and Oxford University Press; 2005.
knee. Arthrits Care Res (Hoboken) 2012,
64(4):465-474. 13. World Health Organisation Collaborating
Centre for Drug Statistics Methodology:
5. Zhang W, Moskowitz R, Nuki G, Abramson S, Guidelines for ATC classification and DDD
Altman R, Arden N, Bierma-Zeinstra S, Brandt assignment 2010. In. Oslo; 2009.
K, Croft P, Doherty M et al: OARSI
recommendations for the management of hip 14. Noordman J, Verhaak P, van Dulmen S:
and knee osteoarthritis, Part II: OARSI Discussing patient's lifestyle choices in the
evidence-based, expert consensus guidelines. consulting room: analysis of GP-patient
Osteoarthritis Cartilage 2008, 16(2):137-162. consultations between 1975 and 2008. BMC
Fam Pract 2010, 11:87.
6. Smink A, van den Ende C, Vliet-Vlieland T,
Swierstra B, Kortland J, Bijlsma J, Voorn T, 15. Cardol M, van Dijk L, de Jong J, de Bakker D,
Schers H, Bierma-Zeinstra S, Dekker J: Westert G: Tweede Nationale Studie naar
"Beating osteoARThritis": development of a ziekten en verrichtingen in de huisartspraktijk.
stepped care strategy to optimize utilization Huisartsenzorg: wat doet de poortwachter? In.
and timing of non-surgical treatment Utrecht/Bilthoven; 2004.
modalities for patients with hip or knee
osteoarthritis. Clin Rheumatol 2011, 16. Smink A, Bierma-Zeinstra S, Dekker J, Vliet
30(12):1623-1629. Vlieland T, Bijlsma J, Swierstra B, Kortland J,
Voorn T, van den Ende C, Schers H:
7. Groenewegen P, de Jong J, Delnoij G: The Agreement of general practitioners with the
Dutch health insurance law; the accumulation guideline-based stepped care strategy for

List of publicationsjyhbg |39

 patients with osteoarthritis of the hip or knee: physical therapy. Phys Ther 2008, 88(8):936-
a cross-sectional study. BMC Fam Pract 2013, 946.
14:33.
20. Scheele J, Vijfvinkel F, Swinkels I, Bierma-
17. O'Neil C, Hanlon J, Macus Z: Adverse effects Zeinstra S, Koes B, Luijsterburg P: Direct
of analgesics commonly used by older adults access to physical therapy for patients with low
with osteoarthritis: focus on non-opioid and back pain in the Netherlands: prevalence and
opioid analgesics. Am J Geriatr Pharmacother predictors. Phys Ther 2014, 97(3).
2012, 10(6):331-342.
21. Smink A, Dekker J, Vliet-Vlieland T, Swierstra
18. Croft P, Porcheret M, Peat G: Managing B, Kortland JH, Bijlsma J, Teerenstra S, Voorn
osteoarthritis in primary care: the GP as public TB, Bierma-Zeinstra S, Schers HJ et al: Health
health physician and surgical gatekeeper. Br J care use of patients with osteoarthritis of the
Gen Prac 2011, 61(589):485-486. hip or knee after implementation of a stepped
care strategy: An observational study. Arthrits
19. Leemrijse C, Swinkels I, Veenhof C: Direct Care Res (Hoboken) 2013, DOI:
access to physical therapy in the Netherlands: 10.1002/acr.22222.
results from the first year in community-based

40
Factors associated with referral to
3
secondary care in patients with
osteoarthritis of the hip or knee
after implementation of a
stepped-care strategy

Published in: Arthritis Care Res (Hoboken). 2017 Feb;69(2):216-225

[Link]

Di-Janne Barten
Agnes Smink
Ilse Swinkels
Cindy Veenhof
Henk Schers
Thea Vliet Vlieland
Dinny de Bakker
Cornelia van den Ende
Abstract

Objective We introduced a Stepped-Care-Strategy (SCS) for hip/knee

osteoarthritis focusing on delivery of high quality, and stepped care. In this study,
we aimed to identify factors associated with various steps of the SCS.

Methods We used data from a two-year observational prospective cohort study,

including 313 patients visiting their general practitioner (GP) with a new episode of
hip/knee osteoarthritis. We used logistic multilevel analyses to identify factors at
the level of the patient, the GP, and the general practice, related to 1) treatment
limited to primary care, 2) referral to non-surgical secondary care, 3) surgical
procedures.

Results Patients whose treatment had been limited to primary care, tended to
function physically better (OR1.03). Furthermore, they less often received exercise
therapy (OR.46), intra-articular injections (OR.08), and radiological assessments
(OR.06). Continuation of non-surgical care after referral was more likely in
employed patients (OR2.90) and patients who had no exercise therapy (OR.19)
nor non-steroidal anti-inflammatory drugs (OR.35). Surgically treated patients
more often received exercise therapy (OR7.42). Referral and surgical treatment
depend only to a limited extent on the GP or the general practice.

Conclusion After implementation of the SCS in primary care, the performance of

exercise therapy seems to play a key role in the decision whether or not to refer
for (non)-surgical treatment in secondary care, rather than disease severity or
psychological factors. To optimize patient-tailored treatment, future research
should be adressed to determine the optimal moment of switch from primary to
secondary care in patients with hip/knee osteoarthritis.

Stepped OA Care |43

Introduction

Management of osteoarthritis (OA) of the hip or knee comprises non-pharmacological,

pharmacological and surgical interventions1. Clinical guidelines recommend to combine non-
pharmacological and pharmacological treatment options prior to the consideration of surgical
interventions1-4. Surgical treatment, by way of Total Joint Replacement (TJR), is an effective
and cost-effective final treatment option in severe hip/knee OA5,6. Surgery rates in patients
with hip/knee OA differ between 16-23% in primary care cohorts (follow-up time 1-4 years)7,8
and between 18-50% in patients referred to secondary care (follow-up time 1-3 years)9-12. The
absolute number of TJRs increases annually, partly due to the increasing prevalence of OA 13.
However, some have suggested that TJR is overutilized14.

To support physicians to apply non-surgical interventions adequately and to reduce

inappropriate use of advanced care including TJRs, several treatment algorithms are developed
in OA care15,16. These algorithms potentially increase the appropriateness of advanced care,
and ascertain that patients with hip/knee OA will receive the right care at the right time. The
Stepped-Care-Strategy (SCS) is an example of a treatment algorithm in hip/knee OA. The
extended content of the SCS is described elsewhere16. In short, non-surgical care in patients
with hip/knee OA is prescribed in three steps. Treatment starts in primary care with
stimulating patients’ self-care (step 1). In case of persisting complaints after adequate
application of all step-1 interventions, (topical) NSAIDs or tramadol are applied, supplemented
by prescribing exercise therapy and, in case of overweight, dietary therapy to diminish the
impairments and limitations due to OA (step 2). TENS and intra-articular corticosteroid
injections, and a referral for hospital-based evaluation, may be applied as final non-surgical
interventions (step 3) (Appendix). All modalities of step 1 and step 2 of the SCS can be
performed in primary care. The performance of step-3 interventions requires a referral to
secondary care. TJR should exclusively be considered in case of persisting complaints after
completing the SCS.

In 2010, implementation of the SCS in clinical practice has been started. Prior to nation-wide
implementation, general practices in one region of the Netherlands were invited to implement
the SCS in clinical practice; the implementation strategy was described elsewhere17. In this
cohort, forty-five percent of the patients were referred to an orthopaedic surgeon, 16% visited
a rheumatologist. TJR was applied in 18% of the patients17. To date, it is unknown what factors
determine the choice for surgery. These factors may relate to patients, practitioners and
practice characteristics. Insight into these factors would be useful to optimize nationwide
implementation of the SCS in clinical practice.

Therefore, the aim of the present study is to identify patient-related, GP-related and general
practice related factors associated with 1. treatment limited to primary care (step 1 and 2
SCS), 2. continuation of non-surgical secondary care within the subpopulation that was

44
referred (step 3 SCS), and 3. the application of a surgical procedure including TJR in the total
patient population visiting their GP due to hip/knee OA.

Methods

Study design
Data were obtained from a two-year observational prospective cohort study, composed
within the context of the BART project (Beating osteoARThritis), executed from August 2010
to March 2013 in the region Nijmegen, The Netherlands. The study was designed to test and
to evaluate the implementation of the SCS16. The study was approved by the Medical Ethics
Committee on Research Involving Human Subjects (CMO) Region Arnhem-Nijmegen
(approval number: CMO 2009/246).

Study population

General practitioners
GPs from the Nijmegen University Network of General Practitioners (NUHP) were invited
to participate in the BART-project and subsequently were encouraged to optimize their non-
surgical care prior to a referral for TJR-consideration. The NUHP consists of 157 GPs working
in 70 general practices. It is a practice based research network from the department of Primary
and Community Care at the Radboud University Nijmegen. Additionaly, six general practices
which were not involved in the NUHP were approached to take part in the study.

Patients
Potentially eligible patients were recruited by the participating GPs during visits in their
practice. Individuals at least 18 years of age who were seeking help from their GP with a new
episode of hip or knee complaints due to OA without a preceding visit accounting to the same
complaints in the last three months, were eligible to participate. In addition, patient
recruitment occurred on the basis of extractions from medical records of participating GPs;
patients with symptomatic hip/knee OA who did not visit the GP due to these complaints in
the preceeding three months were selected and approached to participate in this study.
Patients were excluded when they had received a total hip or knee replacement in the past,
were on a waiting list for such a surgery or were inable to complete the questionnaires due
to language barriers or terminal illnesses.

Assessments
Participants completed five similar questionnaires during the time frame of the study (baseline,
6 months, 12 months, 18 months, 24 months after inclusion). Their GPs also completed one
questionnaire at baseline and another questionnaire two months after inclusion in the
researchproject. For the current study, baseline data were used, completed with data

Stepped OA Care |45

regarding patients’ utilized care during the study period. Box 1 summarizes the collected data
at the level of the patient, the GP, and the general practice.

BOX 1. Collected descriptive items at baseline on patient-level, general practitioner level and general practice level
Patient level
Predisposing #
Age Years
Sex Male / Female
Overweight Yes / No
Education level Elementary school / High school; technical or vocational training / Higher eduction
Paid job / No paid job
Employment status Alone / With partner / With partner and children / With children / With others
Household composition Dutch General Self Efficacy Scale (DGSS), range 10-40;
higher score reflects higher self-efficacy 18
Self-efficacy Pain Coping Inventory List (PCI), range 12-48;
higher score reflects more use of active coping style 19
Active pain coping
Enabling #
Health insurance Basic / With additional coverage for physiotherapy
Resident urbanity Rural / Suburban / Urban
Illness related #
Location of osteoarthritis Hip / Knee / Both hip and knee
Affected joints Number
Duration of complaints <1 year / 1-5 years / 5-10 years / >10 years
Pain severity Western Ontario McMaster University Index of Osteoarthritis (WOMAC), range 0-100;
WOMAC, range 0-100 (20, 21); higher score reflect better health status 20,21
Functional disability Dutch Arthritis Impact Measurement Scales 22

Number of comorbidities
General practitioner level †
Age Years
Sex Male / Female
Years of experience as general practitioner Years
Special interest in musculoskeletal disorders Yes / No
Practice nurse involved in osteoarthritis management Yes / No
Structural collaboration with other discipline(s) Yes / No
General practice level †
Practice type Solo / Duo / Group / GP centre / Health centre
Practice urbanity Rural / Suburban / Urban
Additional variables
Previously utilized care
Education Yes / No
Lifestyle advise Yes / No
Dietary therapy Yes / No
Exercise therapy Yes / No
Acetaminophen Yes / No
Glucosaminen Yes / No
Non-steriod anti-inflammatory drugs Yes / No
Tramadol Yes / No
Intra-articular corticosteroid injection Yes / No
Radiological assessment Yes / No
# Classification according to Andersen et al.23, distinguishing predisposing factors, enabling factors and factors related the
illness level

46
Dependent variables

Treatment limited to primary care (yes/no) (step 1, step 2 SCS)

Based on data of all participants concerning health care utilization during the study period, it
was determined whether or not a patient’s treatment due to their hip/knee OA was limited
to primary care. Limitation to primary care was defined as ‘no patient-reported referral to an
orthopaedic surgeon or rheumatologist during the study period’. If a patient reported that
he/she was referred to secondary care only for radiological assessment without visiting an
orthopaedic surgeon or rheumatologist, the treatment episode was indicated as ‘limited to
primary care’ as well.

Continuation of non-surgical treatment after referral to secondary care (yes/no) (Step 3 SCS)
In the subgroup of patients referred to an orthopaedic surgeon or rheumatologist, it was
determined whether or not non-surgical treatment was continued. This variable was scored
positively if a patient reported that he/she did not receive any surgical procedure after a
referral to secondary care during the study period.

Application of a surgical procedure including TJR (yes/no)

The variable ‘application of a surgical procedure including TJR’ was scored positively if any
participant of the total population reported that he underwent a surgical procedure, mainly a
total hip replacement or a total knee replacement in secondary care during the study period.

Figure 1 shows the dependent variabeles in outline.

Core:
Total population

Shield 1, dependent variable 1:

 Non-referred population (limited to primary care)
 Referred population to secondary care
Unknown referral status

Shield 2, dependent variable 2:

 Non-surgically treated population after referral
total population 1 2 3 to secondary care
 Surgically treated population after referral
to secondary care
Unknown surgical status

Shield 3, dependent variable 3:

 Surgically treated population
 Non-surgically treated population
Unknown referral status
Unknown surgical status

FIGURE 1. Dependent variables (limitation to primary care, non-surgical treatment after referral to secondary care,
and surgical treatment) of the study in outline

Stepped OA Care |47

Potential associative factors
Potential factors associated with referrals to secondary care and subsequent treatment were
divided into patient-related factors, factors related to the GP, and factors related to the
general practice. We excluded general practice and general practitioner related factors in the
determination of factors related to the continuation of non-surgical treatment in patients with
hip/knee OA referred to secondary care, since GPs are not involved in the decision to apply
a surgical procedure or not. The selection of patient-related potential factors was based on
previously reported factors in medical literature7,11,12,24-31. Indicated factors were categorized
according to Andersen’ Behavioural Model of Health Care, distinguishing a predisposing,
enabling, and illness level component23. Additionally, the content of previously utilized care
was collected, including the application of radiological assessment (Box 1).

Statistical analyses

Descriptive statistics
Baseline characteristics were described, both on patient level as well on the level of the GP
and the general practice. Subsequently, patient demographics, health status, and disease
characteristics were described separately in patients who were not referred to secondary care
versus patients who were referred to secondary care. Finally, descriptive properties of the
referred group were studied in more detail, comparing referred patients continuing non-
surgical treatment and referred patients receiving a surgical procedure. Differences between
groups were analysed using unpaired t-tests or chi-square tests, when appropriate.

Identifying factors associated with the provision of care

Given the hierarchical structure of the dataset, i.e. patients (level 1) were nested in the sample
of GPs (level 2), who are nested in general practices (level 3), logistic multilevel regression
models were built to identify factors associated with the setting of care. To prevent overfitting
due to the number of factors32, the most important factors were selected before fitting the
final model. This preselection was started by investigating collinearity between potential
factors; variables with a variance inflation factor (VIF) higher than 10 were removed.
Subsequently, the selection procedure was performed by classifying the remaining factors into
four blocks. The first block included predisposing and enabling factors, the second block
comprised disease-related factors, the third block included previously applied treatment
modalities, and the fourth block comprised GP and GP-practice variables. Backward stepwise
regression based on five imputed datasets was used to select the most important variables per
block. Cutoff value for selecting a variable was a mean likelihood ratio 2 < .05. The overall
final model was constructed by entering simultaneously the selected variables from each of
the four blocks and was based on 20 imputed datasets combined using Rubin’s rules33-35. For
each dependent variabele, the percentage of variation (Variance Partition Component (VPC))
belonging to each hierarchical level was calculated, using the latent variable approach36,37.

48
All statistical analyses were performed using Stata 10 (StataCorp LP, College Station, TX, USA)
software.

Results

Study population
Included patients were treated by 70 GPs working in 38 general practices. Participating GPs
were more often male and worked less often in solo-practices compared to Dutch GPs38. In
total, 313 patients with hip/knee OA participated in the study. Nine out of ten included
patients exceeded the age of 51. The majority of the population was female (62%), suffered
from at least knee OA (79%) and was faced with physical complaints more than one year (79%)
(Table 1). Slightly more than half of the patients were referred to secondary care during the
study period. In the subpopulation of referred patients, about one in three patients received
a surgical intervention.

Stepped OA Care |49

 TABLE 1. Baseline characteristics of patients with hip/knee osteoarthritis, their general practitioners and general practice
Baseline characteristic Missing values (n)

Patient characteristics (n=313)

Demographics
Age, years; mean (SD) 64 (10) 0
Sex, male; n (%) 120 (38) 0
Education, higher education; n (%) 37 (20) 126
Employed, paid work; n (%) 96 (31) 1
Household composition, with partner; n (%) 229 (73) 1
Health insurance, with additional coverage; n (%) 282 (91) 2
Patients’ urbanity, rural; n (%) 194 (62) 1
Health status
Overweight, Body Mass Index >25kg/m2; n (%) 218 (71) 4
Number comorbidities range (0-15); median (IQR) 2 (1-3) 0
Self-efficacy (Dutch General Self Efficacy Scale) *; mean (SD) 31 (5) 8
Active pain coping (Pain Coping Inventory List) †; mean (SD) 26 (6) 8
Disease characteristics
Location of osteoarthritis
Hip; n (%) 159 (51) 0
Knee; n (%) 246 (79) 0
Number of affected joints (range ); median (IQR) 1 (1-3) 0
Duration of complaints, > 1 year; n (%) 247 (79) 1
WOMAC pain ‡; mean (SD) 62 (22) 8
WOMAC functioning ‡; mean (SD) 64 (21) 12
GP-characteristics (n=70)
Age, years; mean (SD) 49 (9) 4
Sex, male; n (%) 51 (73) 0
Experience as GP, years; median (IQR) 17 (10-25) 4
Special interest in musculoskeletal disorders, yes; n (%) 6 (10) 12
Practice nurse involved in osteoarthritis managemen, yest; n (%) 13 (22) 12
Structural collaboration with other discipline(s), yes; n (%) 20 (34) 12
General practice characteristics (n=38)
Practice type, solo; n (%) 6 (17) 3
Practice urbanity, rural; n (%) 23 (61) 0
Abbreviations: n=Number, SD=Standard Deviation, IQR=Interquartile Range, WOMAC=Western Ontario McMaster University
Index of Osteoarthritis, GP=General Practitioner

* Higher scores reflect higher self-efficacy (ranging from 10 to 40); † Higher score indicates more use of an active coping
style (ranging from 12 to 48); ‡ Standardized scores were used where higher scores reflect better health status (ranging
from 0 to 100)

Non-referred versus referred patients

In Table 2, patient demographics, health status, disease characteristics and the content of
previously utilized care were shown, both in non-referred patients and patients who were
referred to secondary care due to their hip/knee OA. Non-referred patients reported less
OA-affected joints and less comorbidities, indicated less pain and fewer limitations in activities
compared to referred patients (WOMAC-scores). Moreover, non-referred patients showed
a less active coping style and were treated less extensively by recommended non-surgical
treatment modalities compared to patients referred to secondary care (Table 2).

50
 TABLE 2. Patient-related characteristics and the content of care in patients with hip/knee osteoarthrtis, separate for
non-referred (n=135) and referred (n=154) patients (including surgically treated population) *
Non-referred to Referred to p-value
secondary care secondary care
(n=135) (n=154)
Demographics
Age, years; mean (SD) 64 (11) 63 (10) .25
Sex, male; number (%) 56 (41) 54 (35) .26
Education, higher education; number (%) 15 (17) 22 (22) .42
Employed, paid work; number (%) 38 (28) 50 (33) .41
Health insurance, additional coverage physical therapy; number (%) 119 (89) 142 (93) .24
Household composition, with partner; number (%) 101 (75) 117 (76) .91
Patients’ urbanity, rural; number (%) 85 (63) 97 (63) .94
Health status
Overweight, Body Mass Index >25kg/m2; number (%) 87 (65) 111 (73) .16
Comorbidities, number; median (Interquartile Range) 1 (0-2) 2 (1-3) <.01
Self efficacy , Dutch General Self-efficacy Scale §; mean (SD) 31 (5) 31 (5) .58
Active pain coping, Pain Coping Inventory list †; mean (SD) 25 (6) 27 (5) .02
Disease characteristics
Location of osteoarthritis
Hip; number (%) 72 (53) 77 (50) .58
Knee; number (%) 98 (73) 127 (82) .04
Affected joints, number; median (Interquartile Range) 1 (1-3) 2 (1-3) .02
Duration of complaints, > 1 year; number (%) 98 (73) 129 (84) .02
WOMAC pain ‡; mean (SD) 70 (20) 55 (22) <.01
WOMAC functioning ‡; mean (SD) 73 (19) 57 (20) <.01
Content of previously utilized care, yes; number (%)
Education 97 (72) 136 (91) <.01
Lifestyle advise 86 (64) 117 (79) <.01
Dietary therapy (if overweighted) 7 (8) 18 (17) .06
Exercise therapy 62 (46) 115 (77) <.01
Acetaminophen 98 (73) 137 (90) <.01
Glucosaminen 44 (33) 49 (34) .90
NSAIDs 56 (42) 89 (62) <.01
Tramadol 5 (4) 35 (25) <.01
Intra-articular corticosteroid injection 8 (6) 51 (35) <.01
Radiological assessment 61 (45) 141 (94) <.01
Abbreviations: SD=Standard Deviation, WOMAC=Western Ontario McMaster University Index of Osteoarthritis, NSAID=Non-
Steroidal Anti-Inflammatory Drug
* In 24 patients it was unknown whether or not a referral to secondary care has happened; § Higher scores reflect higher
self-efficacy (ranging from 10 to 40); † Higher score indicates more use of an active coping style (ranging from 12 to 48);
‡ Standardized scores were used where higher scores reflect better health status (ranging from 0 to 100)

About one-third (34%) of the population who had been referred to secondary care, received
a surgical procedure, which represents 18% of the total population. Frequently, the surgical
interventions concerned TJRs. Surgically treated patients were older, were less often
employed, showed a more active coping style and reported less OA-affected joints than non-
surgically treated patients. Acetaminophen, NSAIDs and exercise therapy had been more often
applied in patients who received a surgical procedure compared to non-surgically treated
referred patients. In contrast, glucosamine was less often used by surgically treated patients
than by patients who continued non-surgical care (Table 3).

Stepped OA Care |51

 TABLE 3. Patient-related characteristics and the content of care in patients with
hip/knee osteoarthrtis who were referred to secondary care
Referred to secondary care (n=142)a
No surgical
procedure Surgical procedure p-
(n=93) (n=49) value
Demographics
Age, years; mean (SD) 61 (11) 65 (7) .02
Sex, male; number (%) 30 (32) 20 (41) .31
Education, higher education; number (%) 15 (24) 7 (18) .50
Employed, paid work; number (%) 34 (37) 10 (20) .04
Health insurance, additional coverage for physical therapy; 85 (91) 45 (94) .62
number (%)
Household composition, with partner; number (%) 68 (73) 42 (86) .09
Patients’ urbanity, rural; number (%) 56 (60) 37 (76) .07
Health status
Overweight, Body Mass Index >25kg/m2; number (%) 70 (77) 33 (67) .22
Comorbidities, number; median (Interquartile Range) 2 (1-3) 1 (1-3) .35
Self efficacy , Dutch General Self-efficacy Scale §; mean (SD) 31 (6) 31 (5) .68
Active pain coping, Pain Coping Inventory list †; mean (SD) 26 (5) 28 (5) .02
Disease characteristics
Location of osteoarthritis
Hip; number (%) 42 (45) 27 (55) .26
Knee; number (%) 77 (83) 39 (80) .64
Affected joints, number; median (Interquartile Range) 2 (1-4) 1 (0-3) .02
Duration of complaints, > 1 year; number (%) 79 (85) 40 (83) .80
WOMAC pain ‡; mean (SD) 57 (23) 53 (19) .34
WOMAC functioning ‡; mean (SD) 59 (22) 54 (16) .17
Content of previously utilized care, yes; number (%)
Education 82 (88) 46 (94) .28
Lifestyle advise 72 (77) 38 (78) .99
Dietary therapy (if overweighted) 12 (17) 5 (16) .85
Exercise therapy 62 (67) 45 (92) <.01
Acetaminophen 78 (84) 48 (98) .01
Glucosaminen 35 (38) 9 (19) .02
NSAIDs 49 (53) 37 (77) <.01
Tramadol 20 (22) 13 (27) .48
Intra-articular corticosteroid injection 32 (34) 15 (31) .71
Radiological assessment 84 (90) 48 (98) .09
Abbreviations: SD=Standard Deviation, WOMAC=Western Ontario McMaster University Index of Osteoarthritis, NSAID=Non-
Steroidal Anti-Inflammatory Drug.
a
In 12 patients, it was unknown whether the referral resulted in a surgical procedure / continuation of non-surgical care
§ Higher scores reflect higher self-efficacy (ranging from 10 to 40); † Higher score indicates more use of an active coping
style (ranging from 12 to 48); ‡ Standardized scores were used where higher scores reflect better health status (ranging
from 0 to 100)

52
Factors associated with the provision of care
Thirteen factors were identified which were associated with at least one dependent variable
(Table 4). Most of these factors were related to the patient and to the content of previously
utilized care. No GP-related factors and only one general practice related factor were part of
the final models. Due to collinearity, the variable ‘experience years as GP’ was removed before
starting logistic multilevel analyses.

Treatment limited to primary care (step 1, step 2 SCS)

The analyses showed that a limitation to treatment in primary care was associated with fewer
limitations in physical functioning (p=<.01) and with no-use of exercise therapy (p=.03), no-
use of intra-articular injections (p=<.01) and absence of radiological assessments (p=<.01).
Moreover, a positive trend was indicated with respect to being treated in a solo practice
(p=.06). In total, more than 99.9% of the variance was located at the level of the patient.

Continuation of non-surgical treatment after referral to secondary care (step 3 SCS)

Eight variables at patient level were added to the final model determining factors related to
the continuation of non-surgical care after referral to secondary care. In the subpopulation of
referred patients, having a paid job (p=.04), more OA-affected joints (p=.04), no-use of
exercise therapy (p=<.01) and no-use of NSAIDs (p=.03) were associated with continuing
non-surgical treatment.

Application of surgical procedure

The final model fitted to identify factors associated with the application of a surgical procedure
comprised six variables. Only one factor was indicated to be statistically significantly associated
with a surgical procedure. The application of a surgical procedure was more likely after the
use of exercise therapy (p<.01). Variance was mainly located at the level of the patient (89%),
followed by the GP (10.3%) and the general practice (<1%).

Stepped OA Care |53

TABLE 4. Factors related to the setting of care in patients with hip/knee osteoarthritis (logistic multilevel regression analyses)
Non-referral in Continuation of non-surgical Surgical procedure in primary
primary care population care in referred population care population
(Nyes=135; Nno=154) (Nyes=93; Nno=49) (Nyes=49; Nno=229)
Factors related to: ORa 95% CI p-value ORb 95% CI p-value ORc 95% CI p-value
Patient
Predisposing component
Active pain coping, PCI
(range 12-48)† 1.01 .95 – 1.07 .79 .93 .85 – 1.01 .08 1.06 0.98 – 1.14 .15
Household composition,
with partner; .53 .12 – 1.51 .19
Employed, paid work 2.90 1.05 – 7.82 .04
Illness level component
WOMAC functioning
(range 0-100)‡ 1.03 1.01 – 1.05 <.01 1.01 .99 – 1.04 .35 .99 0.97 – 1.01 .23
Affected joints, number
(range 0-9) 2.00 1.05 – 3.82 .04
Content of care
Use of glucosamine 1.67 .61 – 4.60 .32 .48 .20 – 1.13 .09
Use of exercise therapy .46 .23 – .92 .03 .19 .05 – .65 <.01 7.42 2.38 – 23.11 <.01
Use of NSAIDs .35 .13 – .91 .03 2.19 .93 – 5.16 .07
Use of tramadol .28 .07 – 1.08 .07
Use of radiological
assessment .06 .03 – .16 <.01
Use of intra-articular
injections .08 .03 – .22 <.01
Use of acetaminophen 4.86 .56 – 42.15 .15
GP-(practice)
Solo practice, yes 2.46 .97 – 6.26 .06
Abbreviations: n=Number, OR=Odds Ratio, CI=Confidence Interval, PCI=Pain Coping Inventory list, WOMAC=Western Ontario
McMaster University Index of Osteoarthritis, NSAID=Non-Steroidal Anti-Inflammatory Drug, GP=General Practitioner
† Higher score indicates more use of an active coping style; ‡ Standardized scores were used, higher scores reflect better
health status
a Backward stepwise final model: likelihood ratio 2 = 178.06, df = 8, p<.001; McFadden R-squared = .41;
b Backward stepwise final model: likelihood ratio 2 = 41.12, df = 9, p<.001; McFadden R-squared = .21;
c Backward stepwise final model: likelihood ratio 2 = 57.30, df = 7, p<.001; McFadden R-squared = .19

Discussion

The objective of this study was to identify factors related to the patient, the GP and the general
practice associated with the setting of care in a population of patients with hip/knee OA after
implementation of the SCS. Our study shows that in a two years follow-up period, 47% of the
population visiting their GP with hip/knee OA were exclusively treated in primary care. Non-
referral to secondary care seems to be stronger associated with the content of previously
utilized care than with patient-demographics, disease-related factors or factors related to the
GP or general practice. Based on our study, patients seem to be referred more often to
secondary care after receiving exercise therapy, intra-articular injections, and radiological

54
assessments. This suggests that participating GPs try to act in agreement with the SCS by
offering non-surgical modalities before considering surgical interventions.

The proportion of patients who underwent a surgical procedure in our study was similar to
previously reported data in primary care populations7,8. Several studies indicated more pain
and a higher level of functional limitations as important factors with respect to the decision to
apply a TJR7,11,12,27. However, we did not find an association between receiving a surgical
procedure and any disease characteristics. One difference between our study and previous
studies is that we included additionally GP and general practice related factors as well as the
content of previously utilized care to patient-related factors. Including these factors, the
extent of health care utilization seems to be a stronger predictive factor for the application of
a surgical procedure than disease characteristics. In previous research, it has been shown that
a higher level of illness is accompanied by an increased health care utilization39. For example,
in the two-year period preceding TJR, patients have been using large amounts of painkillers
and more outpatient health care services40. This supports the principles of the SCS, suggesting
TJR as final treatment option when non-surgical modalities no longer establish satisfactory
results.

In our study, 34% of the patients referred to secondary care received a surgical procedure.
Previous work by McHugh et al. demonstrated a surgery-rate of 50% in a referred population11.
The latter discrepancy can possibly be explained by the nature of both populations. The
McHugh study included patients newly referred to orthopaedic surgeons for TJR
consideration. In contrast, in our study, patients could have been referred to both orthopaedic
surgeons and rheumatologists due to many reasons, including TJR consideration and a request
for a second opinion for example. Continuation of non-surgical care after a referral was hardly
related to predisposing or enabling factors. One exception concerns the patients’ employment
status. In contrast to previous findings, we found that ‘having paid employment’ was
significantly associated with non-surgical treatment26. Possibly, in view of the current labour
shortage due to the financial crisis, patients prefer to prevent absenteeism rather than relieving
their complaints by a surgical procedure.

Each additional joint affected with OA doubles the chance to continue non-surgical treatment
after a referral to secondary care. Since previous research in patients with hip/knee OA has
shown that multi-joint complaints adversely impact postoperative results41,42, patients and/or
orthopaedic surgeons possibly are reticent to turn to a surgical intervention and prefer to
continue non-surgical treatment. As the incidence of multi-joint OA is expected to increase
over the next years43, it is desirable to get more insight into effective interventions in this
specific population.

In general, identified factors associated with non-surgical treatment limited to primary care
respectively the application of TJR in secondary care, are mainly located at the level of the
patient (explaining ≥ 89% of the variance). A systematic review by O’Donnell et al. investigating

Stepped OA Care |55

the variation in GP referral rates in the general population, reported the highest attribution
to patient-characteristics as well44. However, it could be discussed whether the content of
care is primarily related to the patient, rather than being determined by a patient’s GP.
Probably, in daily practice, patients agree with the proposed care by the GP, prior to its’ actual
application.
In our study, only one practice-related factor showed a positive trend with respect to
treatment limited to primary care: being treated in a solo-practice. Previous work presented
inconclusive results regarding the association between practice size and referral rate in the
general population44.

This study comprises secondary analyses of data collected by the BART project. To our
knowledge, the BART project is the first project in which multi-levelled factors associated with
the provision of care in patients with hip/knee OA were identified. Our results show that
cohort studies are useful to get insight into health services utilization over time as long as
longitudinal, multidisciplinary medical record registrations are lacking. However, this study has
some limitations as well. Firstly, the content of utilized care was collected by a self-reporting
tool, which could have induced an underestimation of health services utilization. Secondly, data
were not available regarding some potential relevant factors, for example the radiological
severity of hip/knee OA and a patient’s willingness to consider TJR. These factors were
indicated to be determinants for TJR in earlier work7,27,29,30. Thirdly, since our study sample
was limited, we had to preselect the most important factors for inclusion in the final model
instead of using all potential predictive factors. Despite the chosen method comprising
preselection within content-matter motivated blocks using backward selection regression has
been found to be a good alternative, statistical power seems to be limited to detect less strong
associations. Finally, in each of the final models, the McFadden R-squared was low. As a
consequence, data presented in this study could not be used to predict the care pathway of
an individual patient.

In conclusion, after the implementation of the SCS in a primary care population, the
performance of exercise therapy has shown to play a key role in the decision whether or not
to refer for (non)surgical treatment in secondary care. Demographics, disease characteristics
and health status are only slightly associated with the provision of OA care. This suggests that,
in line with the SCS, GPs try to offer non-surgical modalities before appealing to surgical
interventions. In future, research is desirable to determine the optimal moment of switch from
primary care to secondary care, ensuring the best tailored treatment in patients with hip/knee
OA.

56
References osteoarthritis in the primary care setting and
factors associated with general practitioners’
indication for prosthetic replacement within 1
1. Zhang W, Moskowitz RW, Nuki G, Abramson year Osteoarthritis Cartilage. 2008;16:8.
S, Altman RD, Arden N, et al. OARSI
recommendations for the management of hip 9. Conaghan P, D'Agostino M, Le Bars M, Baron
and knee osteoarthritis, Part II: OARSI G, Schmidely N, Wakefield R, et al. Clinical and
evidence-based, expert consensus guidelines. ultrasonographic predictors of joint
OsteoarthritisCartilage. 2008;16(2):137-62. replacement for knee osteoarthritis: results
from a large, 3-year, prospective EULAR study.
2. Fernandes L, Hagen KB, Bijlsma JW, Ann Rheum Dis. 2010;69:4.
Andreassen O, Christensen P, Conaghan PG,
et al. EULAR recommendations for the non- 10. Hamel M, Toth M, Legedza A, Rosen M. Joint
pharmacological core management of hip and replacement surgery in elderly patients with
knee osteoarthritis. Ann Rheum Dis. severe osteoarthritis of the hip or knee. Arch
2013;72(7):1125-35. Intern Med. 2008;168(13):11.

3. Hochberg MC, Altman RD, April KT, Benkhalti 11. McHugh GA, Campbell M, Luker KA. GP
M, Guyatt G, McGowan J, et al. American referral of patients with osteoarthritis for
College of Rheumatology 2012 consideration of total joint replacement: a
recommendations for the use of longitudinal study. Br J Gen Pract.
nonpharmacologic and pharmacologic 2011;61(589):e459-e68.
therapies in osteoarthritis of the hand, hip, and
knee. Arthritis Care Res(Hoboken). 12. Zeni J, Axe M, Snyder-Mackler L. Clinical
2012;64(4):465-74. predictors of elective total joint replacement
in persons with end-stage knee osteoarthritis.
4. National Collaborating Centre for Chronic BMC Musculoskel Disord. 2010;6(11):86.
Conditions. Osteoarthritis: National clinical
guideline for care and management in adults. 13. Cram P, Lu X, Kates S, Singh J, Li Y, Wolf B.
London: Royal College of Physicians; 2008. Total knee arthroplasty volume, utilization,
and outcomes among Medicare beneficiaries,
5. Chang R, Pellisier J, Hazen G. A cost- 1991-2010. JAMA. 2012;308:10.
effectiveness analysis of total hip arthroplasty
for osteoarthritis of the hip. JAMA. 14. Ghomrawi H, Schackman B, Mushlin A.
1996;275(11):8. Appropriateness criteria and elective
procedures - total joint arthroplasty. New Eng
6. Hawker GA, Badley E, Croxford R, Coyte P, J MED. 2012;367(26):3.
Glazier P, Guan J, et al. A population-based
nested case-control study of the costs of hip 15. Bruyère O, Cooper C, Pelletier J, Branco J,
and knee replacement surgery. Med Care. Brandi M, Guillemain F, et al. An algorithm
2009;47(7). recommendation for the management of knee
osteoarthritis inEurope and internationally: a
7. Birrell F, Afzal C, Nahit E, al. e. Predictors of report from a task force of the European
hip joint replacement in new attenders in Society for Clinical and Economic Aspects of
primary care with hip pain. Br J Gen Prac. Osteoporosis and Osteoarthritis (ESCEO).
2003;53(486):26-30. Semin Arthritis Rheum. 2014;44:11.

8. Boutron I, Rannou F, Jardinaud-Lopez M, Meric 16. Smink AJ, van den Ende CH, Vliet Vlieland TP,
G, Revel M, Poiraudeau S. Disability and quality Swierstra BA, Kortland JH, Bijlsma JW, et al.
of life of patients with knee or hip "Beating osteoARThritis": development of a

Stepped OA Care |57

 stepped care strategy to optimize utilization 24. Ang D, James G, Stump T. Clinical
and timing of non-surgical treatment appropriateness and not race predicted
modalities for patients with hip or knee referral for joint arthroplasty. Arthritis Rheum.
osteoarthritis. Clin Rheumatol. 2009;61(12):1677-85.
2011;30(12):1623-9.
25. Gossec L, Tubach F, Baron G, al. e. Predictive
17. Smink AJ, Dekker J, Vliet Vlieland TPM, factors of total hip replacement due to primary
Swierstra BA, Kortland JH, Bijlsma JWJ, et al. osteoarthritis: a prospective 2 year study of
Health care use of patients with osteoarthritis 505 patients. Ann Rheum Dis.
of the hip or knee after implementation of a 2005;64(7):1028-32.
stepped care strategy: an observational study.
Arthritis Care Res (Hoboken). 26. Hanchate A, Zhang Y, Felson D, Ash A.
2014;66(6):817-27. Exploring the determinants of racial and ethnic
disparities in total knee arthroplasty: health
18. Teeuw B, Schwarzer R, Jerusalem M. Dutch insurance, income, and assets. Med Care.
adaptation of the general self-efficacy scale. 2008;46(5):481-8.
[Link]
1994. 27. Hawker G, Guan J, Croxford R, Coyte P,
Glazier P, Harvey B, et al. A prospective
19. Kraaimaat F, Evers A. Pain-coping strategies in population-based study of the predictors of
chronic pain patients: psychometric undergoing total joint arthroplasty. Arthritis
characteristics of the pain-coping inventory Rheum. 2006;54(10):3212-20.
(PCI). Int J Behav Med. 2003;10(4):343-63.
28. Judge A, Welton N, Sandhu J, Ben-Shlomo Y.
20. Bellamy N, Buchanan WW, Goldsmith CH, Equity in access to total joint replacement of
Campbell J, Stitt LW. Validation study of the hip and knee in England: cross sectional
WOMAC: a health status instrument for study. BMJ. 2010;11(341):c4092.
measuring clinically important patient relevant
outcomes to antirheumatic drug therapy in 29. Lievense AM, Bierma-Zeinstra SM, Verhagen
patients with osteoarthritis of the hip or knee. AP, Verhaar JA, Koes BW. Prognostic factors
JRheumatol. 1988;15(12):1833-40. of progress of hip osteoarthritis: a systematic
review. Arthritis Rheum. 2002;47(5):556-62.
21. Roorda L, Jones C, Waltz M, Lankhorst G,
Bouter L, van der Eijken J, et al. Satisfactory 30. Mujica Mota R, Tarricone R, Ciani O, Bridges
cross cultural equivalence of the Dutch J, Drummond M. Determinants of demand for
WOMAC in patients with hip osteoarthritis total hip and knee arthroplasty: a systematic
waiting for arthroplasty Ann Rheum Dis. literature review. BMC Health Serv Res.
2004;63:7. 2012;30(12):225.

22. Riemsma R, Taal E, Rasker J, al. e. Evaluation of 31. Steel N, Clark A, Lang I, Wallace R, Melzer D.
a Dutch version of the AIMS2 for patients with Racial disparities in receipt of hip and knee
rheumatoid arthritis. Br J Rheumatol. joint replacements are not explained by need:
1996;35(8):755-+60. the Health and Retirement Study 1998-2004. J
Gerontol A Biol Sci Med Sci. 2008;63(6):629-
23. Andersen R, Newman JF. Societal and 34.
Individual Determinants of Medical Care
Utilization in the United States. Milbank Mem 32. Harrell FE. Regression modeling strategies:
Fund Q Health Soc. 2005;83(4). with applications to linear models, logistic
regression, and survival analysis: Springer-
Verlag New York, Inc; 2001.

58
 of primary care patients with osteoarthritis.
33. Graham J, Olchoswksi A, Gilreath T. How BMC Health Serv Res. 2007;7:169.
many imputations are really needed? Some
practical clarifications of multiple imputation 40. Berger A, Bozic K, Stacey B, Edelsberg J,
theory. Prev Sci. 2007;8(3):206-13. Sadosky A, Oster G. Patterns of
pharmacotherapy and health care utilization
34. Marshall A, Altman DG, Holder RL, Royston P. and costs prior to total hip or total knee
Combining estimates of interest in prognostic replacement in patients with osteoarthritis.
modelling studies after multiple imputation: Arthritis Rheum. 2011;63(8):2268-75.
current practice and guidelines. BMC Med Res
Methodol. 2009;28(9):57. 41. Hawker G, Badley E, Borkhoff C, Croxford R,
Davis A, Dunn S, et al. Which patients are
35. Rubin D. Multiple imputation for nonresponse most likely to benefit from total joint
in surveys. New York: John Wiley & Sons; arthroplasty? Arthritis Rheum. 2013;65:10.
1976.
42. Perruccio A, Power J, Evans H, Mahomed S,
36. Browne W, Subramanian S, Jones K, Goldstein Gandhi R, Mahomed N, et al. Multiple joint
H. Variance partitioning in multilevel logistic involvement in total knee replacement for
models that exhibit overdispersion. J Roy Stat osteoarthritis: Effects on patient-reported
Soc A Sta. 2005;168(3):599-613. outcomes. Arthritis Care Res (Hoboken).
2012;64(6):8.
37. Goldstein H, Browne W, Rasbach J.
Partitioning variation in multilevel models. 43. Nelson A, Smith M, Golightly Y, Jordan J.
Understand Statist. 2002;1(4):223-31. "Generalized osteoarthritis": a systematic
review Semin Arthritis Rheum. 2014
38. van Hassel D, Kenens R. Cijfers uit de Jun;43(6):8.
registratie van huisartsen - peiling 2012.
Utrecht; 2013. 44. O'Donnell C. Variation in GP referral rates:
what can we learn from the literature? Fam
39. Rosemann T, Joos S, Szecsenyi J, Laux G, Pract. 2000;17:10.
Wensing M. Health service utilization patterns

Stepped OA Care |59

 4
One size fits all in physiotherapy
management of knee
osteoarthritis? A cross-sectional,
clinical vignette study

Under review: Physiotherapy

Di-Janne Barten
Jesper Knoop
Ilse Swinkels
Wilfred Peter
Dinny de Bakker
Cindy Veenhof
Joost Dekker
Abstract

Objective To evaluate to what extent primary care provided physiotherapy is

currently tailored to five phenotypes of knee osteoarthritis (OA), namely: the
minimal joint disease’ phenotype, the ‘strong muscle strength’ phenotype, the
‘severe radiographic OA’ phenotype, the ‘obese’ phenotype, and the ‘depressive
mood’ phenotype respectively.

Design A descriptive, cross-sectional survey in the Netherlands.

Setting Primary care physiotherapy practices

Participants 144 Primary care physiotherapists who treated at least one patient
with knee OA over the last month

Main outcome measures Predefined hypotheses regarding physiotherapists’

optimal treatment strategy in five phenotypes of knee OA represented by five
clinical vignettes.

Results Statistically significant differences are found regarding the content and
amount of care between phenotypes of knee OA. These differences were mainly
in line with our predefined hypotheses. Provided care differed from our hypotheses
in the strong muscle phenotype, in which exercise therapy was provided more
often than expected and the referral rate to secondary care was lower than
expected. In the depressive mood phenotype, the referral rate to psychologists was
higher than expected.

Conclusions Physiotherapists seem to tailor their treatment to the five

phenotypes of knee OA, instead of a ‘one size fits all’ approach. Future research is
recommended on evaluating (cost)effectiveness of stratified interventions and
developing a practical screening tool for physiotherapists to facilitate stratification
of patients with knee OA.

Phenotypes of knee OA |63

Introduction

Knee osteoarthritis (OA) is a heterogeneous, musculoskeletal disorder affecting 3.8% of the

world population1. OA is characterized by variations in pathophysiologic etiologies2, varying
severity of clinical symptoms3,4, different courses over time3 and differences in responses to
treatment4. In consequence, OA is increasingly recognized as a syndrome defined by a group
of diseases or phenotypes with a common end point5-7. Stratification of the OA syndrome into
phenotypes will probably facilitate the determination of patients which are the most likely to
benefit from which treatment, inducing more personalized care and more effective and
efficient use of healthcare facilities8,9.

Knoop et al. distinguished five clinically relevant phenotypes in a large cohort of patients with
knee OA10. This finding was replicated in another knee OA cohort11. The identified phenotypes
were based on four clinical characteristics: radiographic severity of OA, muscle strength, body
mass index, and depression. The phenotypes were named: ‘minimal joint disease’ phenotype,
‘strong muscle strength’ phenotype, ‘severe radiographic OA’ phenotype, ‘obese’ phenotype,
and ‘depressive mood’ phenotype10,11.

Current clinical guidelines minimally account for differences in treatment between phenotypes,
but call for tailoring general treatment recommendations to individual patients12-14. To date, it
is unknown to what extent physiotherapists respond to that call and tailor their treatment to
individuals or subgroups with knee OA.

We hypothesize that the minimal joint disease phenotype could benefit from a physiotherapist-
delivered, exercise program with education focusing on self-management as generally
described by clinical guidelines12-14, since this phenotype is not specifically characterized by
prominent clinical features.
The strong muscle phenotype is distinguished by strong quadriceps muscles10,11. This
phenotype mainly concerns men with a history of a knee trauma who currently demonstrate
high levels of physical activity accompanied by biomechanical overload15. Therefore, treatment
in the strong muscle phenotype is hypothesized to focus on providing education rather than
exercise therapy. Educational topics regarding pain coping and preventing overloading will be
of main interest. As non-surgical treatment options are limited, the number of treatment
sessions will be limited and patients will be more often advised to consult secondary care
compared to other phenotypes.
Degeneration of cartilage and bone is the most distinctive characteristic of the severe
radiographic OA phenotype. As exercise therapy has shown to be effective regardless the
severity of OA on imaging techniques16,17, exercise therapy is hypothesized to be included.
Despite the controversial relationship between radiographic severity and the need for surgical
treatment, it is hypothesized that surgery will be often considered in case of non-effective
exercise therapy18-21.

64
Obesity is the most prominent characteristic of the obese phenotype. Therefore, weight
reduction is hypothesized to be the major topic in these patients22. Besides dietary advises by
the physiotherapist, involvement of a dietician will be considered in many cases. In addition,
treatment will contain low impact, (unloading) muscle strengthening exercises23,24 as well as
exercises to improve general aerobic fitness. Patients are advised to be physically active as
much as possible in their daily living.
Finally, the depressive mood phenotype distinguishes itself by the high prevalence of
depression. It is hypothesized that patients will receive a combination of gradually evolved
exercise therapy and an educational program focused at adopting an adequate coping style.
Positive reinforcement is expected to be one of treatments’ key elements. In some cases,
advice for psychological care will be considered as well to treat depressive symptoms25.

The objective of the present study is to evaluate to what extent primary care provided
physiotherapy is currently tailored to the ‘minimal joint disease’ phenotype, the ‘strong muscle
strength’ phenotype, the ‘severe radiographic OA’ phenotype, the ‘obese’ phenotype, and the
‘depressive mood’ phenotype respectively.

Methods

Design
A descriptive, cross-sectional survey was conducted among Dutch primary care
physiotherapists and exercise therapists. Data were gathered between July and October 2016.

Population
Physiotherapists and exercise therapists who regularly treat patients with knee OA were
invited to participate. Two recruitment strategies were used simultaneously. First,
physiotherapists and exercise therapists joining a Dutch ‘Rheumatology Network’26,27 were
invited to participate (June 2016). Digital reminders were sent four weeks after the initial
invitation. Secondly, physiotherapists were invited through the weekly newsletter of the Royal
Dutch Association of Physical Therapy and by calls on social media in October 2016. The calls
comprised a compact study description and hyperlinks to the introduction page of the digital
questionnaire, which provided more detailed information.
Inclusion criterion for participation in the study was involvement in treatment of at least one
patient with knee OA during the last month.

Clinical vignettes
Five visual clinical vignettes representing five previously mentioned phenotypes were included
in the survey questionnaire (Figure 1 & Appendix 1). Clinical vignettes offer significant
advantages in studying clinical behavior of health care professionals, including high feasibility
and low costs28-30.

Phenotypes of knee OA |65

The content of the clinical vignettes was based on a previously described clinical vignette by
Holden et al.31. One researcher (DB) and two clinical experts in the field of OA research (JK
and WP) adapted the original vignette to five phenotype-specific vignettes.

FIGURE 1. Visual presentation of five clinical phenotypes of knee osteoarthritis10,11

mos = months, yrs = years, K&L = Kellgren & Lawrence, BMI = Body Mass Index

Questionnaire
Participating therapists were asked to complete a series of dichotomous and multiple choice
questions to get insight into the content of their treatment in different phenotypes of knee
OA. The questionnaire started with two filter questions to test for eligibility on recent
treatment of knee OA. Included therapists were asked to report their treatment in a patient
with knee OA in general. Subsequently, the visual vignette of a patient who fits in the minimal
joint disease phenotype was introduced and, again, therapists were asked to report their
treatment in this specific patient. The same question was posed simultaneously regarding
another four patients, representing the remaining four phenotypes. As exercise therapy and
education are expected to be widespread used in physiotherapy practice, the content of those
interventions was questioned in more detail. With respect to exercise therapy, participants
were asked for the type of exercise therapy they would perform, for example exercises
focused on aerobic fitness, local muscle strengthening, or improving activities of daily life.
Regarding education, physiotherapists were asked to prioritize five topics which would be
considered during a patient’s treatment, for instance pain management, weight management,
or the importance of pacing physical activities. Furthermore, participants were asked whether
they would consider involvement of other healthcare providers in each of the five phenotypes.

66
Besides the content of physiotherapy treatment, questions focused on the expected number
of treatment sessions and to what extent the included phenotypes cover the total population
of patients with knee OA in their daily practice.

Finally, demographic questions were captured, including gender, age, clinical expertise in the
field of OA, motivation to join an Rheumatology Network and practice site. Appendix 2
comprises a detailed description of the survey questionnaire.

Data analyses
Descriptive statistics were performed to describe the content and amount of physiotherapy
in patients with knee OA. Subsequently, these descriptive characteristics per phenotype were
used to evaluate the predefined hypotheses concerning the content and amount of care per
phenotype (Table 1). A p-value of less than 0.05 was interpreted as statistically significant.
Cases were excluded from analyses in case of missing data regarding >1 phenotype. Statistical
analyses were executed using Stata 14.0 (StataCorp LP, College Station, TX, USA).

TABLE 1. Overview of applied statistics to test predefined hypotheses in five phenotypes of knee osteoarthritis10,11
Phenotype Hypothesis Applied statistic(s)
Minimal joint Exercise program and education focusing Descriptive statistics,
disease of self-management Spearman’s rank correlation coefficient
Focus of treatment on providing Descriptive statistics,
education rather than exercise therapy Spearman’s rank correlation coefficient
Pain coping and preventing overloading Descriptive statistics
are main issues in education
Strong muscle
High referral rate* to secondary care Descriptive statistics, Kruskall Wallis for equality of
compared to other phenotypes proportions and a Bonferroni post-hoc analysis (2-sided)
Limited number of treatment sessions Descriptive statistics, Kruskall Wallis for equality of
proportions and a Bonferroni post-hoc analysis (2-sided)
Application of exercise therapy regardless Descriptive statistics, Kruskall Wallis for equality of
Severe the severity of OA proportions and a Bonferroni post-hoc analysis (2-sided)
radiographic OA High referral rate to secondary care Descriptive statistics, Kruskall Wallis for equality of
proportions and a Bonferroni post-hoc analysis (2-sided)
Inclusion of low impact, (unloading) Descriptive statistics
muscle strengthening exercises as well as
exercises to improve general aerobic
fitness.
Obese
Importance of advice to be physically Descriptive statistics
active as much as possible in daily living
Involvement of a dietician to achieve Descriptive statistics, Kruskall Wallis for equality of
weight reduction proportions and a Bonferroni post-hoc analysis (2-sided)
Inclusion of a combination of gradually Descriptive statistics, Spearman’s rank correlation
evolved exercise therapy and an coefficient
educational program focused at adopting
Depressive mood
an adequate coping style.
In some cases, advice for psychological Descriptive statistics, Kruskall Wallis for equality of
care proportions and a Bonferroni post-hoc analysis (2-sided)

* A referral by a physiotherapist comprises an advice to the patient and/or his general practitioner to consider involvement
of another healthcare provider

Phenotypes of knee OA |67

Results

In total, 225 physiotherapists accessed the digital questionnaire. Seven physiotherapists were
excluded since they did not met the inclusion criterion. Demographics of these seven
physiotherapists did not significantly differ from the eligible therapists. Subsequently, due to
missing data, 74 cases were excluded from analyses. The remaining 144 physiotherapists were
included for analyses. Fifty-six percent of these respondents were derived via Rheumatology
Networks, 40% via the digital newsflash, 4% via remaining routes. Except for age, recruitment
groups were comparable regarding demographics.

Population
Table 2 shows demographic data of participating physiotherapists. Forty-four percent of the
respondents were male and 74% of the therapists were in the 31-60 years of age category,
which was comparable to the general Dutch population of physiotherapists in primary care 32.
Seventy percent of our population were highly experienced therapists (≥15 years of
experience). Ten percent of the population reported to be an expert in de field of OA.

TABLE 2. Demographics of participating physiotherapists treating patients with knee osteoarthritis (n=144)
Characteristic Number percentage Missing values (number)
Gender (male) 36 (44.4) 63
Age category 64
<30 years 10 (12.5)
31-60 years 59 (73.7)
>60 years 11 (13.8)
Years of experience 63
≤5 years 6 (7.4)
6-15 years 18 (22.2)
>15 years 57 (70.4)
Self-reported expertise in clinical knee OA 66
(Less than) average 38 (48.7)
More than average 32 (41.0)
Expert in the field of OA 8 (10.3)
Member of Arthritis Association (yes) 64 (63.4) 43
Practice site (urban) 55 (69.6) 65

Physiotherapy per phenotype of knee OA

Participating physiotherapists endorse the proposed phenotypes in their clinical practice.
Overall, 21% of their population represents the ‘minimal joint disease’ phenotype, 15% the
‘strong muscle’ phenotype, 23% the ‘severe radiographic OA’ phenotype, 25% the ‘obese’
phenotype, and 15% the ‘depressive mood’ phenotype. Less than 2% of the patients with knee
OA do not fit to any of the phenotypes.

68
Figure 2 shows in what percentage of each phenotype, physiotherapists would consider a
specific intervention. Education and therapeutic exercises were frequently mentioned in all
phenotypes. Statistically significant differences (p≤.05) were found regarding the application of
most of the interventions between phenotypes, except for education, providing braces, and
massage (Figure 2).

Education

Therapeutic exercise *

Improving self-management *

Manual therapy *
Minimal joint disease
(Medical) taping *
Strong muscle
Providing braces
Severe radiographic OA
Massage
Obese
Hydrotherapy *
Depressive mood
TENS *

Remaining interventions §*

0 10 20 30 40 50 60 70 80 90 100
%

FIGURE 2. Applied interventions in patients with clinical knee osteoarthritis (OA) per phenotype
§ Remaining interventions includes electrotherapy, lasertherapy, and heat or ice
* Statistically significant difference between phenotypes (p<0.01)

Table 3 summarizes per phenotype the top-5 therapeutic exercise modalities, the top-5 topics
included in patient education, and the top-5 referrals as well as the median number of
considered treatment sessions per phenotype. Subsequently, these characteristics are used to
evaluate the predefined hypotheses concerning the content and amount of care per phenotype
of knee OA (Table 4).

Minimal joint disease phenotype

In accordance with our hypothesis, a combination of exercise therapy and education was
considered in the majority of the minimal joint disease phenotype (85%) (Table 4). Respecting
the content of exercise therapy, the frequent inclusion of high-loaded muscle strengthening
exercises stands out. The minimal joint disease phenotype would receive less treatment
sessions (median of 1-5 sessions) compared to the severe radiographic OA phenotype, the
obese phenotype, and the depressive mood phenotype, (p<0.01) (Table 3).

Phenotypes of knee OA |69

Strong muscle phenotype
In contrast to our expectations, exercise therapy seems to play an important role in the strong
muscle phenotype. Sixty percent of the physiotherapists would use a combination of exercise
therapy and education, compared to education only in one third of the population (Table 4).
The content of patient education was in accordance with our hypotheses and focused on
pacing physical activities and improving pain coping strategies. The referral rate to secondary
was lower than expected; 8% of the considered referrals concerned a referral to an orthopedic
surgeon, compared to 56% in the severe radiographic OA phenotype (p=<0.01) and 24% in
the obese phenotype (p=0.04). As hypothesized, physiotherapists would use significantly less
treatment sessions in the strong muscle phenotype compared to the severe radiographic OA
phenotype, the obese phenotype, and the depressive mood phenotype (Tables 3 and 4).

Severe radiographic OA phenotype

Eighty-nine percent of the physiotherapists would consider the use of exercise therapy in the
severe radiographic OA phenotype. This application rate did not significantly differ from the
remaining phenotypes, except for the strong muscle phenotype (61%, p<0.01) (Table 4).
Furthermore, the hypothesized high referral rate to secondary care was confirmed (56%)
(p<0.01) (Tables 3 and 4).

Obese phenotype
As hypothesized, unloaded muscle strengthening exercises and aerobic exercises were
represented in the top5 of applied exercise therapy modalities and ‘increasing physical activity
level’ was the second most mentioned topic in patient education. Furthermore, our
expectation concerning involvement of a dietician was confirmed. Eighty-two percent of the
obese phenotype would be advised to visit a dietician, which was significantly higher compared
to the remaining phenotypes (Tables 3 and 4).

Depressive mood phenotype

Gradually evolved exercise therapy was the most considered therapeutic exercise modality in
the depressive mood phenotype. In accordance with our hypothesis, this ‘graded activity’
would be combined with pain coping strategies in a considerable part of the population (54%).
A priori, we expected that psychological care would be considered in ‘some’ cases. However,
participating physiotherapists reported to consider a referral to a psychologist in 56% of the
depressive mood phenotype (Table 4).

70
 TABLE 3. Overview of the content and amount of care in five phenotypes of knee OA
Phenotype Minimal joint disease Strong muscle Severe radiographic OA Obese Depressive mood
Top-5 therapeutic Home-exercises 82 Home-exercises 69 Home-exercises 75 Home-exercises 67 Graded activity * 69
Exercise therapy

exercises (%) HL muscle strength * 76 Coordinating exercises * 60 Functional exercises 61 UL muscle strength * 57 Home-exercises 64
Functional exercises 66 Functional exercises 53 Coordinating exercises * 56 Graded activity * 55 UL muscle strength * 58
(n=108)

Coordinating exercises * 50 HL muscle strength * 36 UL muscle strength * 51 Functional exercises 54 Functional exercises 56
Mobilising exercises * 34 Graded activity * 31 HL muscle strength * 42 Aerobic exercises * 51 Aerobic exercises * 46
Consequences of OA 55 Importance of rest * 57 Treatment options * 50 Weight reduction * 62 Increasing PA level * 57
Top-5 education

Onset of OA 55 Consequences of OA 52 Consequences of OA 49 Increasing PA level * 54 Psychological RF decline * 51

Treatment options * 49 Onset of OA 52 Onset of OA 43 Contact a dietician * 51 Onset of OA 40
topics (%)
Education
(n=108)

Pacing PA * 35 Pacing PA * 45 Pacing PA * 36 Physical RF for decline 39 Consequences of OA 38

Physical RF for decline 20 Decreasing PA level * 36 Physical RF for decline 29 Onset of OA 36 Pain medication 26
GP * 42 GP * 34 Orthopedic surgeon * 56 Dietician * 82 GP * 71
Top-5 referrals (%)

Local sports provider * 27 Local sports provider * 33 GP * 48 GP * 69 Psychologist * 56

Podiatrist 14 Podiatrist 12 Dietician * 22 Orthopedic surgeon * 24 GP nurse mental care * 42
Referral†

Orthopedic surgeon * 6 Orthopedic surgeon * 8 Pharmacologist * 12 Local sports provider * 18 Dietician * 36

(n=85)

Pharmacologist * 4 Psychologist * 2 Local sports provider * 9 Occupational therapist * 16 Multidisciplinar team * 24

Amount of care (%) *

Intake only 12 Intake only 11 Intake only 1 Intake only 1 Intake only 1
1-5 treatment sessions 55 1-5 treatment sessions 67 1-5 treatment sessions 17 1-5 treatment sessions 1 1-5 treatment sessions 1
6-10 treatment sessions 27 6-10 treatment sessions 17 6-10 treatment sessions 40 6-10 treatment sessions 17 6-10 treatment sessions 17
11-15 treatment sessions 6 11-15 treatment sessions 5 11-15 treatment sessions 32 11-15 treatment sessions 31 11-15 treatment sessions 31
16-20 treatment sessions 0 16-20 treatment sessions 0 16-20 treatment sessions 9 16-20 treatment sessions 36 16-20 treatment sessions 33
>20 treatment sessions 0 >20 treatment sessions 0 >20 treatment sessions 1 >20 treatment sessions 15 >20 treatment sessions 17

OA = Osteoarthritis; HL = high loaded; UL = unloaded; PA = physical activity; RF = risk factors; * p≤0.01 between phenotypes;
†
A referral by a physiotherapist comprises an advice to the patient and/or his general practitioner to consider involvement of another healthcare provider
 TABLE 4. Evaluation of hypotheses concerning the content and amount of care in phenotypes of knee OA

Phenotype Hypothesis Identified data Evaluation hypothesis

Minimal joint Exercise program and Application of exercise AND education: 84.7%
disease education focusing of self- Application of exercise OR education: 9.7%
management Application of exercise NOR education: 5.6% +
Spearman’s ρ .52
Strong muscle Focus of treatment on providing Application of education WITHOUT 32.6%
education rather than exercise exercise: 59.7%
therapy Application of education AND exercise: 1.4%
-
Application of exercise WITHOUT education: 6.3%
Application of education NOR exercise: .26
Spearman’s ρ
Pain coping and preventing Importance of rest, pacing activities, and decreasing physical
overloading are main issues in activity level are all top-5 education topics +
education
High referral rate† to secondary Referral rate comprised 8.2%, which was significantly lower
care compared to other than the severe radiographic OA phenotype (56.5%; p<.01) -
phenotypes and the obese phenotype (23.5%; p=0.04)
Limited number of treatment The median number of treatment sessions was 1-5 sessions,
sessions which was significantly lower than the remaining phenotypes +
except for the minimal joint disease phenotype.
Severe Application of exercise therapy Application in severe radiographic OA: 88.9%
radiographic regardless the severity of OA Application in minimal joint disease: 93.1% (p=.27)
disease Application in strong muscle strength: 61.1% (p<.01) +
Application in obese: 89.6% (p=.46)
Application in depressive mood: 88.2% (p=.46)
High referral rate to secondary Referral rate comprised 56.5%, which was significantly higher
care than the remaining phenotypes +

Obese Inclusion of low impact, Both local, unloaded muscle strengthening exercises as well as
(unloading) muscle aerobic exercises were represented in the top-5 most
strengthening exercises as well performed exercise therapies. +
as exercises to improve general
aerobic fitness.
Importance of advice to be Increasing physical activity level was the second most
physically active as much as mentioned topic in patient education +
possible in daily living
Involvement of a dietician to Referral rate comprised 82.4%, which was significantly higher
+
achieve weight reduction than the remaining phenotypes
Depressive Inclusion of a combination of Application of graded activity AND coping: 54.4%
mood gradually evolved exercise Application of graded activity OR coping: 17.8%
therapy and an educational Application of graded activity NOR coping: 27.8% +
program focused at adopting an Spearman’s ρ .62
adequate coping style.
In some cases, advice for Referral rate comprised 56.5%, which was significantly higher
+/
psychological care will be than the remaining phenotypes
-
considered

72
Discussion

Elaborating on the identification of five clinical phenotypes of knee OA10,11, this study aimed to
evaluate to what extent currently provided physiotherapy is tailored to five phenotypes:
‘minimal joint disease’, ‘strong muscle strength’, ‘severe radiographic OA’, ‘obese’, and
‘depressive mood’ respectively. Based on our results, primary care physiotherapists seem to
tailor their treatment to each of the five phenotypes, instead of offering an ‘one size fits all’
treatment. Our hypotheses concerning the content of physiotherapy and the number of
treatment sessions in each specific phenotype were mainly confirmed.

Evaluation of predefined hypotheses

Considering the minimal joint disease phenotype, physiotherapists would frequently use a
combination of exercise therapy and education, implying a good fit to clinical guidelines12-14.
In the strong muscle phenotype, we did not expect a frequent use of high-loaded muscle
strength exercises, since high muscle strength is the most prominent characteristic. Maybe,
physiotherapists are simply used to include muscle strength exercises in their treatment rather
than considering its’ added value in each individual patient. Instead of therapeutic exercises,
biomechanical interventions, like knee braces and wedges soles could be beneficial in
preventing high local stress in the degenerated joint compartment2,33. Subsequently, in case of
insufficient relieve of complaints despite providing appropriate non-surgical treatment, total
joint replacement has shown to be an effective final treatment option34,35. However, in contrast
to our expectations, a minority of the physiotherapists would account for surgical treatment
as the referral rate to secondary care amounted only 8%.
A much higher referral rate was found in the severe radiographic OA phenotype (56%), which
was in accordance with our predefined hypothesis. As the relationship between radiographic
severity and the indication for surgery is controversial20,21, therapists could be recommended
to focus on a patient’s functioning rather than solely on radiographic severity.
In the obese phenotype, the considered content of physiotherapy mainly comprised weight
management which was in line with our expectations. Although aerobic exercises were
mentioned by 51% of the therapists, an even higher appliance rate could have been expected
as this intervention specifically addresses the metabolic syndrome in patients with obesity36.
On the other hand, Christensen et al. suggested dietary guidance to be preferable over
exercise therapy in patients with obesity and knee OA, since the maintaining effect on weight
is higher in dietician-supported patients than patients who exclusively received exercise
therapy24.
In the depressive mood phenotype, sixty percent of the physiotherapists would consider a
referral to a psychologist, which was higher than we expected prior to the study. Besides a
referral for psychological care, physiotherapists also focus on behavioral aspects themselves,
for example by applying behavioral graded activity, improving a patient’s coping style, and
addressing psychological risk factors for functional decline. This approach supports the
fundamental role of central sensitization in the depressive mood phenotype and confirms that

Phenotypes of knee OA |73

knee OA itself not sufficiently explains the complex symptomatology in this subgroup of knee
OA33.

Methodological considerations
First, the high number of missing values should be mentioned. Most missing values were
noticed when questions involved five phenotypes simultaneously. Maybe, these questions were
too complex. Second, we did not ask for a preferred sequence of eligible interventions. It was
previously suggested to apply non-surgical interventions prior to a referral to secondary care37.
A strength of this study, comprised the use of visual and colored clinical vignettes. Visual
communication takes less time to interpret, colored visuals increase willingness to read by
80%, and visual information is supposed to be better remembered38. Although vignettes are
feasible to use and are accompanied with low costs28-30, clinical vignettes may merely reflect
clinicians competences rather than actual clinical practice39. Fourth, the content of our
vignettes were restricted to the phenotypes identified by Knoop et al and replicated by van
der Esch et al.10,11. Although less than 2% of the participating physiotherapists suggested
additional subgroups of knee OA, a recent review has identified 79 different theoretical
phenotypes of knee OA33, subdivided into six main groups. Four of our phenotypes represents
one of those main groups, except for the severe radiographic OA phenotype. Finally, the mode
of stratification should be criticized. In this study, we focused on stratification based on
phenotypes of knee OA. However, the question arises to what extent physiotherapists are
familiar with thinking in terms of phenotypes. They may achieve personalized care based on
other outcomes.

Implications
Primary care physiotherapists seem to tailor their treatment to specific phenotypes of knee
OA, rather than using an ‘one size fits all’ approach. The current study provides a first
indication of which interventions are considered per phenotype; development and systematic
evaluation of more detailed, stratified interventions are recommended both on clinical
effectiveness as well as cost-effectiveness. This may result in phenotype specific
recommendations in general clinical guidelines for knee OA.
To facilitate physiotherapists in tailoring their treatment to specific phenotypes, development
of a triage screening tool could be valuable. In patients with low back pain, the STarT Back
Screening Tool supports clinicians in stratifying patients into subgroups and, in consequence,
optimizes treatment and treatment outcome (8,9,40). This stratification could be based on
phenotypes, but also on other factors, like a patient’s motivation, education level, or previous
experiences.
A final remark concerns the rapid development of eHealth. Although the content of
physiotherapy treatment will not necessarily change when using eHealth, its’ implementation
could have major consequences for the organization of a patient’s treatment. Especially
patients who are eligible for short term, low advanced therapy (i.e. minimal joint phenotype,
strong muscle phenotype) could benefit from (partly) substitution of face-to-face care to online

74
care. (Cost)effectiveness of this blended approach should be examined more thoroughly prior
the implementation in clinical OA care41,42.

In conclusion, based on clinical vignettes representing five different phenotypes of knee OA,
physiotherapists seem to tailor their treatment to each of the five phenotypes, instead of an
‘one size fits all’ treatment. Identified differences in treatment are mainly in accordance with
our predefined hypotheses. Future research is recommended on evaluating (cost)effectiveness
of stratified interventions and developing a practical screening tool to facilitate stratification of
patients with knee OA.

Phenotypes of knee OA |75

References osteoarthritis progression who will respond to
treatment? A focus on epidemiology and
phenotype of osteoarthritis. Drugs Aging 2015
1. Cross M, Smith E, Hoy D, Nolte S, Ackerman Mar;32(3):179-187.
I, Fransen M, et al. The global burden of hip and
knee osteoarthritis: estimates from the global 9. Karsdal MA, Christiansen C, Ladel C,
burden of disease 2010 study. Ann Rheum Dis Henriksen K, Kraus VB, Bay-Jensen AC.
2014 Jul;73(7):1323-1330. Osteoarthritis--a case for personalized health
care? Osteoarthritis Cartilage 2014
2. Felson DT. Identifying different osteoarthritis Jan;22(1):7-16.
phenotypes through epidemiology.
Osteoarthritis Cartilage 2010 May;18(5):601- 10. Knoop J, van der Leeden M, Thorstensson CA,
604. Roorda LD, Lems WF, Knol DL, et al.
Identification of phenotypes with different
3. Bijlsma JW, Berenbaum F, Lafeber FP. clinical outcomes in knee osteoarthritis: data
Osteoarthritis: an update with relevance for from the Osteoarthritis Initiative. Arthritis
clinical practice. Lancet 2011 Jun Care Res (Hoboken) 2011 Nov;63(11):1535-
18;377(9783):2115-2126. 1542.

4. Driban JB, Sitler MR, Barbe MF, 11. van der Esch M, Knoop J, van der Leeden M,
Balasubramanian E. Is osteoarthritis a Roorda LD, Lems WF, Knol DL, et al. Clinical
heterogeneous disease that can be stratified phenotypes in patients with knee
into subsets? Clin Rheumatol 2010 osteoarthritis: a study in the Amsterdam
Feb;29(2):123-131. osteoarthritis cohort. Osteoarthritis Cartilage
2015 Apr;23(4):544-549.
5. Altman R, Asch E, Bloch D, Bole G, Borenstein
D, Brandt K, et al. Development of criteria for 12. McAlindon TE, Bannuru RR, Sullivan MC,
the classification and reporting of Arden NK, Berenbaum F, Bierma-Zeinstra SM,
osteoarthritis. Classification of osteoarthritis et al. OARSI guidelines for the non-surgical
of the knee. Diagnostic and Therapeutic management of knee osteoarthritis.
Criteria Committee of the American Osteoarthritis Cartilage 2014 Mar;22(3):363-
Rheumatism Association. Arthritis Rheum 388.
1986 Aug;29(8):1039-1049.
13. Fernandes L, Hagen KB, Bijlsma JW,
6. Martel-Pelletier J, Lajeunesse D, Fahmi H, Andreassen O, Christensen P, Conaghan PG,
Tardif G, Pelletier JP. New thoughts on the et al. EULAR recommendations for the non-
pathophysiology of osteoarthritis: one more pharmacological core management of hip and
step toward new therapeutic targets. Curr knee osteoarthritis. Ann Rheum Dis 2013
Rheumatol Rep 2006 Feb;8(1):30-36. Jul;72(7):1125-1135.

7. Herrero-Beaumont G, Roman-Blas JA, 14. Hochberg MC, Altman RD, April KT, Benkhalti
Castaneda S, Jimenez SA. Primary M, Guyatt G, McGowan J, et al. American
osteoarthritis no longer primary: three subsets College of Rheumatology 2012
with distinct etiological, clinical, and recommendations for the use of
therapeutic characteristics. Semin Arthritis nonpharmacologic and pharmacologic
Rheum 2009 Oct;39(2):71-80. therapies in osteoarthritis of the hand, hip, and
knee. Arthritis Care Res (Hoboken) 2012
8. Bruyere O, Cooper C, Arden N, Branco J, Apr;64(4):465-474.
Brandi ML, Herrero-Beaumont G, et al. Can
we identify patients with high risk of

76
15. Waarsing JH, Bierma-Zeinstra SM, Weinans H. inflammation, and clinical outcomes among
Distinct subtypes of knee osteoarthritis: data overweight and obese adults with knee
from the Osteoarthritis Initiative. osteoarthritis: the IDEA randomized clinical
Rheumatology (Oxford) 2015 Sep;54(9):1650- trial. JAMA 2013 Sep 25;310(12):1263-1273.
1658.
23. Christensen R, Bartels EM, Astrup A, Bliddal
16. Skou ST, Derosche CA, Andersen MM, H. Effect of weight reduction in obese patients
Rathleff MS, Simonsen O. Nonoperative diagnosed with knee osteoarthritis: a
treatment improves pain irrespective of systematic review and meta-analysis. Ann
radiographic severity. A cohort study of 1,414 Rheum Dis 2007 Apr;66(4):433-439.
patients with knee osteoarthritis. Acta Orthop
2015;86(5):599-604. 24. Christensen R, Henriksen M, Leeds AR,
Gudbergsen H, Christensen P, Sorensen TJ, et
17. Knoop J, Dekker J, van der Leeden M, van der al. Effect of weight maintenance on symptoms
Esch M, Klein JP, Hunter DJ, et al. Is the of knee osteoarthritis in obese patients: a
severity of knee osteoarthritis on magnetic twelve-month randomized controlled trial.
resonance imaging associated with outcome of Arthritis Care Res (Hoboken) 2015
exercise therapy? Arthritis Care Res May;67(5):640-650.
(Hoboken) 2014 Jan;66(1):63-68.
25. Yohannes AM, Caton S. Management of
18. Riddle DL, Jiranek WA. Knee osteoarthritis depression in older people with osteoarthritis:
radiographic progression and associations with A systematic review. Aging Ment Health 2010
pain and function prior to knee arthroplasty: a Aug;14(6):637-651.
multicenter comparative cohort study.
Osteoarthritis Cartilage 2015 Mar;23(3):391- 26. Li LC, Hurkmans EJ, Sayre EC, Vliet Vlieland
396. TP. Continuing professional development is
associated with increasing physical therapists'
19. Verra WC, Witteveen KQ, Maier AB, roles in arthritis management in Canada and
Gademan MG, van der Linden HM, Nelissen the Netherlands. Phys Ther 2010
RG. The reason why orthopaedic surgeons Apr;90(4):629-642.
perform total knee replacement: results of a
randomised study using case vignettes. Knee 27. Peter WF, van der Wees PJ, Hendriks EJ, de
Surg Sports Traumatol Arthrosc 2016 Bie RA, Verhoef J, de Jong Z, et al. Quality
Aug;24(8):2697-2703. indicators for physiotherapy care in hip and
knee osteoarthritis: development and
20. Bedson J, Croft PR. The discordance between clinimetric properties. Musculoskeletal Care
clinical and radiographic knee osteoarthritis: a 2013 Dec;11(4):193-202.
systematic search and summary of the
literature. BMC Musculoskelet Disord 2008 28. Mohan D, Fischhoff B, Farris C, Switzer GE,
Sep 2;9:116-2474-9-116. Rosengart MR, Yealy DM, et al. Validating a
vignette-based instrument to study physician
21. Maillefert JF, Roy C, Cadet C, Nizard R, Berdah decision making in trauma triage. Med Decis
L, Ravaud P. Factors influencing surgeons' Making 2014 Feb;34(2):242-252.
decisions in the indication for total joint
replacement in hip osteoarthritis in real life. 29. Peabody JW, Luck J, Glassman P, Jain S, Hansen
Arthritis Rheum 2008 Feb 15;59(2):255-262. J, Spell M, et al. Measuring the quality of
physician practice by using clinical vignettes: a
22. Messier SP, Mihalko SL, Legault C, Miller GD, prospective validation study. Ann Intern Med
Nicklas BJ, DeVita P, et al. Effects of intensive 2004 Nov 16;141(10):771-780.
diet and exercise on knee joint loads,

Phenotypes of knee OA |77

30. Rutten GM, Harting J, Rutten ST, Bekkering 37. Smink AJ, van den Ende CH, Vliet Vlieland TP,
GE, Kremers SP. Measuring physiotherapists' Swierstra BA, Kortland JH, Bijlsma JW, et al.
guideline adherence by means of clinical "Beating osteoARThritis": development of a
vignettes: a validation study. J Eval Clin Pract stepped care strategy to optimize utilization
2006 Oct;12(5):491-500. and timing of non-surgical treatment
modalities for patients with hip or knee
31. Holden MA, Nicholls EE, Hay EM, Foster NE. osteoarthritis. Clin Rheumatol 2011
Physical therapists' use of therapeutic exercise Dec;30(12):1623-1629.
for patients with clinical knee osteoarthritis in
the United kingdom: in line with current 38. McCabe DP, Castel AD. Seeing is believing: the
recommendations? Phys Ther 2008 effect of brain images on judgments of scientific
Oct;88(10):1109-1121. reasoning. Cognition 2008 Apr;107(1):343-
352.
32. van Hassel D, Kenens R. 2012; Available at:
[Link] 39. Shah R, Edgar DF, Evans BJ. A comparison of
n/Cijfers-uit-de-registratie-van- standardised patients, record abstraction and
[Link]. clinical vignettes for the purpose of measuring
Accessed February, 20th, 2017. clinical practice. Ophthalmic Physiol Opt 2010
May;30(3):209-224.
33. Dell'Isola A, Allan R, Smith SL, Marreiros SS,
Steultjens M. Identification of clinical 40. Hill JC, Dunn KM, Lewis M, Mullis R, Main CJ,
phenotypes in knee osteoarthritis: a systematic Foster NE, et al. A primary care back pain
review of the literature. BMC Musculoskelet screening tool: identifying patient subgroups
Disord 2016 Oct 12;17(1):425. for initial treatment. Arthritis Rheum 2008
May 15;59(5):632-641.
34. Chang RW, Pellisier JM, Hazen GB. A cost-
effectiveness analysis of total hip arthroplasty 41. Bossen D, Kloek C, Snippe HW, Dekker J, de
for osteoarthritis of the hip. JAMA 1996 Mar Bakker D, Veenhof C. A Blended Intervention
20;275(11):858-865. for Patients With Knee and Hip Osteoarthritis
in the Physical Therapy Practice: Development
35. Hawker GA, Badley EM, Croxford R, Coyte and a Pilot Study. JMIR Res Protoc 2016 Feb
PC, Glazier RH, Guan J, et al. A population- 24;5(1):e32.
based nested case-control study of the costs of
hip and knee replacement surgery. Med Care 42. Kloek CJ, Bossen D, Veenhof C, van Dongen
2009 Jul;47(7):732-741. JM, Dekker J, de Bakker DH. Effectiveness and
cost-effectiveness of a blended exercise
36. Lee S, Kim TN, Kim SH, Kim YG, Lee CK, intervention for patients with hip and/or knee
Moon HB, et al. Obesity, metabolic osteoarthritis: study protocol of a randomized
abnormality, and knee osteoarthritis: a cross- controlled trial. BMC Musculoskelet Disord
sectional study in Korean women. Mod 2014 Aug 8;15:269-2474-15-269.
Rheumatol 2015 Mar;25(2):292-297.

78
Appendix 1: Extended description of five phenotypes of knee osteoarthritis (Knoop et al. (2011), van der Esch et al. (2015))
Ann represents the ‘minimal joint disease phenotype’, Bert represents the ‘strong muscle phenotype’, Corine the ‘severe radiographic osteoarthritis phenotype’,
Dory the ‘obese phenotype’, and Erica the ‘depressive mood phenotype’
Appendix 2:
Survey questionnaire ‘Physiotherapy in knee osteoarthritis: one size fits all?’

1. How many patients suffering from osteoarthritis do you treat monthly?

Only by exception / 1 or 2 patients / 3 to 5 patients / 6 to 10 patients / >10 patients

2. What percentage of the total population with osteoarthritis particularly involves knee
osteoarthritis?
0% / About 25% / About 50% / About 75% / 100%

3. Which interventions do you generally use in patients with osteoarthritis of the knee?
(Please mark a maximum of five interventions)
Electrotherapy / Hydrotherapy / Assessing braces / Lasertherapy / Massage / (Medical)
taping / Therapeutic exercise / Manual therapy / TENS / Heat or ice / Ultrasound /
Education regarding effective self-management / Other (please specify)

Suppose that Ann visits your practice. Results of the anamnesis and physical examination
were presented in the figure below.

4. Ann would like to enjoy gardening again, without a continuous painful knee. Suppose,
you decide to start a physiotherapy treatment. Which interventions would you
include? (Please mark a maximum of five interventions)
Electrotherapy / Hydrotherapy / Assessing braces / Lasertherapy / Massage / (Medical)
taping / Therapeutic exercise / Manual therapy / TENS / Heat or ice / Ultrasound /
Education regarding effective self-management / Other (please specify)

In addition to Ann, another four patients with knee OA request for your help: Bert, Corine,
Dory and Erica. Compared to Ann, those patients were of similar age, were all referred for
their GP, visit a physiotherapist for the first time, are former shopkeepers and enjoy
gardening. However, they also differ from each other. Next questions are about
consequences of these differences regarding physiotherapy treatment of Ann, Bert, Corine,
Dory and Erica.

5. Which interventions would you include in each patient’s treatment? (Please mark a
maximum of five interventions for each patient)
Electrotherapy / Hydrotherapy / Assessing braces / Lasertherapy / Massage / (Medical)
taping / Therapeutic exercise / Manual therapy / TENS / Heat or ice / Ultrasound /
Education regarding effective self-management / Other (please specify)

6. If you decide to perform therapeutic exercises, which kind of exercise would you
include in each patient’s treatment? (Please prioritize your five preferred
interventions for each patient)
Home-exercises / Local high-loaded muscle strengthening exercises / Local unloaded muscle
strengthening exercises / Functional therapeutic exercises / Aerobic exercises / Coordinating
exercises / Mobilising exercises / Gradually evolved exercises (graded activity) / Hydrotherapy
/ Other (please specify)

7. If you decided to perform therapeutic exercises, how would you deliver your
intervention? (Please mark which manner(s) fit(s) to each of the five phenotypes)

80
 Individually under close supervision / Individually including performing exercises
independently / By group treatment / By a consult to teach home-exercises / By written
home-exercises / Under supervision of a fitness coach / By referral to a local sports provider
/ Other (please specify)

8. If you decided to include education, which topics would be covered? (Please mark
which topics would be covered in each of the five phenotypes)
Onset of OA / Consequences of OA / Physical risk factors for functional decline / Psychoogical
risk factors for functional decline / Socio-demographic risk factors for functional decline /
Treatment options in OA / Advises regarding use of braces / Advises regarding use of walking
aids / Advises regarding pain medication / Advises regarding food / Advise to contact a
dietician / Advise to contact the GP / Importance of weight reduction / Importance of rest /
Importance of increasing physical activity level / Pacing physical activities / Preventing painful
activities / Decreasing physical activity level / Other (please specify)

9. How many times would you be likely to see each of the five patients? (Please check
the most appropriate box)
Just an intake / 1-5 sessions / 6-10 sessions / 11-15 sessions / 16-20 sessions / 21-50
sessions / >50 sessions

10. To what extent influences a patient’s reimbursement your choice regarding the
number of delivered treatment sessions?
Not at all / To a certain extent / Substantially / Completely

11. Which timeframe fits the best to each patient’s treatment from the first to the last
visit?
<1 month / 1-2 months / 3-4 months / 5-6 months / 6-12 months / >12 months

12. To whom would you be likely to refer each of the five patients? (Please check all
that apply)
GP / Orthopedic surgeon / Rheumatologist / Psychologist / Multidisciplinary pain team /
Pharmacologist / Occupational therapist / Dietician / Podiatrist / Primary care assistant
practitioner mental health care / Local sports provider / Acupuncturist

This questionnaire describes five fictional patients with knee OA who could be presented in
primary care physiotherapy. These patients represents different phenotypes of knee OA

13. In case of ten patients with knee OA would present themselves in your practice, how
many of those patients would fit to the description of Ann, Bert, Corine, Dory and
Erica. (Divide the sum of ten over the five patients, including a remaining category for
patients who do not fit to any of the described phenotypes)

14. If you have indicated that a part of your population of patients with knee OA does
not fit to any of the five phenotypes, could you please describe the group(s) of
patients you are likely to see in your practice?

Finally, we would like to gather some demographic data.

15. I am a
Male / female

Phenotypes of knee OA |81

 16. My age is:
<30 years / 30-40 years / 41-50 years / 51-60 years / >60 years

17. I operate as a:
Physiotherapist / Exercise therapist / Other (please specify)

18. Regarding this profession, my year of graduation was:

<1970 / 1970-1980 / 1981-1990 / 1991-2000 / 2001-2010 / >2010

19. Compared to an average therapist, my level of expertise in OA is

Lower / Similar / Higher / I am an expert in the field of OA

20. My two most important motivations to join an Arthritis Association are:

Interest in the target group / Possibility of exchanging general knowledge regarding
treatment of patients with OA / Possibility to discuss patient cases / Possibility to meet fellow
therapists / Possibility to recruit new patients / Possibility to improve the practice’s bargaining
position with healthcare insurers / Possibility to position your practice against nearby
practices

21. The practice site could be best described by:

Rural / Suburban / Urban

82
 Patient-specific self-assessment
5
instruments for measuring
physical function: a systematic
review of measurement properties

Published in: J Clin Epidemiol. 2012 Jun;65(6):590-601

[Link]

Di-Janne Barten
Martijn Pisters
Palesa Huisman
Tim Takken
Cindy Veenhof
Abstract

Objective To identify patient-specific self-assessment instruments which measure

physical function in patients with musculoskeletal disorders and to evaluate the
descriptive properties as well as the psychometric qualities of these instruments.

Study Design and Setting After a systematic search, included instruments were
evaluated psychometrically by the checklist ‘quality criteria for measurement
properties of health status instruments’

Results Twenty-three studies were included, referring to twelve instruments.

Nine different versions of the Patient Specific Functional Scale (PSFS) were
identified. The practical elaboration of the different versions of the PSFS varied
widely. None of the instruments were tested on all psychometric quality criteria of
the checklist. The PSFS described by Cleland et al. was most extensively investigated
and obtained exclusively positive scores. Overall, construct validity, reliability and
responsiveness were evaluated most frequently.

Conclusion The descriptive properties and psychometric quality of patient-

specific instruments measuring physical function, are only partly investigated. The
PSFS was the most investigated instrument: nine different versions have been
evaluated psychometrically. The version of Cleland et al. was most extensively
investigated, obtained exclusively positive scores following the quality criteria by
Terwee et al. and could be recommended for clinical use therefore. Future research
will be necessary to confirm the psychometric quality of patient-specific
instruments measuring physical function in patients with musculoskeletal disorders.

Personalized measurement |85

Introduction

Musculoskeletal disorders are one of the major health care problems facing the Dutch
population. Low back pain is the most prevalent disorder, with a point prevalence of 24.1% in
the total population1. Frequently, patients with musculoskeletal disorders are faced with
disability which limits them in performing activities of daily living. Disability entails high
economic, societal and personal cost2,3. To diminish disability, non-pharmacological treatments
(such as rehabilitation or physical therapy) are focussed on both a patient’ physical functioning
and / or his context, including his psychological and social functioning4,5. Considering many
contextual factors that determine disability are common across musculoskeletal disorders and
even relevant to any chronic health condition, especially a patient’ physical functioning makes
the difference between the one and the other patient with musculoskeletal disorders5. To
assess a patients’ level of physical functioning and to evaluate the effect of interventions in the
clinical encounter, high quality measurements are necessary6.

Physical functioning can be assessed in different ways. Firstly, general measurement

instruments, like the MOS 36-Item Short Form Health Survey (SF-36) (physical function
subscale)7-9, are often applied. Secondly, disease-specific tools are used, for example the Neck
Disability Index10. Both general and disease-specific tools are instruments which content is
completely set, irrespective a patients’ health problem, request for help or nature of
complaints. Data produced by these so called ‘fixed-item’ tools are convenient and relatively
simple to categorize and compare across patients and settings11.

However, fixed-item tools are often difficult to interpret on an individual patient level. These
tools do not consider patients’ preferences and variability in performance on particular
activities12. For example, the ability to perform gardening will be of low relevance to a patient
who does not own a garden or such as the ability to climb stairs will not be relevant to a
patient who always takes the elevator.

The interest in so-called patient-specific outcomes which address each patient’s priorities in
outcome assessment, is increasing in clinical practice and research12,13. In contrast to fixed-
item instruments, patient-specific instruments can identify relevant issues on an individual level
and allow the evaluation to focus on what is important to each patient12. Similar to fixed-items
instruments, patient-specific instruments have limitations. The question arises to what extent
the outcomes are comparable between patients, because of the individualized content. The
application of statistical techniques is therefore questionable. In addition, floor effects may
occur as patients will choose difficult tasks as ‘most important impaired activities’11.

Nevertheless, because of the increase of a ‘patient-based concept’ in health policy,

individualized outcome tools will become more and more important. Several patient-specific
self-assessment instruments (self-administrated and interview-based) have been developed in

86
patients with musculoskeletal disorders12. However, a complete overview including a
psychometric quality assessment of available patient-specific self-assessment instruments
concerning physical function is lacking.

The objective of the current study is to identify available patient-specific self-assessment

instruments which measure physical function in patients with musculoskeletal disorders and
to evaluate the descriptive properties as well as the psychometric qualities of these
instruments.

Methods

Search strategy
An extensive search strategy was conducted in the electronic databases PubMed (1966 –
December 2011), CINAHL (1982 – December 2011) and Embase (1988 - December 2011).
The search strategy was built upon four elements: 1. outcome assessment; 2. patient-specific
character of outcome assessment; 3. outcome dimension physical function and 4.
psychometric qualities. The search strategy was formulated in PubMed and adapted for use in
other databases (see appendix 1). Additionally, reference lists of all relevant articles were
screened to include potential articles.

Selection criteria
The following inclusion criteria were used:

1. The instrument is a questionnaire, a rating scale or a (semi-structured) interview; 2. The

instrument minimally measures the dimension ‘physical function’; 3. The instrument is applied
to patients with musculoskeletal disorders; 4. The instrument has a patient-specific character;
5. Investigating the measurement properties of the instrument was the main aim of the study;
6. Publications describing different versions of the same test with different items were both
included;
7. For practical reasons, only articles published in English, German, French or Dutch were
considered for inclusion; 8. Only full-length published articles were included.

An article was excluded if: 1. The instrument was a performance based test; 2. The instrument
was a different language version of an original instrument.

Study selection procedure

The study selection process was performed in two stages. The first selection on titles and
subsequently on abstracts, was independently performed by two reviewers (JaB, PaH). The
second step comprised reviewing full-text articles against the mentioned inclusion criteria.
Disagreements concerning selection and inclusion of studies were resolved by discussion. A
third reviewer (MfP) was consulted in case of persisting disagreement.

Personalized measurement |87

Data extraction
Study characteristics (authors, title, year of publication) and descriptive characteristics of the
instrument were extracted from the selected articles, including name of the instrument, target
population, size of the population in which the instrument was applied, purpose of the
instrument, outcome domain(s), referral to a specific time period, mode of administration
(self-administered or interview based), mode of selecting items, number of scales, number of
items per scale, response options, range of scores and time to administer.

Psychometric quality assessment

Psychometric quality was determined using the standardized checklist ‘quality criteria for
measurement properties of health status questionnaires’14 (see appendix 2). This checklist,
developed by Terwee et al., is based upon the criteria of the ‘Scientific Advisory Committee
(SAC) of the Medical Outcomes Trust15’. In contrast to the SAC-list, the list of Terwee et al.
mentions explicit criteria for what constitutes good measurement properties. All
measurement properties were rated as ‘+’ (positive), ‘?’ (doubtful), ‘-‘ (negative) or ‘0’ (no
information available). The following measurement properties were evaluated:

Content validity. Content validity examines the extent to which the domain of interest is
comprehensively represented by the items in the instrument16. This term was operationalized
by describing the measurement aim of the instrument, the target population, the measurement
concept, the way of item selection and the interpretability of the items17,18.

Internal consistency. Internal consistency is a measure of the extent to which items in a

instrument (sub)scale are correlated, thus measuring the same construct. Factor analysis
should have been applied to determine whether the items of a scale measure the same
construct. In addition, Cronbach’s alpha should have been calculated as a measure of internal
consistency. An alpha of ≥.70 was considered acceptable19.

Criterion validity. Criterion validity refers to the extent to which scores on a particular
instrument are related to a gold standard. Positive evaluations were given when the gold
standard was convincingly described as a real gold standard and if the correlation with the gold
standard was at least .70.

Construct validity. Construct validity is a measure of the extent of which scores on a particular
instrument relate to other measures in a manner that is consistent with theoretically derived
hypotheses concerning the concepts that are being measured17,18. Construct validity was
considered adequate if specific hypotheses were defined regarding the relationships with other
measures of physical function and if ≥75% of these hypotheses were confirmed.

Reproducibility. Reproducibility concerns the amount of which repeated measurements in

stable persons provide similar results. Reproducibility can be divided into two aspects:

88
reliability and absolute agreement. Reliability refers to the extent to which patients can be
distinguished from each other, despite measurement errors (relative measurement error).
Intraclass Correlation Coefficient (ICC) (continuous data) or weighted Cohen’s Kappa
(discrete data) were regarded as adequate measures18. A value of .70 was used as a minimum
standard19.
Agreement describes the extent to which the scores on repeated measures are similar to
each other (absolute measurement error). Bland & Altman Limits of Agreement (LOA) and
the smallest detectable change (SDC) were considered adequate measures of agreement19-22.
A positive rating was assigned if the minimal important change (MIC) was outside the LOA or
if the SDC was smaller than the MIC14.

Responsiveness. Responsiveness was defined as the ability of an instrument to detect clinically

important changes over time in the concept to be measured23-25. Responsiveness could be
determined in many different ways14. However, considering responsiveness as a measure of
longitudinal validity, responsiveness is independent of the treatment effect. It was considered
adequate if the SDC was smaller than the MIC, if the responsiveness ratio (RR) was ≥1.9626
or if the area under the ‘receiver operating characteristics’ (ROC) curve was ≥.7023.

Floor and ceiling effects. Floor and ceiling effects were considered to be present if more than
15% of the respondents achieved the lowest or highest possible score, respectively27. A
positive rating was assigned if floor and ceiling effects were absent.

Interpretability. Interpretability is defined as the extent to which one can assign qualitative
meaning to quantitative scores28. To assess interpretability, means and standard deviations
(SD) of relevant groups should have been presented. In addition, the MIC should have been
defined. Interpretability was scored positively if mean scores and SD were presented of at
least four subgroups of patients and if the MIC was defined.

Psychometric quality assessment was conducted by two reviewers independently (JaB, MfP).
When disagreement was found between the two reviewers, the measurement quality which
was subject of disagreement, was discussed. A third reviewer (CV) was consulted in case of
persisting disagreement.

Overall quality
To obtain an overall score for psychometric quality of the identified instruments, the number
of positive ratings out of the total rated items for each instrument was counted.

Personalized measurement |89

Results

Study selection procedure

The literature search identified 1617 unique articles. After the selection procedure, 23 studies
were included, referring to twelve different instruments (figure 1). The full names of the
investigated instruments are presented in Table 1. The percentage of agreement between the
two reviewers amounted 88% after stage one and 75% after stage two of the selection
procedure.

Literature search:1617 unique articles

Title screening: 1617 unique articles

1331 articles excluded

Abstract screening: 286 articles

190 articles excluded before discussion between reviewers

29 articles excluded after discussion between reviewers

Full-text assessment: 67 articles

36 articles excluded before discussion:

No full-text (n=3) (29-31)
No musculoskeletal disorder (n=2) (32;33)
No patient-specific measure (n=17) (34-50)
No description of psychometric quality (n=6) (51-56)
Translated instrument (n=6) (57-62)
No physical function (n=1) (63)
Data combined from several studies (n=1) (64)

6 articles excluded after discussion:

No musculoskeletal disorder (n=1) (65)
No patient-specific measure (n=2) (66;67)
No description of psychometric quality (n=3) (11;68;69)

2 articles excluded after assessment of 3rd reviewer:

No patient-specific measure (70;71)

23 articles included in the review

FIGURE 1. Selection procedure of a systematic review on patient-specific instruments measuring physical functioning

90
 TABLE 1. Full names of the included instruments
Abbreviation Full name

COPM Canadian Occupational Performance Measure

I-HAQ-DI Individualized Health Assessment Questionnaire Disability Index
IMAS Individualized Milliken Activities of Daily Living Scale
I-PSI Interviewer-administrated Patient-Specific Index
I-WOMAC Individualized Western Ontario and McMaster Universities Osteoarthritis Index
MACTAR McMaster Toronto Arthritis Patient Preference Disability Questionnaire
PSA Patient-Specific Approach
PSAQ Patient-Specific Activity Questionnaire
PS-DASH Patient-Specific Disabilities of the Arm, Shoulder and Hand questionnaire
PSFS Patient-Specific Functional Scale
SMCS Severity of the main complaints scale
S-PSI Self-reported Patient-Specific Index

Description of the instruments

Three instruments, derived from eleven articles, were found which represented minor
variations of the same construct measured, namely: patients rate their most important
disabilities and evaluate these disabilities over time. The most common example to measure
this construct is the Patient-Specific Functional Scale (PSFS). Nine different versions of the
PSFS were included in this review72-80. These versions differed in target population, mode of
administration (interview based versus self-administrated) and scoring method. The Severity
of the Main Complaints Scale (SMCS)81 measures the same construct, as well as the McMaster
Toronto Arthritis Patient Preference Disability Questionnaire (MACTAR) does. However, the
MACTAR additionally investigates patients’ general, emotional and social health status82,83.
More extended versions of the PSFS rate, beside the most important impaired activities, the
severity of these impairments and / or the frequency of the impaired activities in daily life.
Hereby, two measurement tools offer patients the possibility to define their most important
impaired activities84,85; four measurement tools permit patients to choose impaired activities
out of a pre-defined list12;86-88 and also four measurement tools score severity and importance
of all activities named in a pre-defined list13,89-91.
Three studies described instruments which were derived from existing instruments: the
Patient-specific Arm, Shoulder and Hand questionnaire (PS-DASH)90, the Individualized Health
Assessment Questionnaire Disability Index (I-HAQ-DI)12 and the Individualized Western
Ontario and McMaster Universities Osteoarthritis Index (I-WOMAC)88. The original DASH,
HAQ and WOMAC are extensively psychometrically evaluated and widely used in clinical
practice.
The majority of the identified instruments are used to evaluate changes over time 13,72,73,75-78,81,84-
91
. Only the PSFS described by Westaway and colleagues and the I-HAQ-DI have a
discriminative purpose12,79.

All descriptive characteristics of the included measurement tools are presented in Table 2.

Personalized measurement |91

 TABLE 2. Description of the included patient-specific instruments
Measurement Purpose Target population Size* Outcome Mode of Mode of No. of Number of Response Range of Time to
tool domain(s) administration item scales items per scale options scores administer
selection
COPM 84;92 Evaluative Various musculo- 237 Physical Interview-based Patient 2 Unlimited 0-10 0-100 30-45
skeletal disorders functioning minutes
(small part neuro- (performance)
muscular
disorders)
I-HAQ-DI 12 Discrimi- Rheumatoid 370 Physical Self- Investigator 2 5 0-3 0-45 ?
native arthritis functioning administrated + Patient
(capacity)
IMAS 89 Evaluative Post-surgical 45 Physical Self- Patient 6 Meal: 8 5+3 47-235 10 minutes
traumata upper functioning administrated Hygiene: 9
extremity (capacity & Dressing: 8
performance) Manipulation: 9
Cleaning: 7
Else: 6
I-PSI 13 Evaluative Total Hip 74 Physical Self- Investigator 4 6, 7, 15, 28 7 0-100 16 ± 5.5
Replacement functioning administrated + Patient percen- minutes
(capacity & tile score
performance)
I-WOMAC 88 Evaluative Hip / knee 1218 Physical Self- Investigator 1 5 0-100 0-500 ?
osteoarthritis functioning administrated + Patient
(performance)
MACTAR 82,83 Evaluative Rheumatoid 155 Physical Interview-based Patient 2 Trans: 0-10 1-7 11-47 15 minutes
arthritis, Chronic functioning Status: 5-9
low back pain 100 (performance)
Health status
PSA by Evaluative Low back pain 150 Physical Self- Patient 1 3 0-100 0-300 ?
Beurskens 86 functioning administrated
(performance)
PSA by Rollman Evaluative Temporo- 132 Physical Self- Patient 1 1 0-100 0-100 ?
85 mandibular functioning administrated
disorders (performance)
Psychometric quality assessment
Psychometric properties of the included patient-specific instruments are presented in Table 3.
None of these instruments has been tested on all psychometric quality criteria of the checklist.
One instrument has been tested positively on four items73, nine on two items 13,72,78-80,84,85,87,91,92,
five on one item74,77,81,82,88 and six instruments has not been evaluated positively on any of the
items12,75,76,86,89,90.

Content validity. None of the instruments scored positively on content validity, unless the
measurement aim, the concept being measured and the target population were mostly
described well. Content validity was often assessed as ‘unknown’, because it was unclear
whether both the patient and an investigator or an expert were involved in item selection.

Internal consistency. Two instruments (Individualized Western Ontario and McMaster

Universities Osteoarthritis Index (I-WOMAC) and I-HAQ-DI)12,88 were assessed on internal
consistency. Although Cronbach’s alpha was .93 and .87 respectively, internal consistency was
scored as ‘doubtful’, because confirmatory factor analysis was not performed.

Criterion validity. The self-reported version of the Patient Specific Index (S-PSI) was positively
related to the interview-based version of the Patient Specific Index (I-PSI) (Pearson product
moment correlation (r) = .78) and therefore obtained a positive score on criterion validity91.
The Individualized Milliken Activities of Daily Living Scale (IMAS) instrument was not related
to a convincible gold standard and was scored as ‘doubtful’89. Criterion validity was not
assessed for the remaining instruments.

Construct validity. Ten studies presented specific hypotheses regarding the strength and
direction of expected correlations with other measurement tools which are supposed to
measure physical function13,72-74,78-80,82,84,90. Eight out of these ten studies obtained a positive
score. Two studies obtained a negative score, because less than 75% of the hypotheses was
confirmed79,90. The Individualized Western Ontario and McMaster Universities Osteoarthritis
Index (I-WOMAC) and the I-HAQ-DI were assessed on construct validity without
hypotheses-testing, but with comparison to other the original instrument12,88. In nine studies,
construct validity was not assessed.

Reproducibility – reliability. The PSFS described by Chatman, Cleland, Stratford and Westaway,
as well the I-WOMAC, the I-PSI, the S-PSI and the Patient Specific Approach (PSA) described
by Rollman seem to be reliable instruments to assess patient-specific physical function (.72 ≤
ICC ≤ .92)13,72,73,78,79,85,88,91. The PSFS described by Young showed an ICC of .17 and was rated
negatively as a consequence80. One study used Pearson correlations to express reliability (r
=0.91) and was therefore scored as ‘doubtful’89.

Personalized measurement |93

Reproducibility – agreement. Two instruments obtained positive scores on agreement73,79. In
both cases, the SDC was smaller than the MIC. The PSFS by Stratford et al.78 presented a
Standard Error of Measurement (SEM) of .41, but did not define a MIC and subsequently
obtained a ‘doubtful’ score.

Responsiveness. Information on responsiveness lacked in four studies74,89-91. The PSFS by

Cleland indicated an AUC of .99, whereas the PSFS by Stewart and by Young indicated both
an AUC of .7173,77,78. The Canadian Occupational Performance Measure (COPM) showed an
AUC of .79, the SMCS showed an AUC of .82, the Patient-Specific Activity Questionnaire
85-87
(PSAQ) of .75 and the PSA described by Rollman presented an AUC of .91 . An AUC ≥.70
is required for a positive score. The PSFS by Pengel et al. was also scored positively, based on
a Guyatt Responsiveness Ratio of 2.176. Conform the criteria of Terwee et al., the remaining
studies used an inadequate design to assess responsiveness, such as a sensitivity-to-change
coefficient, the standardized response mean and the standard error.

Floor and ceiling effects. The PSAQ and the PSFS by Chatman et al. were evaluated with respect
to floor and ceiling effects. The PSAQ scored positively on this item, because only six percent
of the respondents had the lowest possible score87. Chatman et al. did not indicate a specific
percentage72 and was therefore assessed as ‘doubtful’.

Interpretability. None of the studies met all criteria concerning interpretability. Because mean
scores and standard deviations of at least four subgroups lacked, as well information about the
MIC, interpretability was scored as ‘no information available’ in all studies.

94
 TABLE 3. Psychometric quality of the included patient-specific instruments
Measurement tool Content Internal Criterion Construct Reproducibility Responsive Floor and Interpre- No. of positive
validity consistency validity validity Agreement Reliability -ness ceiling tability ratings / total no.
effects of ratings

COPM 84;92 0 0 0 + 0 0 + 0 0 2/2

I-HAQ-DI 12 0 ? 0 ? 0 0 ? 0 0 0/3

IMAS 89 0 0 ? 0 0 ? 0 0 0 0/2

I-PSI 13 0 0 0 + 0 + ? 0 0 2/3

I-WOMAC 88 0 ? 0 ? 0 + ? 0 0 1/4

MACTAR 82;83 ? 0 0 + 0 0 ? 0 0 1/3

PSA by Beurskens 86 0 0 0 0 0 0 ? 0 0 0/1

PSA by Rollman 85 0 0 0 0 0 + + 0 0 2/2

PSAQ 87 0 0 0 0 0 0 + + 0 2/2

PS-DASH 90 0 0 0 - 0 0 0 0 0 0/1

PSFS by Chatman 72 0 0 0 + 0 + ? ? 0 2/4

PSFS by Cleland 73 0 0 0 + + + + 0 0 4/4

PSFS by Gross 74 0 0 0 + 0 0 0 0 0 1/1

Discussion

Twenty-three studies referring to twelve different instruments which investigated

psychometric properties of patient-specific instruments measuring physical function in patients
with musculoskeletal disorders were identified. The PSFS is the most described tool, with nine
different variations. Extended versions of the PSFS were developed to additionally prioritize
the patient-mentioned impaired activities. In none of the instruments all measurement
properties proposed by Terwee et al. have been evaluated14. The PSFS described by Cleland
et al.73 achieved the highest score with four positively scored items. The PSAQ, the PSFS’s by
Cleland, Gross and Stewart, the SMCS, the S-PSI, the COPM and the PSA described by Rollman
obtained positive scores on all measurement properties which have been
evaluated73,74,77,81,84,85,87,91,92.

One of the treatment goals in patients with musculoskeletal disorders who are disabled in
performing activities of daily living, is to improve physical functioning. Assessing impaired
activities and evaluating them over time is an adequate method to meet this goal.
Nevertheless, the practical elaboration of this method varies across the different instruments.
Some instruments only identify impaired activities and rank them with respect to severity72-
76,78-82,85
, whereas other instruments additionally investigate the importance of the performance
on the impaired activities12,84,88,90 or the frequency of the impaired activities in daily life86,87.

The differences in practical elaboration become even more clear in the variations between
different versions of the PSFS. Nine identified studies comprised different versions of the
PSFS72-80. Although all these instruments are designated as ‘PSFSs’, the scoring method and
mode of administration vary between the measurement tools. This hampers comparison
between different patients or conditions.

The method of identifying impaired activities differed between the identified instruments. Most
instruments allow patients to formulate their own impaired activities. On the other hand, the
SMCS, PSAQ, PS-DASH, IMAS, I-WOMAC, I-HAQ, I-PSI, S-PSI and MACTAR use predefined
lists of potential activities12,13,81-83,87-91. The advantage of the application of predefined lists is the
facilitated possibility for comparison with and between different populations and settings.
Furthermore, change over time can be indicated easier in disabilities with potential for
improvement. On the contrary, the application of predefined lists entails the risk of missing
important activities11.

Psychometric quality was extracted using a checklist developed by Terwee et al.14. In total, 189
items were assessed. One instrument achieved four out of eight positive scores73; six
instruments noted no positive score12,75,76,86,89,90. Hereby, the overall psychometric quality of
patient-specific instruments which measure physical function seems to be low. However, only
three items obtained an insufficient score. The vast majority (75%) has been assessed as ‘no

96
information available’. The relative short existence of patient-specific measures might explain
this phenomenon. Nevertheless, despite the scarcity of psychometric data concerning patient-
specific instruments, construct validity, reliability as well as responsiveness were investigated
in more than half of the studies. Construct validity was positively assessed in eight out of
twelve studies in which this quality was evaluated13,72-74,78,80,82,84. It can be concluded that patient-
specific instruments seem to measure the same construct as disease-specific or generic
physical function-tools. Eight instruments appeared to be reliable13,72,73,78,79,85,88,91. However, in
case of evaluative tools, responsiveness is possibly a more appropriate property than reliability.
The most adequate approach for evaluating responsiveness is still unclear25. Therefore, it is
not surprising that we found many different ways in which responsiveness had been
determined. Conform the criteria of Terwee et al.14, eight studies used an adequate method
(ROC-curve or Guyatts RR)73,76,77,80,81,85,87,92. Nine studies used inadequate methods, like the
SRM and the ES. However, the SRM and ES are widely administered in psychometric research
to assess responsiveness. Therefore, the question arises whether this measurement property
is not judged too strictly.
Floor and ceiling effects have been poorly investigated yet. However, patients may indicate
very difficult activities as their most impaired activities and, as a consequence, rate these
activities with the lowest possible score. The likelihood of improvement of these impossible
activities is small, even as the improvement in rating11. Therefore, more solid research on floor
and ceiling effects of patient-specific instruments is needed.

Some limitations of this systematic review should be acknowledged. First, patient-specific

instruments are developed to enable clinicians to measure changes in activities which really
matter to individual patients. However, to determine the psychometric quality of patient-
specific instruments, evaluation should occur in clinical practice in addition to evaluation in
population studies.’ Moreover, this review was limited to studies with a primary goal to
determine psychometric qualities; studies in which several measurement tools were applied
to detect differences in physical function might provide complementary evidence concerning
measurement properties. Third, the checklist developed by Terwee et al. is not a gold standard
to determine psychometric quality of a instrument. It is just a tool to check whether the
properties are clear and systematically presented. Thereby, the criteria to obtain a positive
score are very stringent sometimes. For example, content validity is exclusively supposed to
be positive if both the measurement aim, the target population, the concepts that are being
measured, the item selection and the involvement in item selection were sufficiently described.
Missing just one of these criteria induce a ‘doubtful’ score. A final remark concerning the
checklist of Terwee et al. is the unavailability to determine the overall best instrument.
Nevertheless, based on the current evidence, the PSFS described by Cleland et al.73 could be
recommended to use in clinical practice to monitor a patients’ progress in a patient-specific
way. This instrument has been evaluated most comprehensively and obtained just positive
scores. In future, more studies, performed in clinical practice, are needed to confirm the
psychometric quality of patient-specific instruments in general and the PSFS described by
Cleland in particular.

Personalized measurement |97

In conclusion, although twelve instruments have been developed to measure patient-specific
function in patients with musculoskeletal disorders, the descriptive properties and
psychometric quality of these instruments are only partly investigated. The PSFS is the most
investigated instrument: nine different versions have been evaluated psychometrically. The
version of Cleland et al.73 was most extensively investigated, obtained exclusively positive
scores following the quality criteria by Terwee et al.14 and could be recommended for clinical
use therefore. Future research will be necessary to confirm the psychometric quality of
patient-specific instruments measuring physical function in patients with musculoskeletal
disorders.

98
References diverse patient groups. Med Care 1994; 32:40-
66.

1. Picavert H, Schouten J. Musculoskeletal pain in 9. Ware J, Sherbourne C. The MOS 36-Item

the Netherlands: prevalences, consequences Short-Form Health Survey (SF-36), I:
and risk groups, the DMC(3)-study. Pain 2003; conceptual framework and item selection. Med
102(1-2):167-178. Care 1992; 30:473-483.

2. Dunlop D, Semanik P, Song J, Manheim L, Shih 10. Vernon H, Mior S. The Neck Disability Index:
V, Chang R. Risk factors for functional decline a study of reliability and validity. J Manupulative
in older adults with arthritis. Arthritis Rheum Physiol 1991; 14:409-415.
2005; 52(4):1274-1282.
11. Jolles BM, Buchbinder R, Beaton DE. A study
3. Miller M, Rejeski W, Reboussin B, Ten Have T, compared nine patient-specific indices for
Ettinger W. Physical activity, functional musculoskeletal disorders. J Clin Epidemiol
limitations, and disability in older adults. J Am 2005; 58(8):791-801.
Geriatr Soc 2000; 48(10):1264-1272.
12. Seror R, Tubach F, Baron G, Guillemin F,
4. American College of Sports Medicine, Ravaud P. Measure of function in rheumatoid
Chodzko-Zaijko W, Proctor D, Fiatarone arthritis: individualized or classical scales? Ann
Singh M, Minson C, Nigg C et al. American Rheum Dis 2010; 69(1):97-101.
College of Sports Medicine position stand.
Exercise and physical activity for older adults. 13. Wright J, Young N. The patient-specific index:
Med Sci Sports Exerc 2009; 41(7):1510-1530. asking patients what they want. J Bone Joint
Surg Am 1997; 79:974-983.
5. Weigl M, Cieza A, Cantista P, Reinhardt J,
Stucki G. Determinants of disability in chronic 14. Terwee C, Bot S, de Boer M, van der Windt
musculoskeletal health conditions: a literature D, Knol D, Dekker J et al. Quality criteria were
review. Eur J Phys Rehabil Med 2008; 44:67-79. proposed for measurement properties of
health status questionnaires. J Clin Epidemiol
6. Tugwell P, Bombardier C, Buchanan W, 2007; 60:34-42.
Goldsmith C, Grace E, Hanna B. The
MACTAR Patient Preference Disability 15. Scientific Advisory Committee of the Medical
Questionnaire--an individualized functional Outcome Trust. Assessing health status and
priority approach for assessing improvement quality of life instruments: attributes and
in physical disability in clinical trials in review criteria. Qual Life Res 2002; 11:193-
rheumatoid arthritis. J Rheumatol 1987; 205.
14(3):446-451.
16. Guyatt G, Feeny D, Patrick D. Measuring
7. McHorney C, Ware J, Raczek A. The MOS 36- health related quality of life. Ann Intern Med
Item Short-Form Health Survey (SF-36), II: 1993; 118:622-629.
psychometric and clinical tests of validity in
measuring physical and mental health 17. Kirshner B, Guyatt B. A methodological
constructs. Med Care 1993; 31:247-263. framework for assessing health indices. J
Chronic Dis 1985; 38:27-36.
8. McHorney C, Ware J, Lu R, Donald
Sherbourne C. The MOS 36-item short-form 18. Streiner D, Norman G. Health Measurement
health survey (SF-36): III. Tests of data-quality, Scales; a practical guide to their development
scaling assumptions, and reliability across and use. Fourth edition ed. Oxford: Oxford
University Press, 2008.

Personalized measurement |99

19. Nunnally J. Psychometric theory. 2nd edition of scientific review criteria. Clin Ther 1996;
ed. New York: McGraw-Hill, 1978. 18:979-992.

20. Beckerman H, Roebroeck M, Lankhorst G, 29. Carpenter L, Baker GA, Tyldesley B. The use
Becher J, Bezemer P, Verbeek A. Smallest real of the Canadian occupational performance
difference, a link between reproducibility and measure as an outcome of a pain management
responsiveness. Qual Life Res 2001; 10:571- program. Can J Occup Ther 2001; 68(1):16-22.
568.
30. Hefford C, Lodge S, Elliott K, Abbott JH.
21. Bland J, Altman D. Statistical methods for Measuring patient-specific outcomes in
assessing agreement between two methods of musculoskeletal clinical practice: a pilot study.
clinical measurement. Lancet 1986;307-310. N Z J Physiother 2008; 36(2):41-48.

22. de Vet H, Bouter L, Bezemer P, Beurskens A. 31. Sterling M, Brentnall D. Patient specific
Reproducibility and responsiveness of functional scale. Australian Journal of
evalutive outcome measures. Int J Technol Physiotherapy 2007; 53(1):65.
Assess Health Care 2001; 17:479-487.
32. Juniper EF, Buist AS, Cox FM, Ferrie PJ, King
23. Deyo R, Centor R. Assessing the DR. Validation of a standardized version of the
responsiveness of functional scales to clinical Asthma Quality of Life Questionnaire. Chest
change: an analogy to diagnostic test 1999; 115(5):1265-1270.
performance. J Chron Dis 1986; 39:897-906.
33. Nakawatase Y, Taru C, Tsutou A, Shiotani H,
24. Mokkink L, Terwee C, Knol D, Stratford P, Kido Y, Ohara T et al. Development of an
Alonso J, Patrick D et al. The COSMIN evaluation scale for self-management behavior
checklist for evaluating the methodological related to physical activity of type 2 diabetic
quality of studies on measurement properties: patients. Diabetes Care 2007; 30(11):2843-
A clarification of its content. BMC Medical 2848.
Research Methodology 2010; 10(22).
34. Akai M, Doi T, Fujino K, Iwaya T, Kurosawa H,
25. Terwee C, Dekker F, Wiersinga W, Prummel Nasu T. An outcome measure for Japanese
M, Bossuyt P. On assessing responsiveness of people with knee osteoarthritis. J Rheumatol
health-related quality of life instruments: 2005; 32(8):1524-1532.
Guidelines for instrument evaluation. Qual Life
Res 2003; 12:349-362. 35. Alexander M, Franko OI, Makhni EC,
Zurakowski D, Day CS. Validation of a modern
26. Guyatt G, Walter S, Norman G. Measuring activity hand survey with respect to reliability,
change over time: assessing the usefulness of construct and criterion validity. J Hand Surg
evaluative instruments. J Chron Dis 1987; Eur Vol 2008; 33(5):653-660.
40(2):171-178.
36. Alonso J, Lamarca R, Marti V. The pain and
27. McHorney C, Tarlov A. Individual-patient function of the Hip (PFH) scale: A patient-
monitoring in clinical practice: are available based instrument for measuring outcome after
health status surveys adequate? Qual Life Res total hip replacement. Orthopedics 2000;
1995; 4:293-307. 23(12):1273-1277.

28. Lohr K, Aaronsson N, Alonso J, Burnam M, 37. Bago J, Perez-Grueso FJ, Les E, Hernandez P,
Patrick D, Perrin E et al. Evaluating quality of Pellise F. Minimal important differences of the
life and health status instruments: development SRS-22 Patient Questionnaire following

100
 surgical treatment of idiopathic scoliosis. Eur lower-limb amputees. Arch Phys Med Rehabil
Spine J 2009;18(12):1898-1904. 2011;92(8):1305-1310.

38. Davis SE, Chung KC. Validity and 46. El Miedany Y, El Gaafary M, El Aroussy N,
responsiveness of the Jebsen-Taylor Hand Ahmed I, Youssef S, Palmer D. Patient
Function Test. J Hand Surg Am 2010; 35(1):30- reported outcomes in ankylosing spondylitis:
37. development and validation of a new
questionnaire for functional impairment and
39. Dawson J, Doll H, Boller I, Fitzpatrick R, Little quality of life assessment. Clin Exp Rheumatol
C, Rees J et al. The development and validation 2011; 29(5):801-810.
of a patient-reported questionnaire to assess
outcomes of elbow surgery. J Bone Joint Surg 47. Gabel C, Melloh M, Burkket B, Michener L.
Br 2008;90B(4):466-473. Lower Limb Functional Index: Development
and Clinimetric Properties. Phys Ther 2011;
40. Fries JF, Cella D, Rose M, Krishnan E, Bruce B. Nov(3).
Progress in assessing physical function in
arthritis: PROMIS short forms and 48. Katz P, Radvanski D, Allen D, Buyske S, Schiff
computerized adaptive testing. J Rheumatol S, Nadkarni A. Development and validation of
2009; 36(9):2061-2066. a short form of the valued life activities
disability questionnaire for rheumatoid
41. Goodacre L, Smith J, Meddis D, Goodacre J. arthritis. Arthritis Care Res (Hoboken) 2011;
Development and validation of a patient- 63(12):1667-1671.
centred Measure of Activity Limitation (MAL)
in rheumatoid arthritis. Rheumatology 2007; 49. Thorborg K, Holmich P, Christensen R,
46(4):703-708. Petersen J, Roos E. The Copenhagen Hip and
Groin Outcome Score (HAGOS):
42. Irrgang JJ, Snyder-Mackler L, Wainner RS, Fu development and validation according to the
FH, Harner CD. Development of a patient- COSMIN checklist. Br J
reported measure of function of the knee. J Sports Med 2011; 45(6):478-491.
Bone Joint Surg Am 1998; 80(8):1132-1145.
50. Ornetti P, Dougados M, Paternotte S, Logeart
43. Moe RH, Garratt A, Slatkowsky-Christensen I, Gossec L. Validation of a numerical rating
B, Maheu E, Mowinckel P, Kvien TK et al. scale to assess functional impairment in hip and
Concurrent evaluation of data quality, knee osteoarthritis. Ann Rheum Dis 2011;
reliability and validity of the 70(5):740-746.
Australian/Canadian Osteoarthritis Hand
Index and the Functional Index for Hand 51. Andrew WD, Jane KS, Sebastian JP, Rajkumar
Osteoarthritis. Rheumatology 2010; S, Bennetts K. Performance problems of
49(12):2327-2336. patients with chronic low-back pain and the
measurement of patient-centered outcome.
44. Van Der Wees P., Hendriks E, Van Beers H, Spine (Phila Pa 1976 ) 2004; 29(1):87-93.
Van Rijn R, Dekker J, De Bie R. Validity and
responsiveness of the ankle function score 52. Bearon LB, Crowley GM, Chandler J, Robbins
after acute ankle injury. Scand J Med Sci Sports MS, Studenski S. Personal functional goals: A
2010. new approach to assessing patient-centered
outcomes. J Appl Gerontol 2000; 19(3):326-
45. de Laat F, Rommers G, Geertzen J, Roorda L. 344.
Construct validity and test-retest reliability of
the questionnaire rising and sitting down in 53. Gossec L, Dougados M, Rincheval N,
Balanescu A, Boumpas DT, Canadelo S et al.

Personalized measurement |101

 Elaboration of the preliminary Rheumatoid Assessment Questionnaire in patients
Arthritis Impact of Disease (RAID) score: a undergoing surgical or conservative inpatient
EULAR initiative. Ann Rheum Dis 2009; treatment. Quality of life research : an
68(11):1680-1685. international journal of quality of life aspects of
treatment, care and rehabilitation 2006;
54. Hefford C, Lodge S, Elliott K, Abbott JH. 15(7):1233-1241.
Measuring patient-specific outcomes in
musculoskeletal clinical practice: a pilot study. 63. Cott CA, Teare G, McGilton KS, Lineker S.
N Z J Physiother 2008; 36(2):41-48. Reliability and construct validity of the client-
centred rehabilitation questionnaire. Disabil
55. Marx RG, Jones EC, Atwan NC, Closkey RF, Rehabil 2006; 28(22):1387-1397.
Salvati EA, Sculco TP. Measuring improvement
following total hip and knee arthroplasty using 64. Hall AM, Maher CG, Latimer J, Ferreira ML,
patient-based measures of outcome. J Bone Costa LO. The patient-specific functional scale
Joint Surg Am 2005; 87A(9):1999-2005. is more responsive than the Roland Morris
disability questionnaire when activity limitation
56. Pouchot J. Toward individualized quality-of-life is low. Eur Spine J 2011; 20(1):79-86.
assessment. Jt Bone Spine 2000; 67(2):83-85.
65. Berghuis-Kelley D, Scherer S. Research
57. Berendes T, Pilot P, Willems J, Verburg H, te corner. Outcome measures in
SR. Validation of the Dutch version of the cardiopulmonary physical therapy: use of the
Oxford Shoulder Score. J Shoulder Elbow Surg Patient Specific Functional Scale. Cardiopulm
2010; 19(6):829-836. Phys Ther J 2007; 18(3):21-23.

58. Bizzini M, Gorelick M. Development of a 66. Amjadi SS, Maranian PM, Paulus HE, Kaplan
German version of the knee outcome survey RM, Ranganath VK, Furst DE et al. Validating
for daily activities. Arch Orthop Trauma Surg and assessing the sensitivity of the Health
2007; 127(9):781-789. Assessment Questionnaire-Disability Index-
derived Short Form-6D in patients with early
59. Costa LOP, Maher CG, Latimer J, Ferreira PH, aggressive rheumatoid arthritis. J Rheumatol
Ferreira ML, Pozzi GC et al. Clinimetric testing 2009; 36(6):1150-1157.
of three self-report outcome measures for low
back pain patients in Brazil: Which one is the 67. Dawson J, Coffey J, Doll H, Lavis G, Cooke P,
best? Spine 2008; 33(22):2459-2463. Herron M et al. A patient-based questionnaire
to assess outcomes of foot surgery: validation
60. Kjeken I, Dagfinrud H, Uhlig T, Mowinckel P, in the context of surgery for hallux valgus.
Kvien TK, Finset A. Reliability of the Canadian Qual Life Res 2006; 15(7):1211-1222.
Occupational Performance Measure in patients
with ankylosing spondylitis. J Rheumatol 2005; 68. Clinch J, Tugwell P, Wells G, Shea B.
32(8):1503-1509. Individualized functional priority approach to
the assessment of health related quality of life
61. Kohlmann T, Bullinger M, Kirchberger- in rheumatology. J Rheumatol 2001; 28(2):445-
Blumstein I. The German version of the 451.
Nottingham Health Profile (NHP): Method of
translation and psychometric validation. SOZ - 69. Sanchez K, Papelard A, Nguyen C, Jousse M,
PRAVENTIVMED 1997; 42(3):175-185. Rannou F, Revel M et al. Patient-Preference
Disability Assessment for Disabling Chronic
62. Wollmerstedt N, Kirschner S, Faller H, Konig Low Back Pain. Spine 2009; 34(10):1052-1059.
A. Reliability, validity and responsiveness of the
German Short Musculoskeletal Function

102
70. Chan Ci En M, Clair DA, Edmondston SJ. patients: a report of a patient specific measure.
Validity of the Neck Disability Index and Neck Physiotherapy Canada 1995; 47(4):258-263.
Pain and Disability Scale for measuring
disability associated with chronic, non- 79. Westaway MD, Stratford PW, Binkley JM. The
traumatic neck pain. Man Patient-Specific Functional Scale: validation of
Ther 2009; 14(4):433-438. its use in persons with neck dysfunction. J
Orthop Sports Phys Ther 1998; 27(5):331-338.
71. Hoving JL, O'Leary EF, Niere KR, Green S,
Buchbinder R. Validity of the neck disability 80. Young IA, Cleland JA, Michener LA, Brown C.
index, Northwick Park neck pain Reliability, construct validity, and
questionnaire, and problem elicitation responsiveness of the neck disability index,
technique for measuring disability associated patient-specific functional scale, and numeric
with whiplash-associated disorders. Pain 2003; pain rating scale in patients with cervical
102(3):273-281. radiculopathy. Am J Phys Med Rehabil 2010;
89(10):831-839.
72. Chatman AB, Hyams SP, Neel JM, Binkley JM,
Stratford PW, Schomberg A et al. The patient- 81. Beurskens AJ, de Vet HC, Koke AJ.
specific functional scale: Measurement Responsiveness of functional status in low back
properties in patients with knee dysfunction. pain: a comparison of different instruments.
Phys Ther 1997; 77(8):820-829. Pain 1996; 65(1):71-76.

73. Cleland JA, Fritz JM, Whitman JM, Palmer JA. 82. Verhoeven A, Boers M, van der Lindern S.
The reliability and construct validity of the Validity of the MACTAR Questionnaire as
neck disability index and patient specific Functional Index in a Rheumatoid Arthritis
functional scale in patients with cervical Clinical Trial. J Rheumatol 2000; 27:2801-2809.
radiculopathy. Spine 2006; 31(5):598-602.
83. Sanchez K, Papelard A, Nguyen C,
74. Gross DP, Battie MC, Asante AK. The Patient- Bendeddouche I, Jousse M, Rannou F et al.
Specific Functional Scale: Validity in Workers' McMaster-Toronto Arthritis Patient
Compensation Claimants. Arch Phys Med Preference Disability Questionnaire Sensitivity
Rehabil 2008; 89(7):1294-1299. to Change in Low Back Pain: Influence of Shifts
in Priorities. Plos One 2011; 6(5):e20274.
75. McMillan CR, Binhammer PA. Which Outcome
Measure is the Best? Evaluating Responsiveness 84. Dedding C, Cardol M, Eyssen IC, Dekker J,
of the Disabilities of the Arm, Shoulder, and Beelen A. Validity of the Canadian
Hand Questionnaire, the Michigan Hand Occupational Performance Measure: a client-
Questionnaire and the Patient-Specific centred outcome measurement. Clin Rehabil
Functional Scale Following Hand and Wrist 2004; 18(6):660-667.
Surgery. Hand (N Y ) 2009; 4(3):311-318.
85. Rollman A, Naeije M, Visscher CM. The
76. Pengel LHM. Responsiveness of Pain, Disability, reproducibility and responsiveness of a
and Physical Impairment Outcomes in Patients patient-specific approach: a new instrument in
with Low Back Pain. evaluation of treatment of temporomandibular
disorders. J Orofac Pain 2010; 24(1):101-105.
77. Stewart M, Maher CG. Responsiveness of pain
and disability measures for chronic whiplash. 86. Beurskens AJ, de Vet HC, Koke AJ, Lindeman
E, van der Heijden GJ, Regtop W et al. A
78. Stratford P, Gill C, Westaway M, Binkley J. patient-specific approach for measuring
Assessing disability and change on individual functional status in low back pain. J Manip
Physiol Ther 1999; 22(3):144-148.

Personalized measurement |103

87. Frost MH, Reeve BB, Liepa AM, Stauffer JW, of Daily Living Scale. J Occup Rehabil 2005;
Hays RD. What is sufficient evidence for the 15(3):343-351.
reliability and validity of patient-reported
outcome measures? Value Health 2007; 10 90. Vranceanu AM, Kadzielski J, Hwang R, Ring D.
Suppl 2:S94-S105. A patient-specific version of the Disabilities of
the Arm, Shoulder, and Hand Questionnaire. J
88. Seror R, Tubach F, Baron G, Falissard B, Hand Surg Am 2010; 35(5):824-826.
Logeart I, Dougados M et al. Individualising the
Western Ontario and McMaster Universities 91. Wright J. Evaluating the outcome of treatment.
osteoarthritis index (WOMAC) function Shouldn't we be asking patients if they are
subscale: incorporating patient priorities for better? J Clin Epidemiol 2000; 53:549-553.
improvement to measure functional
impairment in hip orknee osteoarthritis. Ann 92. Eyssen IC, Steultjens M, Oud T, Bolt E,
Rheum Dis 2008; 67:494-499. Maasdam A, Dekker J. Responsiveness of the
Canadian Occupational Performance Measure.
89. Seaton MK, Groth GN, Matheson L, Feely C. J Rehabil Res Dev 2011; 48(5):517-528.
Reliability and validity of the Milliken Activities

104
Appendix 1: Search strategy Pubmed (1966 through December 2011)

Outcome assessment (health care) (MeSH) OR Questionnaires (MeSH) OR Assessment

Measure* OR Scale* OR Test* OR Instrument* OR Index OR Indices OR “Measurement
scale” OR “Rating scale” OR “Self reported scale” OR “Self reported questionnaire” OR “Self
administrated questionnaire”

AND

“Patient-centered care” (MeSH) OR Individualized OR “Individualized outcome” OR

“Individualized scale” OR “Client centered” OR “Patient based” OR “Patient specific” OR
“Patient specific measures” OR “Patient preference” OR “Patient reported” OR “Self
reported”

AND

“Activities of Daily Living” (MeSH) OR “Mobility limitation” (MeSH) OR “Motor activity”

(MeSH) OR Functional* OR “Functional index” OR “Functional stat*” OR “Functional ability”
OR “Functional disability” OR Ability OR Disability OR “Physical ability” OR “Physical
disability” OR “Physical function” OR “Physical functioning” OR “Physical performance” OR
“Performing activities” OR “Limited activities” OR “Functional limitation” OR “Activity
limitation”

AND

“Reproducibility of results” (MeSH) OR Psychometrics OR Psychometric* OR “Psychometric

evaluation” OR “Psychometric properties” OR “Psychometric analysis” OR “Psychometric
characteristics” OR “Psychometric assessment” OR “Measurement properties” OR Valid* OR
Reliability OR Reliable OR Reproducibility OR Reproducible OR Responsive* OR Sensitivity
OR Sensitive OR Agreement OR “Internal consistency”

Search date: 6th December 2011

Limits: none
Hits: 161

Personalized measurement |105

Appendix 2: Checklist ‘Quality criteria for measurement properties of health status questionnaires’ (Terwee et al. (2007))

Property Definition Quality criteriaa,b

1 Content validity The extent to which the domain of interest + A clear description is provided of the measurement aim, the target population, the concepts that are being measured,
is comprehensively sampled by the items and the item selection AND target population and (investigators OR experts) were involved in item selection
in the questionnaire ? A clear description of above-mentioned aspects is lacking OR only target population involved OR doubtful design or
method
- No target population involvement
0 No information found on target population involvement.
2 Internal The extent to which items in a (sub)scale + Factor analyses performed on adequate sample size (7 * # items and >100) AND Cronbach’s alpha(s) calculated per
consistency are intercorrelated, thus measuring the dimension AND Cronbach’s alpha(s) between 0.70 and 0.95;
same construct ? No factor analysis OR doubtful design or method;
- Cronbach’s alpha(s) <0.70 or >0.95, despite adequate design and method;
0 No information found on internal consistency.
3 Criterion validity The extent to which scores on a particular + Convincing arguments that gold standard is ‘‘gold’’ AND correlation with gold standard ≥0.70;
questionnaire relate to a gold standard ? No convincing arguments that gold standard is ‘‘gold’’ OR doubtful design or method;
- Correlation with gold standard <0.70, despite adequate design and method;
0 No information found on criterion validity.
4 Construct The extent to which scores on a particular + Specific hypotheses were formulated AND at least 75% of the results are in accordance with these hypotheses;
validity questionnaire relate to other measures in ? Doubtful design or method (e.g., no hypotheses);
a manner that is consistent with - Less than 75% of hypotheses were confirmed, despite adequate design and methods;
theoretically derived hypotheses 0 No information found on construct validity.
concerning the concepts that are being
measured
5 Reproducibility
Agreement The extent to which the scores on + MIC ≥ SDC OR MIC outside the LOA OR convincing arguments that agreement is acceptable;
repeated measures are close to each ? Doubtful design or method OR (MIC not defined AND no convincing arguments that agreement is acceptable);
other (absolute measurement error) - MIC < SDC OR MIC equals or inside LOA, despite adequate design and method;
0 No information found on agreement.
Reliability + ICC or weighted Kappa ≥ 0.70;
 Property Definition Quality criteriaa,b

The extent to which patients can be ? Doubtful design or method (e.g., time interval not mentioned);
distinguished from each other, despite - ICC or weighted Kappa < 0.70, despite adequate design and method;
measurement errors (relative 0 No information found on reliability.
measurement error)
6 Responsive-ness The ability of a questionnaire to detect + SDC < MIC OR MIC outside the LOA OR RR>1.96 OR AUC ≥ 0.70;
clinically important changes over time ? Doubtful design or method;
- SDC or SDC ≥ MIC OR MIC equals or inside LOA OR RR≤ 1.96 OR AUC < 0.70, despite adequate design and methods;
0 No information found on responsiveness.
7 Floor and ceiling The number of respondents who achieved + ≤ 15% of the respondents achieved the highest or lowest possible scores;
effects the lowest or highest possible score ? Doubtful design or method;
- > 15% of the respondents achieved the highest or lowest possible scores, despite adequate design and methods;
0 No information found on interpretation.
8 Interpretability The degree to which one can assign + Mean and SD scores presented of at least four relevant subgroups of patients and MIC defined;
qualitative meaning to quantitative ? Doubtful design or method OR less than four subgroups OR no MIC defined
scores 0 No information found on interpretation.
MIC = minimal important change; SDC = smallest detectable change; LOA = limits of agreement; ICC = Intraclass correlation; SD, standard deviation.
a
+ = positive rating; ? = indeterminate rating; - = negative rating; 0 = no information available.
b
Doubtful design or method = lacking of a clear description of the design or methods of the study, sample size smaller than 50 subjects (should be at least 50 in every (subgroup) analysis),
or any important methodological weakness in the design or execution of the study.

Personalized measurement |107

Validity and responsiveness of the
6
Dutch McMaster Toronto Arthritis
patient preference questionnaire
(MACTAR) in patients with
hip/knee osteoarthritis

Published in: J Rheumatol. 2012 May;39(5):1064-73

[Link]

Di-Janne Barten
Martijn Pisters
Tim Takken
Cindy Veenhof
Abstract

Objective To determine the content validity, the construct validity and the
responsiveness of the Dutch MACTAR in patients with osteoarthritis (OA) of the
hip or knee.

Methods The MACTAR comprises two parts: a transitional part and a status part.
Content validity was investigated by comparing patient-elicited activities to items
on the ‘Western Ontario and McMaster Universities Osteoarthritis Index’
(WOMAC) and the ‘Medical Outcome Survey Short Form 36’ (SF-36). Construct
validity was determined by correlating MACTAR outcomes with WOMAC/SF-36
outcomes. Responsiveness was investigated by correlating MACTAR, WOMAC
and SF-36 change scores with Patient Global Assessment (PGA) scores and plotting
a Receiver Operating Characteristics (ROC) curve.

Results Eleven percent of the 894 impaired activities identified by 192 patients
were not represented in either the WOMAC or the SF-36. The correlations (rs)
investigated for the MACTAR-transitional part varied between .27 and -.40; the
status part correlated moderately with the general health scale of the SF-36
(rs=.44). MACTAR change scores correlated better with PGA than WOMAC/SF-
36 change scores. The area under the ROC-curve amounted to .90.

Conclusion Our results suggest that the MACTAR exhibits moderate construct
validity and good responsiveness in a population of patients with OA of the hip or
knee. Furthermore, the MACTAR is potentially better able to detect changes over
time in activities that are important to individual patients compared to other tools
measuring physical function (WOMAC, SF-36). Therefore, clinicians could use the
MACTAR to evaluate clinically relevant changes over time in patient-specific
physical functioning.

MACTAR |111
Introduction

Osteoarthritis (OA) is a common chronic musculoskeletal disorder1, which can result in

moderate to severe limitations in physical functioning2. Limited physical functioning can lead
to a diminished quality of life3-5. OA treatment guidelines recommend exercise therapy to
reduce impairments in physical function due to OA6,7. Exercise therapy can thus enable
individuals to better meet the demands of daily living8-11.

A number of tools are available to clinicians to evaluate the effect of exercise therapy on
physical function. General, disease-specific and patient-specific tools can be distinguished, all
of which are applied as either (self-reported) questionnaires or performance-based tests. A
systematic review of the psychometric quality of both questionnaires and performance-based
tests in patients with OA of the hip or knee has been published recently12,13. The reviews
recommended the application of the ‘Western Ontario and McMaster Universities
Osteoarthritis Index’ (WOMAC) (14), the ‘Medical Outcome Survey Short Form 36’ (SF-36)15-
17
and multi-activity tests when evaluating physical function in patients with OA12,13.

Standardized tools, applied to all patients in an identical manner, are recommended for the
evaluation of physical functioning. Data produced by these tools may be conveniently and
relatively easily categorized and compared between patients and across settings18. However,
standardized tools are often difficult to interpret at individual level and fail to take account of
individual preferences and variation in the performance of particular activities18. Patient-specific
tools measuring physical function have been developed based on the need for a more ‘patient
centered’ approach as set out in health care policy and to enable clinicians to measure changes
in activities which really matter to individual patients19. In contrast with standardized tools,
patient-specific instruments can identify the relevant issues at individual level and allow the
evaluation to focus on what is important to each individual patient18. Although the possibilities
to compare statistical data between patients are minimal, the application of patient-specific
tools may improve the validity and responsiveness for the assessment of physical function18,19.

The ‘McMaster Toronto Arthritis Patient Preference Disability Questionnaire’ (MACTAR) is

one example of a patient-specific scale measuring physical function20,21 (See Appendix 1). The
objective of the MACTAR is to identify individual disabilities due to the disease and their
relative importance to the patient, complemented by questions on general health status20. The
MACTAR has been described as a highly responsive and valid tool for the evaluation of physical
function in patients with rheumatoid arthritis21. A recent psychometric evaluation of the
questionnaire in patients with chronic lower back pain and patients with systemic sclerosis
showed moderate correlations with general and disease-specific tools which measure physical
function22,23.

112
To enable clinicians to use the MACTAR when evaluating physical function in patients with
OA of the hip or the knee, the psychometric properties of the questionnaire in this specific
population must be determined. Therefore, the objective of the present study is to determine
the content validity, the construct validity and the responsiveness of the MACTAR in patients
with OA of the hip or knee.

Methods

Study design
Data reported in this study were collected from a cluster randomized controlled trial of 200
patients with OA of the hip or knee over a 12-week period (maximum 18 sessions), that
compared behavioral graded activity with usual care in accordance with the Dutch physical
therapy guidelines24. The content of the interventions has been described elsewhere24. The
Medical Ethics Committee of the VU University Medical Center, Amsterdam, The Netherlands
approved the study. For the purposes of the present validation study, data on ‘physical
function’ were used, as well as the descriptive data on the study population.

Study population
Participants were recruited between November 2001 and May 2003 through participating
physical therapists and local newspapers. Dutch-speaking patients with OA of the hip or knee
(based on the criteria of the American College of Rheumatology25,26) aged between 50 and 80
years who experienced diminished physical function were included in the study24. Participants
who completed both baseline and follow-up (week 13) measurements were eligible for
inclusion in the present psychometric evaluation.

Measurements

Demographic and clinical data

Demographic and clinical data , including age, gender, duration of symptoms, OA location and
OA grade, according to Kellgren & Lawrence27, were collected from participating patients.

Physical function

Dutch MACTAR - The objective of this interview-based measurement tool is to evaluate changes
in patient-specific physical function over time. It comprises two parts. The baseline interview
starts with a transitional part. In this part, a trained interviewer asks the patient to identify up
to ten activities in which he / she experiences difficulties because of OA, such as activities in
domestic care, professional life and social interaction. The identified activities are ranked by
the patient from 1 to 10 in order of importance: ‘1’ for the activity the patient most wishes to
be able to do without pain or discomfort due to OA, ‘2’ for the next most important activity

MACTAR |113
and so on. The top-five prioritized activities are evaluated at follow-up. The second part of
the MACTAR (status) collects information on health status. Perceived overall health, as well
as psychological, emotional and social well-being is measured by five questions (Likert type
rating scale); when a question obtains a less than optimal score, a follow-up question probes
whether this is due to OA.
At the follow-up interview (week 13), changes in physical function are investigated. Patients
evaluate progress on their five most important activities as indicated in the transitional part of
the baseline interview, by evaluating each activity as ‘less of a problem’ (3 points), ‘the same’
(2 points) or ‘more of a problem’ (1 point). Patients also rate the perceived change in their
OA on a 7-point Likert scale. The status part reassesses patients’ health status.

It is quite difficult to allocate a total score to the MACTAR tool, because each part measures
different domains. Moreover, the transitional part and the status part employ different scoring
methods. While the transitional part measures change in physical function between baseline
and follow-up, the status part investigates current health status, at both baseline and follow-
up. The scoring method is presented in Table 1. Because of the differences between the
transitional and status parts, scores were not added together, but presented separately. The
MACTAR was translated into Dutch by Verhoeven et al. and validated in a population with
rheumatoid arthritis21.

TABLE 1. Scoring method of the McMaster Toronto Arthritis patient preference questionnaire
Transitional part Status part
Baseline - Min: 5 (poor health status)
Max: 25 (good health status)
Follow-up Min: 6 (maximum deterioration) Min: 5 (poor health status)
Max: 22 (maximum improvement) Max: 25 (good health status)
Change score The same as the follow-up score Min: -20 (maximum deterioration)
Max: +20 (maximum improvement)

Dutch WOMAC - The physical function subscale of the WOMAC contains 17 items which
represent common activities in daily living14,28. Patients are asked how much difficulty they have
had performing the activities mentioned. Each item is scored on a categorical scale, from ‘no
difficulty’ (score 0) to ‘extreme difficulty’ (score 4). The total score varies from 0 (no
difficulties) to 68 (extreme difficulties). Change scores on the WOMAC physical function
subscale can vary between -68 (maximum improvement) to +68 (maximum deterioration).
The WOMAC has been shown to be reliable and valid in patients with OA of the hip or
knee14,28 and the Dutch WOMAC permits valid international Dutch-English comparisons after
differential item functioning28.

Dutch SF-36 - The SF-36 investigates quality of life15-17,29. It comprises eight subscales, three of
which were used in the present validation study: physical functioning, role-physical and general
health. Scores on each subscale range from 0 to 100; higher scores reflect better health status.
The SF-36 has been validated for patients with various diagnoses, including those with OA30;

114
the Dutch language version has proved to be practical, reliable and valid for use in general
population surveys29.

Self-perceived change - At the follow-up evaluation, self-perceived change in physical function

was assessed by a ‘Patient Global Assessment’ (PGA) score. Patients were asked to rate their
overall perception of improvement since the start of the intervention on a scale ranging from
‘1’ (vastly deteriorated) to ‘8’ (completely recovered). PGA scores provide reliable
assessments of health transition in people with musculoskeletal disorders31.

Statistical analyses
Descriptive statistics were applied to describe the study population. PGA ratings were
dichotomized as ‘improved’ (PGA score 5, 6, 7 or 8) versus ‘not improved’ (PGA score 1, 2,
3 or 4). For continuous data, independent t-tests were used to calculate differences at baseline
between those patients who improved and those who did not. For categorical data, Mann-
Whitney U tests were used to compare between groups.

Content validity
Content validity examines the extent to which the domain in question is comprehensively
represented by the items in the questionnaire32,33. To determine whether the items in the
MACTAR refer to relevant aspects of the construct and are relevant to the purpose of the
instrument, the impaired activities mentioned by patients were compared with items on the
WOMAC and the SF-36 physical functioning subscale34.

Construct validity
There is currently no gold standard for attributes such as disability and functional status35-38.
Therefore, construct validity rather than criterion validity was assessed. Construct validity
refers to the extent to which scores on a particular instrument relate to other assessment
tools in a manner that is consistent with theoretically derived hypotheses39.
To investigate the construct validity of the MACTAR in patients with OA, change scores on
the transitional part of the MACTAR were correlated with change scores on both the
WOMAC and the SF-36 physical function subscales, as well as the SF-36 role-physical subscale.
Furthermore, follow-up scores on the status part of the MACTAR were correlated with
follow-up scores on the SF-36 general health subscale. For normally distributed data, Pearson
correlation coefficients (r) were used to express these correlations40. Spearman’s rank
correlation coefficients (rs) were applied when data were not distributed normally.
The following hypotheses were tested: 1. The change score on the physical function subscale
of the WOMAC is negatively correlated (r(s) ≤-.5)41(p<.05) with the change score on the
transitional part of the MACTAR; the correlation was expected to be negative, because the
WOMAC and the MACTAR use reverse scales. 2. Change scores on the physical functioning
and role-physical subscales of the SF-36 are positively correlated (r(s) ≥.5)41 (p<.05) with the
change scores on the transitional part of the MACTAR, and 3. Follow-up scores on the general

MACTAR |115
health subscale of the SF-36 are positively correlated (r(s) ≥.4)41 (p<.05) with the follow-up
scores on the status part of the MACTAR.

Responsiveness
Responsiveness can be assessed in many different ways. However, one can distinguish two
definition groups42. The first group describes responsiveness as ‘the ability to detect clinically
important change’42,43. In this group, an instrument is indicated as high responsive if it is able
to distinguish real change from measurement error. Hereby, responsiveness is calculated as
the magnitude of a treatment effect in which the Standardized Response Mean (SRM) and
Effect Size (ES) could be very useful42,43. The second group defines responsiveness ‘as the ability
to detect changes over time in the construct to be measured’34,42. In this case, responsiveness
is independent from any treatment effect and is interpreted as longitudinal validity. It should
be assessed in analogy to construct validity34. Therefore, predefined hypotheses concerning
change scores on the MACTAR, WOMAC and SF-36 in relation to PGA-scores were tested.
In the case of normally distributed change scores, parametric statistics were applied; non-
parametric variants were applied for data that were not distributed normally. It was
hypothesized that 1. The correlation between change scores on the MACTAR (transitional
part) and the PGA will be better than that between change scores on the PGA and the
WOMAC physical function subscale, the SF-36 physical functioning subscale and the role-
physical subscale respectively (p<.05). 2. Change scores on the MACTAR (transitional part)
for patients who have improved according to PGA will differ significantly (p<.05) from change
scores for those who have not improved according to PGA.

Second, responsiveness was determined by plotting a Receiver Operating Characteristics

(ROC) curve. The first step in its construction was to calculate sensitivity and specificity
statistics for MACTAR change scores in patients identified as improved (PGA score > 4) and
patients identified as non-improved (PGA score ≤ 4). Next, the true positive rate (sensitivity)
was plotted in functions of the false positive rate (1-specificity) for different cut-off points.
The best possible cut-off point would yield a point in the upper left corner of the ROC space,
representing 100% sensitivity, 100% specificity and an ‘Area Under the Curve’ (AUC) of 1.0.
An instrument is indicated as ‘highly responsive’ if the AUC is >.90, ‘moderately responsive’
where the AUC is between .70 and .90 and ‘lowly responsive’ if the AUC is between .50 and
.7044.

All analyses were performed using PASW Statistics 18.0. If patients were unable to identify at
least five impaired activities on the transitional part of the MACTAR, missing activity scores
were filled with a score indicating a ‘no change situation’ (2 points); data from patients who
mentioned less than three impaired activities were excluded from the responsiveness analyses.
Furthermore, in cases of just one missing follow-up item for the status part of the MACTAR,
the score obtained on the equivalent question in the baseline interview was also used for the
follow-up.

116
Following the initial analyses, a sensitivity analysis was performed on various cut-off points of
the dichotomized PGA-score, the aim of which was to determine whether the chosen cut-off
point was the optimal point to dichotomize.

Results

Study population
A total of 192 patients participated in both the baseline and the first follow-up assessment and
were included for content and construct validity analyses. The median PGA score of these 148
females and 44 males was ‘5’, representing ‘slightly improved’. Baseline characteristics of the
study population are presented in Table 2.

TABLE 2. Baseline characteristics total population and subgroups #

Total population Improved group Non-improved
Characteristic p-value
(n=192) (n=144)† group (n=48)†
Female sex 148 (77) 118 (82) 30 (63) .01
Age, mean (± SD) years 64.7 (7.9) 65.5 (7.9) 62.3 (7.6) .02
Location of OA .58
Knee 126 (66) 96 (67) 30 (63)
Hip 49 (26) 36 (25) 13 (27)
Both 17 (9) 12 (8) 5 (10)
Radiologic severity of OA* .62
No features / doubtful 44 (23) 35 (24) 9 (19)
Minimal / moderate 91 (47) 67 (47) 24 (50)
Severe or prothese 6 (3) 5 (4) 1 (2)
Unknown 51 (27) 37 (26) 14 (29)
Duration of symptoms .55
<1 year 44 (23) 36 (25) 8 (17)
1-5 years 71 (37) 52 (36) 19 (40)
>5 years 75 (39) 55 (38) 20 (42)
# Values are represented as number (percentage), unless otherwise indicated; * Kellgren and Lawrence score27,47; †
Improved group = Patient Global Assessment score (PGA) > 4, non-improved group PGA ≤ 4; SD = standard deviation;
OA = osteoarthritis;

Outcomes
Table 3 shows absolute scores on the MACTAR, WOMAC, SF-36 and PGA at baseline and
follow-up for both the total population and improved / non-improved patients. At baseline,
there were no differences on any of the outcome measures between patients who indicated
that they had improved and patients who indicated that they had not improved. At follow-up,
MACTAR scores (both transitional and status parts), WOMAC physical function scores and
SF-36 physical functioning scores differed significantly between improved and non-improved
patients. The measurement variation was higher in the WOMAC and SF-36 compared with
the MACTAR, at both baseline and follow-up (Table 3).

MACTAR |117
 TABLE 3. Baseline, follow-up and changes scores on the outcome measures (mean ± sd)
Outcome measure Total population Improved group Non-improved p-value
(n=192) (n=144) group (n=48)
MACTAR – transitional
Follow-up 16.6 (2.8) 17.6 (2.0) 13.6 (2.6) <.01
MACTAR – status
Baseline 19.3 (4.0) 19.3 (4.1) 19.1 (3.9) .76
Follow-up 20.5 (3.6) 20.8 (3.4) 19.4 (3.8) .03
Change score 1.2 (4.3) 1.5 (4.2) 0.1 (4.5) .05
WOMAC – physical function
Baseline 28.6 (11.0) 28.3 (10.9) 29.8 (11.3) .42
Follow-up 23.0 (11.4) 21.3 (10.6) 27.9 (12.1) <.01
Change score -5.6 (8.9) -6.8 (9.0) -1.9 (7.7) <.01
SF-36 – physical function
Baseline 48.7 (20.1) 49.1 (19.5) 47.6 (22.1) .68
Follow-up 56.0 (21.7) 58.0 (21.2) 50.1 (22.1) .03
Change score 7.6 (17.6) 9.4 (18.5) 2.0 (12.9) .01
SF-36 – role physical subscale
Baseline 42.5 (41.1) 41.2 (41.0) 46.1 (41.5) .48
Follow-up 55.9 (42.0) 59.1 (41.5) 46.1 (42.5) .07
Change score 14.7 (46.0) 19.2 (48.3) 1.1 (35.3) .02
SF-36 – general health
Baseline 50.8 (19.4) 51.8 (19.7) 47.9 (18.8) .24
Follow-up 48.9 (16.5) 50.0 (17.2) 45.7 (14.1) .13
Change score -2.1 (17.3) -2.1 (17.1) -2.1 (17.9) >.99
PGA (median (range)) 5 (6) 6 (3) 4 (2) -
MACTAR = McMaster Toronto Arthritis patient preference questionnaire; WOMAC = Western Ontario and McMaster
Universities Osteoarthritis Index; SF-36 = Medical Outcome Survey Short Form 36; PGA = Patient Global Assessment

Content validity
The study population (n=192) identified a total of 894 impaired activities, which indicates a
mean of 4.6 impaired activities per patient. Seventy-one patients (37%) were unable to identify
at least five impaired activities; one patient could name only one impaired activity; 10 patients
identified two impaired activities; and 33 patients were able to name a maximum of three
impaired activities. ‘Walking’ was most frequently mentioned as the most impaired activity
(43%). Overall, seventy-two percent of the identified impaired activities comprised activities
in the category of ‘mobility’. Table 4 summarizes all the activities mentioned, ranked by
category.
All items from both the WOMAC and the SF-36 physical function subscales were represented
in the impaired activities list based on the MACTAR questionnaire. However, 27% of the
activities mentioned by patients during the MACTAR interview were not represented in the
WOMAC, and 41% were not represented in the SF-36. Eleven percent of the impaired
activities mentioned were not covered either by items of the WOMAC or the SF-36: examples
of which include gardening and activities related to professional life.

118
 TABLE 4. Patient-mentioned impaired activities on the MACTAR-questionnaire
Category Mentioned as most Mentioned in total
Activity impaired activity n (%) n (%)
Housekeeping activities †# 4 (2.3) 70 (8.5)
Vacuum cleaning, mopping, washing windows or dishes, lifting buckets,
etcetera
Leisure activities 19 (11.0) 102 (12.3)
Gardening 5 (2.9) 30 (3.6)
Remaining leisure activities (cultural activities, shopping †) 4 (2.3) 17 (2.2)
Sports (jogging, tennis, swimming, fitness, riding a horse, dancing)# 10 (5.8) 54 (6.5)
Mobility 134 (77.9) 591 (71.5)
Bicycling (including getting up/off) # 19 (11.0) 81 (9.8)
Climbing stairs †# 13 (7.6) 106 (12.8)
Driving (including getting in/out the car) † 4 (2.3) 52 (6.3)
Getting up from the floor/ a chair, getting out of bed † 10 (5.8) 53 (6.4)
Inability to stand for long † 1 (0.6) 54 (6.5)
Kneeling down, bending over, reaching down †# 12 (7.0) 77 (9.3)
Remaining mobility activities 1 (0.6) 39 (4.7)
Walking †# 74 (43.0) 129 (15.6)
Professional activities 1 (0.6) 13 (1.6)
Remaining activities 1 (0.6) 2 (0.2)
Self-care activities 7 (4.1) 82 (9.9)
Dressing (socks, underwear, trousers) † 6 (3.5) 74 (9)
Remaining self-care activities # 1 (0.6) 8 (1.0)
Sexuality 1 (0.6) 5 (0.6)
Sleeping and resting, including turning around in bed † 5 (2.9) 18 (2.2)
Social roles - 11 (1.3)
TOTAL 172 (100) 894 (100)
MACTAR = McMaster Toronto Arthritis Patient Preference Disability Questionnaire; † item is represented in Western Ontario
and McMaster Universities Osteoarthritis Index; # item is represented in Medical Outcome Survey Short Form 36

Construct validity
Correlations (rs) between change scores on the transitional part of the MACTAR and change
scores on the physical function subscales of the WOMAC and the SF-36 were -.40 (p<.01)
and .27 (p<.01) respectively. Change scores on the transitional part of the MACTAR and the
role-physical subscale of the SF-36 were also moderately correlated (rs=.27 (p<.01)).
Spearman’s rs between follow-up score of the MACTAR status part and the general health
subscale of the SF-36 was .44 (p<.01) (Table 5).

MACTAR |119
 TABLE 5. Correlation matrix change scores / follow-up score outcome measures (n=189) #

SF-36 – physical functioning

WOMAC – physical function

SF-36 – general health

MACTAR – transitional

SF-36 – role physical

MACTAR – status
Change scores
MACTAR- transitional part 1.00 -.40 .27 .27
WOMAC- physical function subscale 1.00 -.36 -.30
SF36- physical functioning subscale 1.00 .32
SF36- role physical subscale 1.00
Follow-up scores
MACTAR- status part 1.00 .44
SF36- general health subscale 1.00
# Spearman’s rank correlation coefficient; All correlations were significant at the .01 level
MACTAR = McMaster Toronto Arthritis Patient Preference Disability Questionnaire; WOMAC = Western Ontario and
McMaster Universities Osteoarthritis Index; SF36 = Medical Outcome Survey Short Form 36

Responsiveness
Data from 133 patients (82% female, mean age 64.0 ± 8.1 years) were used in the
responsiveness analyses. Seventy-seven percent of these patients indicated that they had
improved following treatment (PGA score >4) while 23% reported that they had not improved
(PGA score ≤4). With the exception of age, the improved and non-improved group had similar
baseline characteristics. Absolute change scores on physical function outcomes are presented
in Table 6. Change scores for patients who indicated that they had improved differed
significantly from patients who indicated that they had not improved on all outcome measures
(Table 6).
Correlations between change scores on the physical function outcomes and the PGA score
are also presented in Table 6. As hypothesized, change scores on the MACTAR correlate
better with PGA (rs = .69) than change scores on the WOMAC (rs = -.39) and SF-36 (rs = .26
and .25 respectively) (Table 6).

120
 TABLE 6. Change scores on physical function measures and correlation coefficients (rs) with PGA
Absolute change score (mean (95% CI)) Correlation with PGA score**
MACTAR - transitional part*
Total population (n=133) .69
Improved group (n=102) 17.7 (17.3 – 18.2)
Non-improved group (n=31) 13.7 (12.7 – 14.7)
WOMAC - physical function*
Total population -6.1 (-7.7 – -4.5) -.39
Improved group -7.2 (-9.0 – -5.3)
Non-improved group -2.7 (-5.8 – 0.4)
SF-36 - physical functioning*
Total population 8.3 (5.2 – 11.3) .26
Improved group 9.7 (6.0 – 13.5)
Non-improved group 3.5 (-0.6 – 7.6)
SF-36 - role-physical*
Total population 14.9 (6.8 – 22.9) .25
Improved group 19.7 (10.1 – 29.2)
Non-improved group -0.9 (-14.2 – 12.4)
* Significant difference between improved and non-improved group (p<.05); ** Significant at p<.05 level, unless
otherwise indicated. PGA = Patient Global Assessment; MACTAR = McMaster Toronto Arthritis Patient Preference Disability
Questionnaire; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-36 = Medical Outcome
Survey Short Form 36

Figure 1 presents a ROC-curve of the change scores for the MACTAR (transitional part), in
which the sensitivity of the MACTAR amounted to its 1-specificity. The AUC was .90 (95% CI
.89 - .96) with a standard error of .03.

FIGURE 1. Receiver Operating Characteristics curve of the change score on the McMaster Toronto Arthritis
Patient Preference questionnaire (MACTAR) (transitional part); AUC = Area Under the Curve

The sensitivity analysis showed that the cut-off point for the PGA-score dichotomization (>4),
was chosen correctly. Higher and lower cut-off points resulted in less optimal responsiveness
values.

MACTAR |121
Discussion

The aim of the present study was to investigate the content validity, the construct validity and
the responsiveness of the MACTAR in patients with OA of the hip or knee.

The content validity of the MACTAR seems to be good. Specifically, the majority of the
impaired activities identified correlate with items on the WOMAC and / or SF-36, which also
aim to assess physical function. However, the MACTAR fits better with the WOMAC
questionnaire than the SF-36. This is not surprising, since the WOMAC is aimed specifically at
patients with osteoarthritis, whereas the SF-36 is of a more general nature. Data for 11% of
the activities identified are gathered only by the MACTAR and are not represented in either
the WOMAC or the SF-36. These comprised activities in the areas of leisure, professional life
and social interaction. Indeed, participation in these fields varies widely between individuals.
Disease-specific and general instruments do not take account of individual limitations, but
patient-specific measures such as the MACTAR allow clinicians to evaluate physical functioning
at individual level.
The majority of the activities identified by the MACTAR questionnaire comprised activities in
the ‘mobility’ domain, which corresponds with the majority of activities in daily life. Recent
validation studies on the MACTAR questionnaire in patients with chronic low back pain and
rheumatoid arthritis showed comparable results21,23. The most frequently mentioned impaired
activity in patients with chronic low back pain was ‘taking part in sports activities23; in patients
with hip/knee OA ‘walking’ was the most commonly cited impaired activity.

Although the content validity of the MACTAR seems to be good in patients with OA, the
construct validity is less convincing. Moderate associations between the transitional part of
the MACTAR and presumed comparable outcomes (rs≤.40) might be explained by an
unbalanced distribution of impaired activities across the various activity-categories. Specifically,
the ‘mobility’ category comprised almost 72% of all reported impaired activities, whereas the
mobility domain in the WOMAC contains only 58% and in the SF-36, 60% of the total
questionnaire. Thus, the transitional part of the MACTAR covers one specific part of the
physical function domain extensively, whereas disease-specific and general tools account for a
broader spectrum of this domain. Another explanation for the moderate construct validity
could be the narrow variance around the mean on the MACTAR, compared with a wide
variance in WOMAC and SF-36 scores. The variance is caused by patients who tend to assign
the same disability score to very different impaired activities. The difference in variance
impedes a comparison between a patient-specific instrument on the one hand and a disease
specific / generic instrument on the other.
As hypothesized, the status part of the MACTAR was moderately correlated with the general
health subscale of the SF-36 (rs=.44). Previous studies identified comparable correlation
coefficients between the MACTAR and other physical function measures. Sanchez et al. found
a correlation (rs) of .40 between the MACTAR and the Quebec Back Pain Disability Scale37 in

122
patients with chronic low back pain23 and a correlation (rs) of .38 (p=.002) was found between
the MACTAR and the Health Assessment Questionnaire (HAQ)45 in patients with systemic
sclerosis22. Verhoeven et al.21 showed a correlation coefficient (r) of .73 (p<.0003) between
the MACTAR and the HAQ in patients with rheumatoid arthritis.

The MACTAR was developed to evaluate patient-specific physical function over time. With
this goal in mind, responsiveness is the most important psychometric property. For that
reason, we evaluated the responsiveness of the questionnaire. As hypothesized, change scores
for the MACTAR correlated better with the PGA score than change scores on the WOMAC
and SF-36 do, leading to the conclusion that the MACTAR is better able to detect changes
over time in patients with hip / knee OA than the WOMAC or SF-36. It has also been
demonstrated that the MACTAR is capable of distinguishing patients who reported an
improvement from those patients who reported no improvement. An AUC of .90 confirms
the high responsiveness of the MACTAR in patients with hip / knee OA. Verhoeven et al.21
also investigated the responsiveness of the MACTAR, concluding that it showed a high degree
of responsiveness, based on a SRM of 3.5. However, a SRM is not an appropriate measure of
assessing responsiveness as the ability to detect changes over time in the construct to be
measured, but can be used to detect clinically important change34;42.

One limitation of our study is the use of PGA scores as an external criterion to distinguish
patients who improved from those who did not. Guyatt et al.46 showed that patients are unduly
influenced by their current health status when they complete transition ratings such as PGA
scores. Moreover, the reproducibility of a single-item transitional scale is probably lower than
for a more extended measurement tool47. Finally, ‘a little better’ is not, as a matter of course,
equivalent to an ‘important change48. However, better external criteria to discriminate
between improved and non-improved patients have not yet been elaborated.
Although the MACTAR appears to have some advantages over the WOMAC and the SF-36
in assessing physical function in individual patients, it also has some limitations. The need for a
trained interviewer to apply the MACTAR, as well as its complicated scoring method may
reduce the likelihood that the MACTAR will become the instrument of first choice in clinical
practice. Furthermore, patient-specific measures, including the MACTAR, do not take account
of shifts in patient priorities that can occur over time in cases of change in disease status.
Therefore, further studies should take account of the application of patient specific measures
at long-term follow-up.

In conclusion, our results suggest that the MACTAR exhibits moderate construct validity and
good responsiveness in a population of patients with OA of the hip or knee. Furthermore,
the MACTAR is potentially better able to detect changes over time in activities that are
important to individual patients compared to other tools measuring physical function
(WOMAC, SF-36). Therefore, clinicians could use the MACTAR to evaluate clinically relevant
changes over time in patient-specific physical functioning.

MACTAR |123
References 9. Fransen M, McConnell S. Exercise for
osteoarthritis of the knee. Cochrane Database
Sys Rev 2008; 8(4):CD004376.
1. Poos M, Gommer A. Hoe vaak komt artrose
voor en hoeveel mensen sterven eraan? In: 10. Fransen M, McConnell S, Hernandez-Molina G,
Volksgezondheid Toekomst Verkenning, Reichenbach S. Exercise for osteoarthritis of
Nationaal Kompas Volksgezondheid. RIVM, the hip. Cochrane Database Sys Rev 2009;
editor. 2009. Bilthoven. 8(3):CD007912.

2. Bellamy N, Kirwan J, Boers M, et al. 11. Tan J, Horn S. Practical manual of physical
Recommendations for a core set of outcome medicine and rehabilitation: diagnostics,
measures for future phase III trials in knee, hip, therapeutics, and basis problems. Second
and hand osteoarthritis. Consensus Edition ed. St. Louis: Elsevier, 2006.
development at OMERACT III. J Rheumatol
1997; 24:799-802. 12. Terwee C, Mokkink L, Steultjens M, Dekker J.
Performance-based methods for measering the
3. Dekker J, Boot B, van der Woude L, Bijlsma J. physcial function of patients with osteoarthritis
Pain and disability in osteoarthritis: a review of of the hip or knee: a systematic review of
biobehavioural mechanisms. J Behav Med 1992; measurement properties. Rheumatology 2006;
15:189-214. 45:890-902.

4. Felson D, Lawrence R, Dieppe P, Hirsch R, 13. Veenhof C, Bijlsma J, van den Ende C, van Dijk
Helminck C, Jordan J et al. Osteoarthrtitis: G, Pisters M, Dekker J. Psychometric
new insights. Part 1. The disease and its risk evaluation of osteoarthritis questionnaires: a
factors. Ann Intern Med 2000; 133:635-646. systematic review of the literature. Arth
Rheum / Ar C Res 2006; 55(3, June 15):480-
5. McAlindon T, Cooper C, Kirwan J, Dieppe P. 492.
Determinants of disability in osteoarthritis of
the knee. Ann Rheum Dis 1993; 52:258-262. 14. Bellamy N, Buchanan W, Goldsmith C,
Campbell J, Stitt L. Validation study of
6. Jordan K, Arden N, Doberty M, Bannwarth B, WOMAC: a health status instrument for
Bijlsma J. EULAR recommendations 2003: an measuring clinically important patient relevant
evidence based approach to the management outcomes to antirheumatic drug therapy in
of knee osteoarthritis: report of a task force of patients with osteoarthritis of the hip or knee.
the Standing Committee for International J Rheumatol 1988; 15:1833-1840.
Clinical Studies Including Therapeutic Trials
(ESCICIT). Ann Rheum Dis 2003; 62:1145- 15. McHorney C, Ware J, Lu R, Donald
1155. Sherbourne C. The MOS 36-item short-form
health survey (SF-36): III. Tests of data-quality,
7. Vogels E, Hendriks H, van Baar M, Dekker J, scaling assumptions, and reliability across
Hopman-Rock M, Oostendorp R et al. KNGF- diverse patient groups. Med Care 1994; 32:40-
richtlijn artrose heup-knie. 2001. Amersfoort, 66.
KNGF.
16. McHorney C, Tarlov A. Individual-patient
8. Brandt K. The importance of nonpharmalogic monitoring in clinical practice: are available
approaches in management of osteoarthritis. health status surveys adequate? Qual Life Res
Am J Med 1998; 105:39S-44s. 1995; 4:293-307.

17. Ware J, Sherbourne C. The MOS 36-Item

Short-Form Health Survey (SF-36), I:

124
 conceptual framework and item selection. Med the knee. Arthritis Rheum 1986; 29:1039-
Care 1992; 30:473-483. 1049.

18. Seror R, Tubach F, Baron G, Guillemin F, 26. Altman R, Alarcon G, Appelrouth D, Bloch D,
Ravaud P. Measure of function in rheumatoid Borenstein D, Brandt K et al. The American
arthritis: individualized or classical scales? Ann College of Rheumatology criteria for the
Rheum Dis 2010; 69(1):97-101. classification and reporting of osteoarthritis of
the hip. Arthritis Rheum 1991; 34:505-514.
19. Jolles BM, Buchbinder R, Beaton DE. A study
compared nine patient-specific indices for 27. Kellgren J, Lawrence J. Radiological assessment
musculoskeletal disorders. J Clin Epidemiol of osteo-arthritis. Ann Rheum Dis 157;
2005; 58(8):791-801. 15(4):494-502.

20. Tugwell P, Bombardier C, Buchanan W, 28. Roorda L, Jones C, Waltz M, Lankhorst G,

Goldsmith C, Grace E, Hanna B. The Bouter L, van der Eijken J et al. Satisfactory
MACTAR Patient Preference Disability cross cultural equivalence of the Dutch
Questionnaire--an individualized functional WOMAC in patients with hip osteoarthritis
priority approach for assessing improvement waiting for arthroplasty. Ann Rheum Dis 2004;
in physical disability in clinical trials in 63:36-42.
rheumatoid arthritis. J Rheumatol 1987;
14(3):446-451. 29. Aaronson N, Muller M, Cohen P, Essink-Bot M,
Fekkes M, Sandermann R et al. Translation,
21. Verhoeven A, Boers M, van der Lindern S. Validation, and Norming of the Dutch
Validity of the MACTAR Questionnaire as Language Version of the SF-36 Health Survey
Functional Index in a Rheumatoid Arthritis in Community and Chronic Disease
Clinical Trial. J Rheumatol 2000; 27:2801-2809. Populations. J Clin Epidemiol 2011;
51(11):1055-1068.
22. Mouthon L, Rannou F, Bérezné A, Pagnoux C,
Guilpain P, Goldwasser F et al. Patient 30. Kosinski M, Keller S, Hatoum H, Kong S, Ware
Preference Disability Questionnaire in JE Jr. The SF-36 Health Survey as a generic
Systemic Sclerosis: A Cross-Sectional Survey. outcome measure in clinical trials of patients
Arth Rheum / Ar C Res 2008; 59(7):968-973. with osteoarthritis and rheumatoid arthritis:
tests of data quality, scaling assumptions and
23. Sanchez K, Papelard A, Nguyen C, Jousse M, score reliability. Med Care 1999; 37(5
Rannou F, Revel M et al. Patient-Preference (Suppl)):MS 10-22.
Disability Assessment for Disabling Chronic
Low Back Pain. Spine 2009; 34(10):1052-1059. 31. Kamper S, Ostelo R, Knol DC, Maher C, de
Vet H, Hancock M. Global Perceived Effect
24. Veenhof C, Koke J, Dekker J, Oostendorp R, scales provided reliable assessments of health
Bijlsma J, van Tulder M et al. Effectiveness of transition in people with musculoskeletal
behavioral graded activity in patients with disorders, but ratings are strongly influenced
osteoarthritis of the hip and/or knee: a by current status. J Clin Epidemiol 2010;
randomized clinical trial. Arthritis Rheum 63(7):760-766.
2006; 55(6):925-934.
32. Guyatt G, Feeny D, Patrick D. Measuring
25. Altman R, Asch E, Bloch D, Bole G, Borenstein health related quality of life. Ann Intern Med
D, Brandt K et al. Development of criteria for 1993; 118:622-629.
the classification and reporting of
osteoarthritis: classification of osteoarthritis of 33. Mokkink L, Terwee C, Patrick D, Alonso J,
Stratford P, Knol D et al. International

MACTAR |125
 consensus on taxonomy, terminology,and 41. Portney L, Watkins P. Foundations of Clinical
definitions of measurement properties for Research. Second Edition ed. New Jersey:
health-related patient-reported outcomes: Prentice-Hall Inc., 2000.
results of the COSMIN study. J Clin Epidemiol
2010; 63(7):737-745. 42. Terwee C, Dekker F, Wiersinga W, Prummel
M, Bossuyt P. On assessing responsiveness of
34. Mokkink L, Terwee C, Knol D, Stratford P, health-related quality of life instruments:
Alonso J, Patrick D et al. The COSMIN Guidelines for instrument evaluation. Qual Life
checklist for evaluating the methodological Res 2003; 12:349-362.
quality of studies on measurement properties:
A clarification of its content. BMC Medical 43. Streiner D, Norman G. Health Measurement
Research Methodology 2010; 10(22). Scales; a practical guide to their development
and use. Fourth edition ed. Oxford: Oxford
35. Beurskens AJ, de Vet HC, Koke AJ. University Press, 2008.
Responsiveness of functional status in low back
pain: a comparison of different instruments. 44. Hanley J, McNeil B. The meaning and use of the
Pain 1996; 65(1):71-76. area under receiver operating charcteristic
(ROC) curve. Radiology 1982; 143:29-39.
36. Chatman AB, Hyams SP, Neel JM, Binkley JM,
Stratford PW, Schomberg A et al. The patient- 45. Fries J, Spitz P, Young D. Dimensions of health
specific functional scale: Measurement outcome: the Health Assessment
properties in patients with knee dysfunction. Questionnaire, Disability and pain scales. J
Physical Therapy 1997; 77(8):820-829. Rheumatol 1982; 9:727-735.

37. Kopec J, Esdaille J, Abrahamowicz M, et al. The 46. Guyatt G, Norman G, Juniper E, Griffith LE. A
Quebec Backpain Disability Scale. critical look at transition ratings. J Clin
Measurement properties. Spine 1995; 20:341- Epidemiol 2002; Sep; 55(9):900-908.
352.
47. Norman G, Stratford P, Regehr G.
38. Stratford P, Gill C, Westaway M, Binkley J. Methodological problems in retrospective
Assessing disability and change on individual computation of responsiveness to change: the
patients: a report of a patient specific measure. lessons of Cronbach. J Clin Epidemiol 1997;
Physiotherapy Canada 1995; 47(4):258-263. 50:869-879.

39. Kirshner B, Guyatt B. A methodological 48. Terwee C, Roorda L, Dekker J, Bierma-

framework for assessing health indices. J Zeinstra S, Peat G, Jordan K et al. Mind the
Chronic Dis 1985; 38:27-36. MIC: large variations among populations and
methods. J Clin Epidemiol 2010; 63:524-534.
40. Nunnaly JJ. Introduction in psychological
measurment. New York: 1970.

126
Appendix 1: McMaster Toronto Arthritis patient preference questionnaire (MACTAR)

Baseline interview

Interviewers note:
Read the questions and response categories provided. Tick the response given.

1a. How would you say your overall health has been during the last 2 weeks?
You think of it as 3 very good
2 pretty good
1 not too good

2. Osteoarthritis may cause restrictions in several areas of your daily life. For different
people the impact of osteoarthritis is also different. We will ask you to name activities in
which you experience difficulties because of your osteoarthritis. What matters here, is what
your personal experience has been. Please, think of what became a problem, now that you
have osteoarthritis.

Interviewers note:
In order to elicit a comprehensive list of activities: First, give the patient opportunity to react
spontaneously. Then read the probes. Record the exact phrases of the patient on the line
hereunder.

To support you in naming any problems caused by osteoarthritis, I will read you a number of areas
of your daily life that might be affected.

Does your osteoarthritis limit:

 any (other) activities around the house such as cooking, housework etc.?
 any activity related to dressing such as buttoning, pulling a sweater over your heat etc.?
 any (other) activities at your work (outside the home), drive a care or other transportation?
 any (other) leisure activities. Either sport such as bowling, swimming, golf, etc. or non-sports
such as needlework, woodwork etc.?
 any (other) social activities. Such as visiting, plying cards, going to church, etc.?
 sexuality?

Are there changes in the relationship with your family?

If you live together with a hushand / wife / partner, are there changes in the relation with him / her?
If you have children living at home, are there changes in your relationship with them?

…………………………………………………………………………………………
The line above is printed 10 times

MACTAR |127
Interviewers note:
To rank the activities in order of importance to the patient, ask the following questions:

Which of these activities would you most like to be able to do without pain or discomfort of our
osteoarthritis?
Show and read the list to the patient. Write ‘1’ next to the activity the patient chose.
After (read activity 1), which activity would you next most like to be able to do without pain or
discomfort of your osteoarthritis?
Show and read the list to the patient, with exception of the activity with priority 1.
Write ‘2’ next to the activity the patient chose.
After (read activity 1 and 2), which activity would you next most like to be able to do without pain or
discomfort of your osteoarthritis?
Show and read the list to the patient, with exception of the activity with priority 1 and 2.
Write ‘3’ next to the activity the patient chose.

Continue like this, until all activities are ranked. The 5 with the highest priority will return in
the follow-up interview.

3a. In general, how satisfying do you find the way you spend your life?
Over the last week you think of it as: 3 completely satisfying →go to Q4a
2 pretty satisfying
1 not very satisfying

3b. Is your life not completely satisfying

because of your osteoarthritis? 0 yes 1 no

4a. How would you say your overall physical functioning has been?
Over the last week you think of it as: 5 good →go to Q5a
4 fair to good
3 fair
2 fair to poor
1 poor

4b. Is your physical function not as good as is might be

because of your osteoarthritis? 0 yes 1 no

5a. How would you say your overall social functioning has been?
Over the last week you think of it as: 5 good →go to Q6a
4 fair to good
3 fair
2 fair to poor
1 poor

5b. Is your social functioning not as good as is might be

because of your osteoarthritis? 0 yes 1 no

128
6a. How would you say your overall emotional functioning has been?
Over the last week you think of it as: 5 good →stop here
4 fair to good
3 fair
2 fair to poor
1 poor

6b. Is your emotional functioning not as good as is might be

because of your osteoarthritis? 0 yes 1 no

Follow-up interview

1a. How would you say your overall health has been during the last 2 weeks?
You think of it as 3 very good
2 pretty good
1 not too good

1b. Have you noticed any change in your osteoarthritis since we talked during the first interview?
 yes
 no → ‘no change’ at Q1d, go to Q2a

1c. Please describe how your osteoarthritis has changed?

…………………………………………………………………………………………
…………………………………………………………………………………………

1d. When you think of your osteoarthritis during the two weeks before the first interview,
how much better or worse overall has your osteoarthritis become?
7 a great deal better
6 moderately better
5 slightly better
4 no change
3 slightly worse
2 moderately worse
1 a great deal worse

You may remember the first time we spoke. You told me which activities were at that time problems
due to your arthritis. I will ask you again about the five most important.

2a. Since the first interview, have you noticed any change in your ability to (activity 1)?
3 less of a problem
2 the same
1 more of a problem

2b. Since the first interview, have you noticed any change in your ability to (activity 2)?
3 less of a problem
2 the same
1 more of a problem

MACTAR |129
2c. Since the first interview, have you noticed any change in your ability to (activity 3)?
3 less of a problem
2 the same
1 more of a problem

2d. Since the first interview, have you noticed any change in your ability to (activity 4)?
3 less of a problem
2 the same
1 more of a problem

2e. Since the first interview, have you noticed any change in your ability to (activity 5)?
3 less of a problem
2 the same
1 more of a problem

Questions 3a to 6b are equal to the equally numbered questions in the baseline

interview.

130
 7
General discussion
Personalized treatment is recognized as one of the solutions to face the complex and
continuously expanding healthcare demand of patients with (multiple) chronic conditions1. It
is one of the key elements of the renewed view on health, introduced by Huber et al.: patients
are challenged to manage their health themselves and to adapt to new situations2. Clinicians
are supposed to strengthen a patient’s functioning, flexibility, and self-control. Personalized
treatment is often presented as a ‘new phenomenon’ in healthcare. However, in 1997, Wright
et al. already presented a paper which was subtitled: ‘… asking patients what they want’ 3. This
paper demonstrates that personalized care has caught attention for 20 years and over.
Therefore, it could be expected that aspects of personalized care are already widespread used
in clinical practice.

In this thesis, we especially focused on personalized treatment in patients with osteoarthritis

(OA) of the hip and the knee. Due to its’ heterogeneity, OA would be particularly appropriate
for personalized treatment. The results of the studies captured in Chapter 2 to Chapter 6 of
this thesis show that personalized treatment is indeed widely applied in current clinical practice
of OA care. Nevertheless, there is still room for improvement. In this final chapter, we will
reflect on the results presented in the previous chapters and will provide recommendations
for clinical practice, policy and future research. We focus on two aspects of personalized care
specifically: timing of care and focus of care.

Timing of care

Clinicians are challenged to phase interventions over time and reduce inappropriate use of
advanced care4. Thereby, phasing treatment could offer benefits both in clinical perspective as
well as financial perspective. The recently developed Stepped-Care-Strategy (SCS) phases non-
surgical care in three steps5. The first step comprises the optimization of patients’ self-care in
primary care by education, lifestyle advices, and the prescription of acetaminophen. In step 2,
(topical) NSAIDs are added or a referral for allied health care (exercise therapy, dietary
therapy) will be considered (step 2). Finally, step 3 comprises the use of TENS and intra-
articular corticosteroid injections or a referral for interprofessional evaluation in secondary
care. Interventions belonging to subsequent steps should only be considered in patients with
persisting complaints, despite the use of all recommended interventions from the lower step
or steps. Although the SCS has not yet been widely implemented in clinical practice, Dutch
general practitioners already seem to phase their interventions over time. They apply less
advanced interventions, like lifestyle advices and the prescription of NSAIDs (step-1,2), more
often than for instance exercise therapy or a referral for Total Joint Replacement (TJR) (step-
2,3) (Chapter 2). Thus, their usual sequence of care seems to be mainly in accordance with
the SCS. However, appropriate timing of care could still be improved. This improvement
mainly concerns the appropriate timing of a referral to secondary care (Chapter 2 & 3).
Several results from this thesis endorse an overuse of orthopedic care and, in consequence,

General Discussion |133

an underuse of non-surgical care in patients with hip/knee OA. We will subsequently address
three results which illustrate that the timing of care is not appropriate yet: 1. the insufficient
use of all recommended stepped-care interventions, 2. the insufficient use of physical therapy,
and 3. the underutilization of non-surgical treatment in secondary care.

Insufficient use of all recommended interventions

According to the SCS, a patient should be referred to secondary care (step-3) in case of
persisting complaints despite the use of all recommended interventions from the lower step or
steps5. Our retrospective study on currently provided care showed that less than 3% of the
patients referred to secondary care did receive all step-1 and step-2 interventions of the SCS
(Chapter 2). Recently, Hofstede et al. confirmed the low use of all recommended
interventions in hip/knee OA6. In their study, an amount of 6% actually received all
recommended interventions of the SCS prior to surgery. In both studies, clinicians were
unaware of the SCS. A study by Smink et al., which has been performed after implementation
of the SCS, showed that the SCS potentially increases the use of recommended non-surgical
interventions7. In their study, 28% of the patients received all interventions from previous
steps prior to a referral to secondary care. Hence, still the majority of patients referred to
secondary care did not receive all interventions captured in step-1 and step-2. Probably, a
great deal of effort is requested to achieve nationwide implementation of the SCS.
However, it could also be questioned whether the SCS is currently feasible to apply in clinical
practice. By way of illustration, the SCS does not yet take into account the ‘Direct Access
Regulation’ to physical therapy8,9. Direct access allows patients to directly contact a physical
therapist (step-2) without a visit at a general practice (step-1). In consequence, patients using
direct access potentially will not receive adequate pain medication or appropriate lifestyle
advises (step-1) as they turn directly to a physical therapist (step-2). To fit better to current
clinical practice, Koenders et al. slightly modified the SCS, resulting in the ‘Stepped-Care
Osteoarthritis algorithm’10. The Stepped-Care Osteoarthritis algorithm indicates the use of
physical therapy (step-2) in case of completing all recommended step-1 interventions or in case
of severe limitations in physical functioning, operationalized by a score >4 on the
Algofunctional Index11. Patients with minor limitations in physical functioning who visit a
physical therapist by direct access, return to step-1 care performed by the GP10 (Koenders).
Further research should provide insight whether or not this modification on the SCS attributes
to an improved implementation of stepped-care in patients with hip/knee OA.

Insufficient use of physical therapy

A large body of evidence supports the effectiveness of exercise therapy regarding
improvement of functioning in patients with hip/knee OA12-14. Therefore, a referral to physical
therapy is included in step-2 of the SCS. Based on the subsequent steps of the SCS, we could
have expected a higher referral rate to physical therapists than to medical specialists in
secondary care. However, one of our studies indicated that in current primary healthcare,
referrals to physical therapists (step 2, 4%) lags behind compared to referrals to medical

134
specialists, like orthopedic surgeons and rheumatologists (step 3, 13%) (Chapter 2). The low
referral rate to physical therapy could be due to the introduction of the previously mentioned
Direct Access Regulation8,9 - this may have led to patients directly visiting the physical
therapist, without a referral by the GP. However, taking into account the self-referrers,
physical therapy still seems to be underused in current clinical practice. Maybe, financial issues
play a role as well. Until 2006, physical therapy due to hip/knee OA was fully reimbursed for
each Dutch inhabitant. Since 2006, reimbursement for physical therapy was removed from
standard health insurance packages. Patients increasingly have to cover their physical therapy
treatment themselves, for example by voluntary, additional health insurance packages. It is
conceivable that, due to financial issues, some patients have totally ignored physical therapy
(step-2) or only used physical therapy to a limited extent. In consequence, these patients may
not have received appropriate care. Previous research in dietetics care showed that
restrictions in reimbursement resulted in a mean decrease of 16% in treatment time per
patient15. To get insight into the impact of restrictions of reimbursement in physical therapy
practice, we asked physical therapists to what extent a patient’s reimbursement influences the
number of treatment sessions they would provide (Chapter 4). Thirty-five percent of the
therapists indicated that a patient’s reimbursement did not influence the number of provided
sessions, 41% indicated a limited influence, and the remaining 24% of the therapists reported
that a patient’s reimbursement determines the amount of care to a large extent. It is assumed
that these results reflect a patient’s opinion as well since ‘reimbursement’ is often discussed
by the physical therapist and the patient during a treatment episode. Therefore, it seems a
welcome development that from January 2018 and onward physical therapy treatment due to
hip/knee OA is included for full reimbursement again by standard health insurance16. On the
other hand, reimbursement by standard health insurance involves a compulsory deductible.
This compulsory deductible might exceed the monthly fee for additional health insurance
packages. Future research should provide insight to what extent readmission of physical
therapy treatment in hip/knee OA in standard health insurance stimulates the use of physical
therapy instead of use of secondary care facilities, and consequently attributes to an
appropriate timing of care.

Underutilization of non-surgical treatment in secondary care

As mentioned previously, a considerable part of the OA population treated by GPs is referred
for evaluation in secondary care. Our study in Chapter 3 showed that 35% of those patients
subsequently received a surgical intervention. This percentage was recently confirmed by a
publication of the Dutch National Health Care Institute in patients with hip/knee OA17.
Unfortunately, it is unknown whether or not orthopedic surgeons have determined the
adequate use of non-surgical interventions when deciding for a surgical intervention. Based on
two recently published reports, there seems room for improvement. First, the Dutch
Orthopaedic Association dedicated one of its’ five ‘sensible choices for future care’ to the
implementation of stepped-care: ‘No total joint replacement without adequate non-surgical
treatment’18. Second, in a recent publication of the Dutch National Health Care Institute

General Discussion |135

regarding ‘improvements in OA care’, it was stated that providing care in line with the stepped-
care principle will reduce the number of total joint replacements by 15%19.
To improve the implementation of stepped-care in both primary care and secondary care,
interprofessional collaboration is needed during the entire clinical OA pathway. In view of this
collaboration, the Dutch Orthopaedic Association currently revises the interprofessional
guideline ‘Non-surgical treatment in patients with osteoarthritis of the hip or knee’. Collaborating
partners in this revision process are for instance professional associations of physical
therapists, general practitioners, rheumatologists, and patients with hip/knee OA are
represented as well. In addition to recommendations regarding the use of stepped-care in OA,
the guideline will include recommendations regarding roles and responsibilities of healthcare
providers in the clinical OA pathway. By including these recommendations a further step
towards wide implementation of stepped-care and subsequently personalized care is taken.

Focus of care

In addition to an appropriate timing of care, appropriate focus of care is also supposed to

attribute to personalized treatment. In this thesis, we operationalized appropriate focus of
care by 1. treatment stratification and 2. personalized measurement (over time).

Treatment stratification
Due to the heterogeneity of especially knee OA, it has been supposed that the knee OA
population consists of different subgroups or phenotypes20,21. Each of those phenotypes
comprises one distinguishing feature which hypothetically assumes a different treatment
approach22-25. One of the questions which we have addressed in this thesis was to what extent
physical therapists currently focus their treatment at different subgroups of knee OA rather
than providing ‘one size fits all’ care (Chapter 4). The major conclusion drawn from our
clinical vignette study is that physical therapists in primary care do not apply a ‘one size fits all’
approach. Physical therapists indicated to apply different treatment strategies in five previously
identified phenotypes of knee OA. By way of illustration, patients with obesity were more
likely to be referred for dietary therapy compared to the remaining phenotypes; patients with
depressive symptoms are more often considered for psychological care; and patients with
strong quadriceps muscles would receive the least number of exercise treatment sessions in
comparison with other phenotypes.
The rationale for identifying subgroups in the knee OA population was to allow tailored
treatment for specific subgroups of patients and, subsequently, the identification of more
effective interventions4. Previous research has already shown that stratified care is both
clinically effective and cost-effective in patients with low back pain26. At this moment, Knoop
et al. are conducting a pilot study which evaluates the applicability of stratified treatment
approaches in 50 patients with knee OA (In Dutch: ‘Artrosebehandeling op maat’). Although
stratification is promising for clinical practice, one remark should be addressed as well. One

136
of the risks of stratification is that clinicians are provoked to pigeonholing. As an individual
patient practically never fits exclusively to one specific subgroup, stratification may will
compromise a patient’s individual needs rather than it will encourage personalized healthcare.
Physical therapists are challenged to use stratified treatment approaches as a practical tool
during their clinical reasoning process while maintaining an open mind on a patient’s individual
situation.

Personalized measurement
Personalizing the focus of care in patients with hip/knee OA could also be operationalized by
determining patients’ priorities or preferences in diagnostic and evaluative procedures.
Commonly used measurement tools mainly contain fixed items, which do not account for
patient-specific priorities and preferences27. Personalized measurement tools are able to
address issues which specifically influence an individual’s functioning in daily life. Since an
overview of available personalized measurement tools was lacking on the domain of physical
functioning, we performed a systematic review on this theme and, additionally, identified the
psychometric qualities of the included personalized tools. The review in Chapter 5 indicated
twelve instruments measuring physical functioning in a personalized way. However,
psychometric qualities are moderately reported yet. Construct validity, reliability and
responsibility are most often assessed. One measurement tool, the Patient Specific Functional
Scale (PSFS) operationalized by Cleland et al.28 obtained exclusively positive scores on the
assessed items. Therefore, it not surprising that the PSFS is currently well implemented in
clinical practice of physical therapists. The next step is to integrate a personalized tool like the
PSFS in the SCS to guarantee continuity and homogeneity during the total clinical pathway in
hip/knee OA.

Another measurement tool which was identified by our systematic literature review, was the
McMaster Toronto Arthritis patient preference questionnaire (MACTAR)29. The MACTAR is
a patient-specific measurement tool, which aims to collect a patient’s health status and to
evaluate the severity of restricted activities over time by a structured interview. Since the
MACTAR was exclusively validated in patients with rheumatoid arthritis29, we have conducted
a psychometric evaluation of the MACTAR in patients with hip/knee OA (Chapter 6). Based
on our results, it could be concluded that the MACTAR is a high responsive tool to evaluate
patient-specific functioning over time. Furthermore, the MACTAR seems to provide additional
insight into a patient’s individual physical functioning as a considerable part of the mentioned
priorities were not captured by fixed-item questionnaires used in patients with hip/knee OA.

Although personalized measurement potentially provide better insight in patients’ priorities,

the use of personalized-tools takes practice by both patients and clinicians. By way of
illustration, a considerable part of the patients struggles with the identification of problem
activities. Research by Stevens et al. has illustrated that high-literated patients who prefer an
active role did not indicate any difficulties in recalling problem activities, but patients who
lacked these characteristics had many difficulties to identify problem activities30. Clinicians are
General Discussion |137
challenged to facilitate patients in identifying their patient-specific limitations in daily life. In
some patients, it may be helpful to keep distance from specific activities but rather pay
attention to a patient’s meaning of life. Subsequently, patients are asked to mention physical
activities which hinder them in expressing their meaning of life. For instance, a patient values
his physical independency in daily life. One precondition for physical independency in his daily
life is the ability to walk for at least five minutes to reach his car. As he is currently able to
walk for only three minutes, this patient-specific and patient-relevant limitation could be used
to set personalized goals for treatment. Personalized goal setting has for a long time been
recognized to be an effective way of achieving behavioral change31.

Critical considerations

In this thesis, we studied to what extent elements of personalized care are represented in
current clinical practice. In consequence, we mainly used data which have been gathered
previously. For example, electronic health records of a large population treated in primary
health care were used to provide insight into current primary OA care (Chapter 2). Data
gathered as part of the regional implementation of the SCS in clinical practice were used to
determine the appropriate timing of care in hip/knee OA (Chapter 3). The use of previously
gathered data has many advantages. For example, it guarantees the availability of a sufficient
amount of data in a short time at low costs. Furthermore, the risk for selection bias is small.
However, the use of previously gathered data also has its downsides. One disadvantage is the
impossibility to influence the design of the study or the operationalization of outcomes.
Furthermore, the use of existing data has contributed to the wide timeframe of the studies
presented in this thesis. For instance, data used for the psychometric evaluation of the
MACTAR (Chapter 6) were gathered in the context of a randomized controlled clinical trial
applied in the early noughties32. Fortunately, demographic and clinical data of patients
participating in this trial were comparable to recently published studies in primary care
populations with hip/knee OA33,34.

In one study, we used visual clinical vignettes to gather data concerning the content of physical
therapy in several subgroups of knee OA (Chapter 4). Clinical vignettes offer significant
advantages in studying clinical behavior, including high feasibility and low costs35-37. We
specifically used colorized, visual clinical vignettes as visual communication takes less time to
interpret, it increases the willingness to read by 80%, and it is supposed to be better
remembered38. However, clinical vignettes also have a major disadvantage. They merely reflect
clinicians competences rather than actual clinical practice39. Future research is recommended
on studying clinicians’ routine practice in patients with hip/knee OA in addition to a therapists’
treatment intention.

138
We have demonstrated ‘personalized care’ mostly from the clinicians’ point of view. In
Chapter 2, 3, and 4, we have analyzed OA care in patients with hip/knee OA as reported by
GPs and physical therapists, in Chapter 5 and 6 we studied psychometric qualities of
personalized measurement tools to enable its’ use by clinicians in daily practice. Although
clinicians play an important role in performing personalized care, personalized care is initially
about the individual patient. Unfortunately, the patient’s point of view on personalized
healthcare has been minimally addressed in this thesis. It would have been interesting to ask
patients to what extent they are concerned in decision making processes during their
treatment due to hip/knee OA. Shared decision making has shown to attribute to personalized
healthcare in several populations including patients with hip/knee OA40-42. Therefore, future
studies on personalized treatment are highly recommended to involve patients’ perspectives
as well.

Personalized treatment: what is next?

Based on the results of this thesis, several recommendations could be formulated regarding
‘timing and focus’ to improve personalized care in patients with hip/knee OA for clinical
practice, policy, and future research. We have briefly touched some recommendations, but
we pay attention to the main recommendations below.

Clinical practice
Personalized management is not an unknown phenomenon in clinical OA practice. Clinicians
have shown to adapt their treatment to a certain degree to individual patients’ needs, both
with respect to the timing of care and focus of care. In this way, clinicians are ahead of clinical
guidelines, which only minimally pay attention to personalized care. Based on their clinical
expertise, they translate general recommendations contained in guidelines to personalized
management for their individual patients. This confirms that clinical expertise should not be
underestimated and could have an added value on evidence based treatment.

However, there is room for improvement. At this moment, the surgical rate in patients with
hip/knee OA is too high. This could partly be attributed to the suboptimal use of non-surgical
treatment strategies both in primary and secondary care. Improved collaboration, both
between healthcare providers and with the patient, could improve the use of non-surgical
treatment and, in consequence, the appropriate timing of care. In that perspective, the revision
of the interprofessional guideline for ‘non-surgical treatment in patients with hip/knee OA’ could
be seen as a timely development. As this revised guideline embraces the stepped care
approach, but additionally includes recommendations regarding clear partition of
responsibilities, it seems an important development towards structural collaboration over the
total clinical pathway.

General Discussion |139

Another recommendation based on this thesis concerns the use of personalized measurement.
In physical therapy practice, its’ use has become generally accepted. However, patient-specific
priorities identified by personalized measurement tools are yet not evaluated during the total
clinical pathway. Therefore, it is recommended to transfer patients’ individual priorities from
one clinician to another clinician, just like a patient’s diagnosis is reported in case of a referral.

The final recommendation for clinical practice concerns the implementation of stratified care.
Therapists have shown to be aware of different subgroups of patients with knee OA and,
subsequently, are able to focus their treatment on these specific subgroups. Although
stratification could be a useful tool to focus a patient’s treatment, clinicians are challenged to
maintain an open mind on a patient’s individual situation.

Policy
Based on this thesis, two recommendations could be formulated for policy. The first
recommendation concerns the nationwide implementation of a stepped care approach in OA
care, both in monodisciplinary as well as interprofessional guidelines. This implementation
could gain momentum, as several guidelines are currently under revision. In these revised
guidelines, stepped care, including interprofessional consultations, will be one of the key
elements.
The second recommendation for policy concerns the attention for funding in OA care. At this
moment, at least three flows of funding are applied to OA care: ‘fee for service’ in physical
therapy practices, ‘shared savings’ in general practices and ‘bundled-payment’ in secondary
care43. Those financial partitions in funding hamper collaboration and increase the complexity
of care. We endorse that it is not easy to look for a quick solution for this complex issue.
However, in order to do justice to the functioning of individual patients with hip/knee OA,
collaboration across flows of funding is necessary, including focus on patient-relevant
outcomes.

Future research
One of the major ‘lessons learned’ from this thesis is that monitoring the total clinical pathway
is necessary to enable the provision of personalized care. Collaboration across the borders of
settings, across the borders of professional domains, and across the borders of funding are
demanded to ensure appropriate timed care. Data from electronic medical records may be
useful in future research to monitor the clinical pathway in hip/knee OA. These data reflect
real data in clinical practice, and are consequently free of recall bias and selection bias. In the
Netherlands, NIVEL Primary Care Database (NPCD) seems suitable to this aim, as this
database holds the opportunity to link data registered by GPs, physical therapists, dieticians,
and pharmacies. Based on data registered in the NPCD, both the use of non-surgical treatment
strategies and its timing can be associated to treatment outcomes in hip/knee OA. At this
moment, such an integrated overview of applied primary healthcare and achieved treatment
outcomes is lacking in OA.

140
The second recommendation for future research concerns stratification in OA care. In this
thesis we have indicated that physical therapists recognize five subgroups in knee OA based
on clinical characteristics. Recently, Dell’Isola et al. presented a systematic review in which 76
phenotypes were described, summarized in six main sets of variables22. The way in which those
phenotypes were identified, differed widely between the included studies of the review.
Deveza et al. indicated that most phenotypes of knee OA were defined based on one single
characteristic rather than combining data from different domains like clinical, imaging, and
laboratory characteristics21. The wide variety in clinical characteristics encourages future
studies on determining clinical relevant phenotypes in knee OA. Subsequently, treatment
strategies could be developed for those clinically relevant subpopulations of patients as this is
one of the aims of personalized treatment4.

Stratification could also be applied in other ways in the pursuit of personalized treatment. In
addition to stratified treatment strategies, stratification could be applied in its mode. In that
context, the rising trend of eHealth applications should be involved in future research. During
the timeframe of the studies described in the current thesis, development of eHealth
technologies have shown exceptional growth. ‘e-Exercise’ for example combines usual physical
therapy with online, personalized education and exercise modules resulting in a ‘blended care’
concept44[. As e-Exercise has shown to be as effective as conventional care, further
development of personalized eHealth applications is recommended. Particularly, insight should
be provided into eligibility criteria to use eHealth applications: which eHealth applications fit
to a specific person and for what purpose are they applied? Moreover, little is known on the
implementation of personalized eHealth applications in usual clinical practice.

The final paragraph of this thesis on ‘personalized care in patients with hip/knee OA’ returns
to the patient with hip/knee OA. This is a patient who suffers from a heterogeneous disease
and is challenged to manage his health himself and to adapt to new situations2. Clinicians are
being consulted to strengthen a patient’s functioning in daily life in a personalized way. The
current thesis has shown that providing personalized care is not a totally new concept in
general practice, physical therapy practice, and in orthopedics. GPs, physical therapists, and
orthopedic surgeons already differentiate treatment both in timing of care as well as focus of
care. Improvement of personalized care might be achieved by better collaboration between
healthcare providers in primary care and secondary care, implementation of personalized
measurement during the total patient journey, and the use of stratification to support the
process of clinical reasoning in individual patients with hip/knee OA. Furthermore, the
determination of the concept ‘personalized care in OA’ has not been finished yet. Personalized
treatment extends beyond the right timing of care and the right focus of care, which have been
addressed in this thesis. Additional research on the identification of patients that are in greatest
need of treatment, the identification of patients who may respond optimally to a specific
intervention and an efficient use of healthcare resources is needed.

General Discussion |141

References 8. Groenewegen P, van der Zee J, Delnoij D.
Track B4: Workshop: the Dutch health
insurance reform. European Journal of Public
1. Wijma AJ, Bletterman AN, Clark JR, Vervoort Health 2006;16(suppl_1):34-36.
SCJM, Beetsma A, Keizer D, et al. Patient-
centeredness in physiotherapy: What does it 9. Wet op de beroepen in de individuele
entail? A systematic review of qualitative gezondheidszorg. Chapter 3, Part 1, Paragraph
studies. Physiother Theory Pract 2017 6, Article 29. 2017; Available at:
Nov;33(11):825-840. [Link]
7-08-01/0/informatie. Accessed 28/12, 2017.
2. Huber M, Knottnerus JA, Green L, van der
Horst H, Jadad AR, Kromhout D, et al. How 10. Koenders N. Stepped Care Strategy for
should we define health? BMJ 2011 Jul patients with hip and/or knee osteoarthritis in
26;343:d4163. Primary Care: a retrospective analysis of
medical record data. 2015.
3. Wright JG, Young NL. The patient-specific
index: asking patients what they want. J Bone 11. Lequesne MG. The algofunctional indices for
Joint Surg Am 1997 Jul;79(7):974-983. hip and knee osteoarthritis. J Rheumatol 1997
Apr;24(4):779-781.
4. Karsdal MA, Christiansen C, Ladel C,
Henriksen K, Kraus VB, Bay-Jensen AC. 12. Fernandes L, Hagen KB, Bijlsma JW,
Osteoarthritis--a case for personalized health Andreassen O, Christensen P, Conaghan PG,
care? Osteoarthritis Cartilage 2014 et al. EULAR recommendations for the non-
Jan;22(1):7-16. pharmacological core management of hip and
knee osteoarthritis. Ann Rheum Dis 2013
5. Smink AJ, van den Ende CH, Vliet Vlieland TP, Jul;72(7):1125-1135.
Swierstra BA, Kortland JH, Bijlsma JW, et al.
"Beating osteoARThritis": development of a 13. Hochberg MC, Altman RD, April KT, Benkhalti
stepped care strategy to optimize utilization M, Guyatt G, McGowan J, et al. American
and timing of non-surgical treatment College of Rheumatology 2012
modalities for patients with hip or knee recommendations for the use of
osteoarthritis. Clin Rheumatol 2011 nonpharmacologic and pharmacologic
Dec;30(12):1623-1629. therapies in osteoarthritis of the hand, hip, and
knee. Arthritis Care Res (Hoboken) 2012
6. Hofstede SN, Vliet Vlieland TP, van den Ende Apr;64(4):465-474.
CH, Nelissen RG, Marang-van de Mheen PJ,
van Bodegom-Vos L. Variation in use of non- 14. Zhang W, Nuki G, Moskowitz RW, Abramson
surgical treatments among osteoarthritis S, Altman RD, Arden NK, et al. OARSI
patients in orthopaedic practice in the recommendations for the management of hip
Netherlands. BMJ Open 2015 Sep and knee osteoarthritis: part III: Changes in
9;5(9):e009117-2015-009117. evidence following systematic cumulative
update of research published through January
7. Smink AJ, Dekker J, Vliet Vlieland TP, Swierstra 2009. Osteoarthritis Cartilage 2010
BA, Kortland JH, Bijlsma JW, et al. Health care Apr;18(4):476-499.
use of patients with osteoarthritis of the hip or
knee after implementation of a stepped-care 15. Tol J, Swinkels I, Leemrijse C, Veenhof C.
strategy: an observational study. Arthritis Care Minder diëtistische behandeling door
Res (Hoboken) 2014 Jun;66(6):817-827. grotendeels schrappen diëtetiek uit de
basisverzekering. (Factsheet). 2012.

142
16. de Groot I, Hermsen L, Miedema H, Osteoarthritis Cartilage 2010 May;18(5):601-
Obradovic M, Vijgen S, de Wit J. Fysio- en 604.
oefentherapie bij artrose aan heup en knie,
reumatoïde artritis en spondyloartritis en 25. Hinman RS, Crossley KM. Patellofemoral joint
radiculair syndroom (hernia) met motorische osteoarthritis: an important subgroup of knee
uitval. 2017;2017009103. osteoarthritis. Rheumatology (Oxford) 2007
Jul;46(7):1057-1062.
17. Zorginstituut Nederland. Afspraken evaluatie
en resultaten uit het Verbetersignalement zorg 26. Foster NE, Mullis R, Hill JC, Lewis M,
voor artrose van knie en heup. 2017. Whitehurst DG, Doyle C, et al. Effect of
stratified care for low back pain in family
18. Nederlandse Orthopaedische Vereniging. practice (IMPaCT Back): a prospective
Verstandige keuzes binnen de orthopedie. population-based sequential comparison. Ann
2015; Available at: Fam Med 2014 Mar-Apr;12(2):102-111.
[Link]
nieuws/artikel/nov-orthopeden-moeten-niet- 27. Seror R, Tubach F, Baron G, Guillemin F,
[Link]. Accessed 12/30, 2017. Ravaud P. Measure of function in rheumatoid
arthritis: individualised or classical scales? Ann
19. Zorginstituut Nederland. Zinnige Zorg Rheum Dis 2010 Jan;69(1):97-101.
Verbetersignalement voor zorg bij artrose van
knie en heup. 2014;2014082026. 28. Cleland JA, Fritz JM, Whitman JM, Palmer JA.
The reliability and construct validity of the
20. Bruyere O, Cooper C, Arden N, Branco J, Neck Disability Index and patient specific
Brandi ML, Herrero-Beaumont G, et al. Can functional scale in patients with cervical
we identify patients with high risk of radiculopathy. Spine (Phila Pa 1976) 2006 Mar
osteoarthritis progression who will respond to 1;31(5):598-602.
treatment? A focus on epidemiology and
phenotype of osteoarthritis. Drugs Aging 2015 29. Verhoeven AC, Boers M, van der Liden S.
Mar;32(3):179-187. Validity of the MACTAR questionnaire as a
functional index in a rheumatoid arthritis
21. Deveza LA, Melo L, Yamato TP, Mills K, Ravi clinical trial. The McMaster Toronto Arthritis.
V, Hunter DJ. Knee osteoarthritis phenotypes J Rheumatol 2000 Dec;27(12):2801-2809.
and their relevance for outcomes: a systematic
review. Osteoarthritis Cartilage 2017 30. Stevens A, Moser A, Koke A, van der Weijden
Dec;25(12):1926-1941. T, Beurskens A. The patient's perspective of
the feasibility of a patient-specific instrument in
22. Dell'Isola A, Allan R, Smith SL, Marreiros SS, physiotherapy goal setting: a qualitative study.
Steultjens M. Identification of clinical Patient Prefer Adherence 2016 Mar 31;10:425-
phenotypes in knee osteoarthritis: a systematic 434.
review of the literature. BMC Musculoskelet
Disord 2016 Oct 12;17(1):425. 31. Locke EA, Bryan JF. The effects of goal-setting,
rule-learning, and knowledge of score on
23. Driban JB, Sitler MR, Barbe MF, performance. Am J Psychol 1966
Balasubramanian E. Is osteoarthritis a Sep;79(3):451-457.
heterogeneous disease that can be stratified
into subsets? Clin Rheumatol 2010 32. Veenhof C, Koke AJ, Dekker J, Oostendorp
Feb;29(2):123-131. RA, Bijlsma JW, van Tulder MW, et al.
24. Felson DT. Identifying different osteoarthritis Effectiveness of behavioral graded activity in
phenotypes through epidemiology. patients with osteoarthritis of the hip and/or

General Discussion |143

 knee: A randomized clinical trial. Arthritis
Rheum 2006 Dec 15;55(6):925-934. 39. Shah R, Edgar DF, Evans BJ. A comparison of
standardised patients, record abstraction and
33. Gronhaug G, Osteras N, Hagen KB. Quality of clinical vignettes for the purpose of measuring
hip and knee osteoarthritis management in clinical practice. Ophthalmic Physiol Opt 2010
primary health care in a Norwegian county: a May;30(3):209-224.
cross-sectional survey. BMC Health Serv Res
2014 Nov 25;14:598-014-0598-x. 40. Bozic KJ, Belkora J, Chan V, Youm J, Zhou T,
Dupaix J, et al. Shared decision making in
34. Fitzgerald GK, Fritz JM, Childs JD, Brennan GP, patients with osteoarthritis of the hip and
Talisa V, Gil AB, et al. Exercise, manual knee: results of a randomized controlled trial.
therapy, and use of booster sessions in physical J Bone Joint Surg Am 2013 Sep 18;95(18):1633-
therapy for knee osteoarthritis: a multi-center, 1639.
factorial randomized clinical trial.
Osteoarthritis Cartilage 2016 Aug;24(8):1340- 41. Klifto K, Klifto C, Slover J. Current concepts
1349. of shared decision making in orthopedic
surgery. Curr Rev Musculoskelet Med 2017
35. Mohan D, Fischhoff B, Farris C, Switzer GE, Jun;10(2):253-257.
Rosengart MR, Yealy DM, et al. Validating a
vignette-based instrument to study physician 42. Sepucha K, Atlas SJ, Chang Y, Dorrwachter J,
decision making in trauma triage. Med Decis Freiberg A, Mangla M, et al. Patient Decision
Making 2014 Feb;34(2):242-252. Aids Improve Decision Quality and Patient
Experience and Reduce Surgical Rates in
36. Peabody JW, Luck J, Glassman P, Jain S, Hansen Routine Orthopaedic Care: A Prospective
J, Spell M, et al. Measuring the quality of Cohort Study. J Bone Joint Surg Am 2017 Aug
physician practice by using clinical vignettes: a 2;99(15):1253-1260.
prospective validation study. Ann Intern Med
2004 Nov 16;141(10):771-780. 43. Berg M, Ikkersheim D, Kruytzer J, Kuperus K,
Wiechers J. Populatiebekostiging: waarom, wat
37. Rutten GM, Harting J, Rutten ST, Bekkering en hoe? 2013.
GE, Kremers SP. Measuring physiotherapists'
guideline adherence by means of clinical 44. Kloek C, Bossen D, Spreeuwenberg P, Dekker
vignettes: a validation study. J Eval Clin Pract J, de Bakker D, Veenhof C, et al. Effectiveness
2006 Oct;12(5):491-500. of a blended physiotherapy intervention in
patients with hip and/or knee osteoarthritis: a
38. McCabe DP, Castel AD. Seeing is believing: the cluster randomized controlled trial. Physical
effect of brain images on judgments of scientific Therapy In press.
reasoning. Cognition 2008 Apr;107(1):343-
352.

144
Summary
Osteoarthritis (OA) of the hip and/or knee is a common musculoskeletal disorder, which
prevalence is expected to increase considerably over the next decades. OA is a heterogeneous
disease. Both the pathogenesis and the clinical presentation differ widely from patient to
patient. To improve the functioning of a rising number of patients with hip/knee OA, there is
a strong demand for efficient and effective treatment. Due to its’ heterogeneity and the
expected increasing burden of disease, OA would be particularly suitable for personalized
treatment.

The general introduction of this thesis (Chapter 1) presents the rationale for personalized
treatment in patients with hip/knee OA. Personalized treatment is recognized as one of the
solutions to face the complex and continuously expanding health care demand of patients with
(multiple) chronic conditions like OA. One motivation to apply personalized treatment is the
identification of patients that are in greatest need of treatment. Furthermore, personalized
treatment potentially enables the development of stratified interventions and efficient use of
healthcare resources. Personalized treatment can be operationalized in many ways. The aim
of this thesis is to explore two drivers which potentially contribute to personalized treatment
in hip/knee OA ‘timing of care’ (part I) and ‘focus of care’ (part II).

Part I: Timing of care

Timing of care concerns the phasing of interventions over time in patients with hip/knee OA,
in particular non-surgical interventions in general practice, physical therapy practice, and
orthopedic care. Although current clinical guidelines provide a clear insight into effective
interventions, those guidelines do not yet provide a clear guidance for the sequence of
interventions. This might be one explanation for the reported underutilization of non-surgical
interventions in current OA care. To facilitate the use of non-surgical interventions prior to a
referral to secondary care, a Stepped-Care-Strategy (SCS) has been developed by Smink et al.
(2011). In the SCS, interventions are applied in three consecutive steps. Step-1 comprises
stimulation of a patient’s self-care. Step-2 adds more advanced interventions, like the
prescription of NSAIDs and a referral for allied healthcare (exercise therapy / dietary therapy).
Step-3 includes, for instance, a referral to secondary care. Patients will only receive more
advanced care in case of insufficient results achieved by less advanced interventions.
The SCS has already been implemented in one region in the Netherlands. Nationwide
implementation has not been finished yet. In preparation to nationwide implementation of the
SCS, we provide insight into current OA care by general practitioners (GPs) and physical
therapists, including the compliance to the SCS.

Research questions of Chapter 2 are:

 What is the content of current general practice care in patients with hip/knee OA,
including the compliance to the SCS?

Summary |147
  To what extent does the content of physical therapy practice in patients who were
referred by their GP differ from self-referred patients with hip/knee OA?

To answer these questions, we retrospectively studied routinely registered data by Dutch GPs
and physiotherapists in NIVEL Primary Care Database. In total, 12 118 electronic medical
records of patients who were treated by GPs and/or physical therapists due to hip/knee OA
were included for analyses. Results showed that Dutch GPs already phase their interventions
over time: they apply less advanced interventions, like lifestyle advises and the prescription of
NSAIDs (step-1,2), more often than for instance exercise therapy or a referral for total joint
replacement (step-2,3). Although they mainly act in accordance with the SCS, there is room
for improvement. First, GPs rarely use all interventions that are recommended within one
step. Furthermore, referrals for physical therapy lag behind referrals to secondary care. The
introduction of the Direct Access Regulation may have led to patients directly visiting the
physical therapist, without a referral by the GP. Within the timeframe of our study, 35% of
the patients with hip/knee OA used direct access. The content of physical therapy differed
only slightly between GP-referred patients and patients who visited a physical therapist on
their own initiative.

Chapter 3 focuses on the setting in which OA care is provided, which is closely related to
the timing of care. As primary care is easily accessible, and cheaper compared to secondary
care, the SCS advises to start OA management in primary care. Previous research has shown
that a considerable part of the OA population is referred to secondary care. Subsequently,
about one third of this referred population receives a total joint replacement. To date, little is
known regarding factors which support the choice to stay in primary care, to refer for
treatment in secondary care or to conduct total joint replacement surgery. These factors may
include patient-characteristics, practitioner-characteristics, or practice characteristics. Insight
into these factors at multiple levels would be useful to optimize nationwide implementation of
the SCS in clinical practice.

Therefore, the question addressed in Chapter 3 is:

 Which individual factors, GP-related factors and general practice related factors are
associated with 1) treatment in primary care, 2) continuation of non-surgical care after
a referral, and (3) surgical treatment in patients visiting their GP due to hip/knee OA?

To answer this question, we performed logistic multilevel analyses on data previously gathered
in a cohort of 313 newly diagnosed patients with hip/knee OA treated by 70 GPs in 38 general
practices in the Netherlands. Overall, significant factors almost exclusively concerned factors
at patient level, in particular the content of previously utilized care. Treatment was limited to
primary care in 47% of the patients. Patients who stayed in primary care tended to show better
physical functioning (Odds Ratio (OR) 1.03). Furthermore, they less often received exercise
therapy (OR 0.46) or intra-articular injections (OR 0.08), and were less often registered for

148
radiological assessments (OR 0.06). These results are in accordance with the SCS, which
prefers the use of less advanced care prior to the use of more advanced interventions including
a referral to secondary care.
Two out of three patients who have been referred to secondary care were exclusively treated
by non-surgical interventions. The decision to continue non-surgical treatment after a referral
was more likely in employed patients (OR 2.90). Furthermore, each additional joint affected
with OA doubles the chance to continue non-surgical treatment in secondary care. As the
incidence of multi-joint OA is expected to increase over the next decades, further research
on this topic is justified.
No factors related to the patient, the GP nor the general practice were identified with respect
to the use of surgical interventions, except the previous use of exercise therapy. The
application of exercise therapy in the past offers a substantial higher probability to receive a
surgical intervention (OR 7.42). This result could have been expected based on the SCS.
Moreover, it strengthens the important role of orthopedic surgeons as they actually seem to
consider the use of non-surgical interventions prior to decide for a surgical intervention.

Part II: Focus of care

In this thesis, focus of care has been operationalized by ‘treatment stratification’ and
‘personalized measurement’. Treatment stratification aims to better account for different
subgroups within the population with knee OA as treatment is adapted to those different
subgroups. Personalized measurement aims to incorporate patients’ preferences and priorities
since commonly used fixed-item tools lack this possibility.

Treatment stratification
The clinical presentation of patients with hip/knee OA differs from patient to patient. Knoop
et al. (2011) and van der Esch et al. (2015) have succeeded in distinguishing five clinical
phenotypes of knee OA: minimal joint disease phenotype, strong muscle phenotype, severe
radiologic OA phenotype, obese phenotype and depressive mood phenotype. As both the
origin of complaints as well as the course of complaints differ between those phenotypes, it is
supposed that each phenotype would benefit from specific management. It is unknown to what
extent physical therapists currently differentiate their treatment between phenotypes of knee
OA.

Therefore, the research question of Chapter 4 is:

 To what extent is physical therapists’ treatment tailored to the ‘minimal joint disease’
phenotype, the ‘strong muscle strength’ phenotype, the ‘severe radiographic OA’
phenotype, the ‘obese’ phenotype, and the ‘depressive mood’ phenotype in patients
with knee OA?

Summary |149
Hypotheses were constructed for each phenotype regarding preferred treatment strategies,
the referral policy, and the considered number of applied treatment sessions. To test these
hypotheses, a clinical vignette study has been conducted. We composed five visual clinical
vignettes which represented the previously mentioned phenotypes of knee OA. Subsequently,
144 Dutch physical therapists and exercise therapists were recruited to indicate their content
and amount of treatment in each of the five clinical vignettes. Overall, physical therapists and
exercise therapists seemed to tailor their management to specific phenotypes of knee OA.
Statistically significant differences were found regarding the content and the amount of care
between phenotypes of knee OA. These differences were mainly in accordance with our
predefined hypotheses, with the exception of the strong muscle phenotype and the depressive
mood phenotype. In the strong muscle phenotype, exercise therapy was provided more often
than hypothesized and the referral rate to secondary care was lower than expected. In the
depressive mood phenotype, the referral rate to psychologists exceeded our expectations.
Further research was recommended on the development of stratified interventions in knee
OA and its evaluation of (cost)effectiveness.

Personalized measurement
In addition to stratified treatment, personalized measurement could also attribute to a
personalized focus of care in patients with hip/knee OA. At this moment, most tools applied
in diagnostics or evaluation of limitations in physical functioning in patients with hip/knee OA
are fixed-item tools. By those tools, patients’ preferences and variability in performance on
particular activities can not be measured. As a consequence, clinicians face the risk of missing
patient-relevant priorities which could potentially be used in personalized goal-setting. To
provide care with focus on each individual patient, it is desirable to consider the use of a
personalized instrument in addition to a fixed-item tool. At this moment, it is unclear which
personalized-measurement tools are available in patients with musculoskeletal disorders,
including hip/knee OA.

Therefore, we performed a systematic review to indicate those tools (Chapter 5).

Subsequently, we assessed the psychometric qualities of the included instruments.

Corresponding research questions are:

 Which patient-specific self-assessment instruments are available to measure physical
function in patients with musculoskeletal disorders?
 What are the descriptive properties as well as the psychometric qualities of patient-
specific instruments measuring physical function in patients with musculoskeletal
disorders?

A comprehensive search was conducted in several databases. Studies were included if (1) the
main aim of the study was to investigate measurement properties, (2) the described instrument
concerned a questionnaire, a rating scale or a (semi-structured) interview measuring at least

150
physical functioning in patients with musculoskeletal disorders, and (3) the instrument had a
patient-specific character. We included 23 out of 1617 studies referring to twelve different
instruments. The Patient-Specific Functional Scale was the most reported patient-specific
measurement tool in patients with musculoskeletal disorders. In addition to a description of
the included instruments, the psychometric quality of the eligible studies was assessed by the
‘Quality criteria for measurement properties of health status questionnaires’. None of the included
instruments has been evaluated on all criteria of the checklist (9 criteria). Eight instruments
obtained exclusively positive scores on the items which have been evaluated (maximum 4
criteria). However, the majority of quality criteria has not been evaluated yet. Therefore, more
research is needed to get insight into a broader range of psychometric properties of
instruments measuring physical functioning in patients with musculoskeletal disorders in a
patient-specific way.

Chapter 6 directly responds to this request by evaluating the psychometric qualities of a

patient-specific measurement tool, originally developed in patients with rheumatic arthritis:
the Dutch McMaster Toronto Arthritis Patient Preference Questionnaire (MACTAR). The
MACTAR aims to evaluate changes in patient-specific physical functioning over time by
reassessing the experienced difficulties in ten self-reported activities of daily living during a
semi-structured interview.

The corresponding research question was:

 What is the content validity, the construct validity and the responsiveness of the Dutch
MACTAR in patients with OA of the hip or knee?

To answer this question, data were used from a previously conducted randomized controlled
trial in patients with hip/knee OA. Content validity of the MACTAR was determined by
comparing the mentioned activities in the MACTAR to the activities included in the Dutch
‘Western Ontario and McMaster Universities Osteoarthritis Index’ (WOMAC) respectively
the Dutch ‘Medical Outcome Survey Short Form 36’ (SF-36). In total, 27% of the activities
gathered by the MACTAR was not represented in the WOMAC, 41% were not represented
in the SF-36, and 11% were not covered in the WOMAC nor SF-36. Construct validity was
determined by testing theoretically derived hypotheses concerning correlations between
change scores on the MACTAR and change scores on the WOMAC respectively the SF-36.
In contrast to our expectations, we identified only moderate correlations between change
scores of the MACTAR, WOMAC and SF-36 (rs=.27 - rs=.44). Responsiveness of the
MACTAR was indicated to be ‘good’, based on an ‘Area Under the Curve’ of .90 and better
correlations between MACTAR change scores and Patient Global Assessment (PGA) scores
than correlations between WOMAC / SF-36 change scores and PGA scores. These results
suggest that the MACTAR is potentially better able to detect changes over time in activities
that are important to individual patients compared to other tools measuring physical
functioning. Therefore, clinicians are recommended to use the MACTAR to evaluate clinically

Summary |151
relevant changes over time in patient-specific physical functioning in addition to existing tools
in patients with hip/knee OA.

Chapter 7 comprises a general discussion on the current state of affairs on personalized

management in patients with hip/knee OA. It can be concluded that aspects of personalized
care are already extensively applied in current clinical OA care. GPs, physical therapists, and
orthopedic surgeons differentiate their treatment both in timing of care and in focus of care.
However, the determination of the concept ‘personalized care’ has not been completed yet,
neither in research nor in healthcare policy or clinical practice. Therefore, Chapter 7
comprises several recommendations for clinical OA practice, policy, and future research to
reach the ultimate goals of personalized healthcare: to treat the right patients, at the right
time, with the right interventions at the lowest possible price.

152
Samenvatting
Artrose van de heup en/of knie is een veelvoorkomende musculoskeletale aandoening.
Verwacht wordt dat in de komende decennia de prevalentie van artrose aanzienlijk zal gaan
toenemen. Artrose is een heterogene aandoening. Zowel de pathogenese als de klinische
presentatie varieert enorm van patiënt tot patiënt. Om het functioneren van een toenemende
aantal patiënten met artrose te verbeteren, is de vraag naar effectieve en efficiënte behandeling
groot. Vanwege de heterogeniteit van artrose en de verwachtte stijging in omvang, lijkt artrose
bijzonder geschikt voor gepersonaliseerde behandeling.

De algemene introductie van dit proefschrift (Hoofdstuk 1) presenteert de rationale voor

gepersonaliseerde behandeling bij patiënten met heup/knie artrose. Gepersonaliseerde
behandeling wordt gezien als één van de oplossingen om het hoofd te kunnen bieden aan de
complexe en steeds groter wordende zorgvraag van mensen met (meerdere) chronische
aandoeningen, inclusief artrose. Één reden om gepersonaliseerde behandeling toe te passen is
de identificatie van patiënten die de grootste behoefte hebben aan zorg. Daarnaast maakt
gepersonaliseerde behandeling het mogelijk om gestratificeerde zorg toe te passen en efficiënt
gebruik te maken van gezondheidzorgbudgetten. Gepersonaliseerde zorg kan op verschillende
manieren vormgegeven worden. Het doel van dit proefschrift is het verkennen van twee
drivers die in potentie bijdragen aan gepersonaliseerde behandeling van patiënten met
heup/knie artrose: ‘timing van zorg’ (deel 1) en ‘focus van zorg’ (deel 2).

Deel I: Timing van zorg

Timing van zorg heeft betrekking op de fasering van interventies over de tijd bij patiënten met
heup/knie artrose. Hierbij richt dit proefschrift zich in het bijzonder op de niet-chirurgische
behandeling door de huisarts, de fysiotherapeut en de orthopeed. Ondanks dat de huidige
klinische richtlijnen een helder overzicht geven van effectieve interventies bij heup/knie
artrose, schenken deze richtlijnen slechts zeer beperkt aandacht aan de volgorde waarin deze
interventies toegepast moeten worden. Dit is misschien één van de verklaringen voor het vaak
gerapporteerde ondergebruik van niet-chirurgische interventies in de huidige artrose-zorg.
Om het gebruik van niet-chirurgische interventies voorafgaand aan een verwijzing naar de
tweedelijns zorg te faciliteren, hebben Smink et al. een stepped-care behandelstrategie
ontwikkeld (Behandelstrategie ARTrose, 2011). In deze stepped-care strategie worden
interventies in drie opeenvolgende stappen aangeboden. Stap-1 omvat interventies om de
zelfzorg te stimuleren. Stap-2 voegt meer geavanceerde interventies toe, zoals de prescriptie
van NSAID’s en een verwijzing voor paramedische zorg (fysiotherapie / diëtiek). Stap-3 bevat
onder andere een verwijzing naar tweedelijns zorg. Het uitgangspunt van de stepped-care
behandelstrategie is dat patiënten alléén meer geavanceerde zorg ontvangen bij het uitblijven
van resultaten door minder geavanceerde interventies.

Samenvatting |155
De stepped-care behandelstrategie is op dit moment geïmplementeerd in één regio in
Nederland. Landelijke implementatie is nog niet afgerond. In voorbereiding op landelijke
implementatie van de stepped-care behandelstrategie geven we in dit proefschrift inzicht in de
huidige artrosezorg door huisartsen en fysiotherapeuten, inclusief de overeenstemming met
de stepped-care behandelstrategie.

Onderzoeksvragen van Hoofdstuk 2 zijn:

 Wat is de inhoud van de huidige zorg door huisartsen voor patiënten met heup/knie
artrose, inclusief de overeenstemming met de stepped-care behandelstrategie?
 In welke mate verschilt de inhoud van fysiotherapeutische zorg tussen patiënten met
heup/knie artrose die verwezen zijn door hun huisarts van patiënten die de
fysiotherapeut op eigen initiatief bezoeken?

Om deze vragen te kunnen beantwoorden hebben we retrospectief data bestudeerd die

routinematig verzameld zijn door Nederlandse huisartsen en fysiotherapeuten ten behoeve
van ‘NIVEL Zorgregistraties eerstelijn’. In totaal zijn 12 118 medisch-elektronische dossiers
geïncludeerd van patiënten die behandeld zijn door hun huisarts en/of fysiotherapeut vanwege
heup/knie artrose. De uitgevoerde, beschrijvende analyses laten zien dat Nederlandse
huisartsen hun interventies op dit moment al faseren in de tijd: ze passen minder geavanceerde
interventies zoals leefstijladviezen en de prescriptie van NSAID’s (stap 1&2), vaker toe dan
bijvoorbeeld een verwijzing voor fysiotherapie of een verwijzing voor gewrichtsvervanging in
de tweedelijn (stap 2&3). Ondanks dat ze grotendeels handelen in overeenstemming met de
stappen van de stepped-care behandelstrategie, is er ruimte voor verbetering. Ten eerste, op
dit moment passen huisartsen zelden alle interventies toe die bij één stap van de stepped-care
behandelstrategie horen. Daarnaast verwijzen huisartsen patiënten vaker naar een orthopeed
dan naar een fysiotherapeut. De introductie van de regeling Directe Toegankelijkheid
Fysiotherapie biedt patiënten de mogelijkheid om additioneel aan een verwijzing door een
(huis)arts op eigen initiatief een fysiotherapeut te bezoeken. Binnen de tijdspanne van dit
onderzoek maakte 35% van de patiënten met heup/knie artrose die de fysiotherapeut bezocht
gebruik van deze regeling. De inhoud van de fysiotherapeutische zorg verschilde niet tussen
de groep zelf-verwijzers en de groep die verwezen werd door de huisarts.

Hoofdstuk 3 focust op de setting waarin de artrose zorg wordt verleend. Deze setting hangt
nauw samen met de timing van zorg. Omdat eerstelijns zorg voor iedereen toegankelijk is en
de zorg in deze setting goedkoper is in vergelijking met tweedelijns zorg, adviseert de stepped-
care behandeslstrategie om artrose-zorg te starten in de eerstelijn. Eerder onderzoek heeft
aangetoond dat een aanzienlijk deel van de artrosepopulatie uiteindelijk verwezen wordt naar
de tweedelijn. Ongeveer een derde deel van deze verwezen populatie ontvangt een
gewrichtsvervangende operatie. Op dit moment is er weinig bekend over factoren die
geassocieerd zijn met de keuze om de patiënt alléén in de eerstelijn te behandelen, te verwijzen
voor niet-chirurgische behandeling in de tweedelijn of een gewrichtsvervangende operatie te
laten ondergaan. Deze factoren kunnen zowel betrekking hebben op karakteristieken van de

156
patiënt, de huisarts als de huisartspraktijk. Inzicht in deze factoren op verschillende niveau’s
kan bruikbaar zijn bij de landelijke implementatie van de stepped-care behandelstrategie in de
dagelijkse zorgpraktijk.

De vraag die in hoofdstuk 3 centraal staat is:

 Welke individuele factoren, huisarts-gerelateerde factoren en huisartspraktijk-
gerelateerde factoren zijn geassocieerd met 1) behandeling in de eerstelijns zorg, 2)
niet-chirurgische behandeling in de tweedelijn en 3) chirurgische behandeling bij
patiënten die hun huisarts bezoeken vanwege heup/knie artrose?

Om deze vraag te kunnen beantwoorden hebben we logistische multilevel analyses uitgevoerd

op verzamelde data binnen een reeds afgeronde cohortstudie. Het cohort bestond uit 313
nieuw gediagnosticeerde patiënten met heup/knie artrose, behandeld door 70 huisartsen in 38
huisartspraktijken in Nederland. Significante factoren geassocieerd met de setting van zorg
waren voornamelijk gerelateerd aan de patiënt, in het bijzonder de reeds ontvangen zorg door
de patiënt. Bij 47% van de patiënten beperkte de zorg zich tot eerstelijn. Dit deel van de
patiënten liet een hoger niveau van fysiek functioneren zien (Odds Ratio (OR) 1,03). Verder
ontving de groep die uitsluitend behandeld werd in de eerstelijn minder vaak oefentherapie
(OR 0,46) of intra-articulaire injecties (OR 0,08) and werd minder vaak een radiologische
beoordeling gerapporteerd (OR 0,06). Deze resultaten stemmen overeen met de stepped-
care behandelstrategie, waarin het gebruik van minder geavanceerde interventies wordt
geprefereerd boven meer geavanceerde interventies, inclusief een verwijzing naar de
tweedelijn.
Twee van de drie patiënten die werden verwezen naar de tweedelijn ontvingen uitsluitend
niet-chirurgische interventies. Het besluit om niet-chirurgische interventies te continueren in
de tweedelijn werd vaker genomen bij patiënten met een werkverband (OR 2.90). Ook werd
duidelijk dat ieder extra gewricht dat aangedaan is door artrose de kans verdubbelt op
continuering van niet-chirurgische behandeling. Aangezien de kans op multi-gewrichtsartrose
toeneemt in de volgende decennia, is nader onderzoek op dit thema gewenst.
Onze studie heeft geen factoren gerelateerd aan de patiënt, de huisarts noch de
huisartspraktijk aan het licht gebracht die geassocieerd zijn met het uitvoeren van een
chirurgische interventie, behalve het gebruik van oefentherapie in het verleden. De toepassing
van oefentherapie in het verleden biedt een aanzienlijk grotere kans op het ontvangen van een
chirurgische interventie (OR 7.42). Dit resultaat komt overeen met de principes van de
stepped-care behandelstrategie. Daarnaast bevestigt het de belangrijke rol van orthopeden in
het timen van een chirurgische ingreep. Dit onderzoek toont aan dat deze professionals het
gebruik van niet-chirurgische interventies overwegen voorafgaand aan het voorstellen van een
chirurgische ingreep.

Samenvatting |157
Deel II: Focus van zorg

In dit proefschrift is de focus van zorg geoperationaliseerd door middel van

‘behandelstratificatie’ en ‘gepersonaliseerd meten’. Behandelstratificatie heeft als doel beter
rekening te houden met verschillende subgroepen binnen de patiëntenpopulatie met knie-
artrose, doordat de behandeling wordt aangepast aan deze verschillende subgroepen.
Gepersonaliseerd meten het als doel het integreren van een patiënt’ zijn of haar prioriteiten
en voorkeuren. Veelgebruikte, zogenoemde ‘fixed-item’ instrumenten missen deze
mogelijkheid.

Behandelstratificatie
De klinische presentatie van patiënten met heup/knie artrose verschilt van patiënt tot patiënt.
Knoop et al. (2011) en van der Esch et al. (2015) zijn erin geslaagd om vijf klinische fenotypen
van knie-artrose te onderscheiden: minimale gewrichtsschade, hoge spierkracht, ernstige
radiologische afwijkingen, obesitas en depressieve stemming. Omdat zowel de oorzaak van de
klachten als het beloop van de klachten verschilt tussen deze fenotypes, wordt verondersteld
dat ieder fenotype gebaat zou zijn bij een specifieke behandeling. Het is onbekend in welke
mate fysiotherapeuten op dit moment al onderscheid maken in hun behandeling tussen de vijf
genoemde fenotypen van knie-artrose.

Daarom luidt de onderzoeksvraag van Hoofdstuk 4:

 In welke mate is de fysiotherapeutische behandeling afgestemd het ‘minimale
gewrichtsschade’ fenotype, het ‘hoge spierkracht’ fenotype, het fenotype met ‘ernstige
radiologische afwijkingen’, het ‘obese’ fenotype en het fenotype met een ‘depressieve
stemming’ binnen de populatie patiënten met knie-artrose?

Voorafgaand aan het onderzoek zijn hypotheses opgesteld voor ieder fenotype in relatie tot
de geprefereerde interventies, het verwijsbeleid en het aantal te geven behandelsessies. Deze
hypotheses zijn getoetst via een klinische vignettudie. We hebben vijf visuele, klinische
vignetten samengesteld die de reeds genoemde fenotypen vertegenwoordigden. Vervolgens
zijn in Nederland 144 fysiotherapeuten en oefentherapeuten geworven om aan de hand van
de vignetten hun voorgenomen behandeling inclusief de omvang van de behandeling te
beschrijven. Over het algemeen lijken fysiotherapeuten en oefentherapeuten hun behandeling
aan te passen aan de verschillende fenotypen van knie-artrose. Er zijn namelijk statistisch
significante verschillen gevonden ten aanzien van de inhoud en omvang van zorg tussen de vijf
fenotypen. Deze verschillen kwamen grotendeels overeen met onze vooraf gestelde
hypotheses, met uitzondering van het sterke spierkracht fenotype en het fenotype met een
depressieve stemming. Met betrekking tot het sterke spierkracht fenotype viel op dat
oefentherapie veel vaker werd toegepast dan vooral gedacht en dat het aantal verwijzingen
naar de tweedelijn juist lager was dan verwacht. Bij het fenotype met een depressieve
stemming oversteeg het aantal verwijzingen naar psychologen onze verwachting. Nader

158
onderzoek wordt aanbevolen met betrekking tot de ontwikkeling van gestratificeerde
interventies bij patiënten met knie-artrose, inclusief de evaluatie op (kosten)effectiviteit.

Gepersonaliseerd meten
Naast gestratificeerde behandeling kan gepersonaliseerd meten ook bijdragen aan een
gepersonaliseerde focus van zorg bij patiënten met heup/knie artrose. Op dit moment worden
met name fixed-item instrumenten gebruikt bij de diagnostiek en evaluatie van beperkingen in
het fysiek functioneren binnen de artrose-populatie. Deze fixed-item instrumenten zijn niet in
staat om voorkeuren en prioriteiten van patiënten mee te nemen. Ook variatie in uitvoering
van specifieke activiteiten kunnen niet gemeten worden. Dit heeft als consequentie dat clinici
patiënt-relevante activiteiten missen, die ze hadden kunnen inzetten bij gepersonaliseerde
doelbepaling. Om zorg te kunnen bieden die gefocust is op de individuele patiënt is het
wenselijk om het gebruik van gepersonaliseerde meetinstrumenten te overwegen in aanvulling
op fixed-item instrumenten. Op dit moment is het niet duidelijk welke gepersonaliseerde
meetinstrumenten beschikbaar zijn voor patiënten met musculoskeletale aandoeningen,
waaronder heup/knie artrose.

Daarom hebben we een systematische literatuurstudie uitgevoerd om deze instrumenten te

identificeren (Hoofdstuk 5). Vervolgens hebben we de psychometrische kwaliteiten van de
instrumenten bepaald.

De onderzoeksvragen waren:
 Welke gepersonaliseerde zelf-assessment instrumenten zijn beschikbaar om fysiek
functioneren te meten bij patiënten met musculoskeletale aandoeningen?
 Wat zijn de kenmerken en psychometrische kwaliteiten van gepersonaliseerde
meetinstrumenten die fysiek functioneren meten bij patiënten met musculoskeletale
aandoeningen?

We voerden een uitgebreid literatuuronderzoek uit in verschillende databases. Studies werden

geïncludeerd als: (1) het hoofddoel van de studie was het onderzoeken van psychometrische
eigenschappen, (2) het beschreven instrument een vragenlijst, een beoordelingsschaal of een
(semi-gestructureerd) interview was die minimaal fysiek functioneren in kaart bracht bij
mensen met musculoskeletale aandoeningen en (3) het instrument een gepersonaliseerd
karakter had. Uiteindelijk zijn 23 van de 1617 geïdentificeerde studies geïncludeerd, verwijzend
naar twaalf verschillende instrumenten. De ‘Patient-Specific Functional Scale’ was het meest
gerapporteerde, gepersonaliseerde meetinstrument bij patiënten met musculoskeletale
aandoeningen. Naast een beschrijving van de geïncludeerde artikelen is de psychometrische
kwaliteit van de studies getest door middel van de lijst ‘Quality criteria for measurement properties
of health status questionnaires’. Geen van de geïncludeerde artikelen is geëvalueerd op alle
criteria van de lijst (9 criteria). Acht instrumenten ontvingen wel uitsluitend positieve scores
op de items die beschreven zijn, met een maximum van 4 criteria. Echter, de meerderheid van

Samenvatting |159
de kwaliteitscriteria is op dit moment nog niet onderzocht. Meer onderzoek is daarom nodig
om inzicht te krijgen in een breder spectrum van psychometrische kwaliteiten van
instrumenten die fysiek functioneren in kaart brengen bij patiënten met musculoskeletale
aandoeningen op een gepersonaliseerde wijze.

Hoofdstuk 6 geeft direct gehoor aan deze oproep door het evalueren van de psycho-
metrische kwaliteit van een gepersonaliseerd meetinstrument, oorspronkelijk ontwikkeld voor
patiënten met reumatoïde artritis. Dit instrument is de ‘Dutch McMaster Toronto Arthritis
Patient Preference Questionnaire’ (MACTAR). De MACTAR heeft als doel om veranderingen
in de tijd te evalueren op het gebied van gepersonaliseerd fysiek functioneren. Dit wordt
gedaan via het herbeoordelen van ervaren problemen met maximaal tien zelf-gerapporteerde
activiteiten van het dagelijks leven in een semi-gestructueerd interview.

De bijbehorende onderzoeksvraag was:

 Wat is de content validiteit, de construct validiteit en de responsiviteit van de
Nederlandse versie van de MACTAR bij patiënten met artrose van de heup of de knie?

Om deze vraag te beantwoorden hebben we gebruik gemaakt van data die reeds eerder
verzameld is in het kader van een gerandomiseerd, gecontroleerd onderzoek in een
patiëntenpopulatie met heup/knie artrose. De content validiteit van de MACTAR is bepaald
door het vergelijken van genoemde problematische activiteiten in de MACTAR met
activiteiten die voorkomen in de Nederlandse versies van de ‘Western Ontario and McMaster
Universities Osteoarthritis Index’ (WOMAC) respectievelijk de ‘Medical Outcome Survey
Short Form 36’ (SF-36). In totaal kwam 27% van de activiteiten verzameld in de MACTAR niet
voor in de WOMAC, 41% niet in de SF-36 en 11% werd niet gedekt in de WOMAC noch in
de SF-36. Construct validiteit werd bepaald door het toetsen van hypotheses over de
correlatie tussen veranderscores op de MACTAR met veranderscores op de WOMAC / SF-
36. In tegenstelling tot onze verwachtingen vonden we uitsluitend matige correlaties tussen
veranderscores op de MACTAR, WOMAC en SF-36 (rs=.27 – rs=.44). Responsiviteit van de
MACTAR werd op basis van twee manieren als ‘goed’ beoordeeld. Ten eerste op basis van
een ‘Area Under the Curve’ van .90. Ten tweede werd de goede responsiviteit van de
MACTAR bevestigd door een hogere correlatie tussen de veranderscores op de MACTAR
en Patient Global Assessment (PGA) scores dan de correlatie tussen de veranderscores op
de WOMAC / SF-36 en de PGA scores. Deze resultaten suggereren dat de MACTAR in
potentie beter in staat is om veranderingen over de tijd te detecteren in activiteiten die
belangrijk zijn voor individuele patiënten dan andere instrumenten die fysiek functioneren
meten. Daarom wordt aanbevolen om, naast de reeds bestaande instrumenten, de MACTAR
te gebruiken in de zorgpraktijk om klinisch relevante veranderingen over de tijd te meten in
gepersonaliseerd fysiek functioneren.

Hoofdstuk 7 bevat een algemene discussie over de huidige stand van zaken met betrekking
tot gepersonaliseerde behandeling van patiënten met heup/knie artrose. Er kan geconcludeerd

160
worden dat aspecten van gepersonaliseerde zorg al uitgebreid toegepast worden in de huidige
klinische artrose-zorg. Huisartsen, fysiotherapeuten en orthopeden differentiëren hun
behandeling zowel in timing van zorg als focus van zorg. Echter, de ontwikkeling van het
concept ‘gepersonaliseerde zorg’ is nog niet afgerond. Dit geldt zowel voor het onderzoek,
het beleid en de klinische praktijk. Daarom bevat hoofdstuk 7 verschillende aanbevelingen voor
klinische artrose-zorg, beleid en onderzoek om het ultieme doel van gepersonaliseerde zorg
te kunnen bereiken: het behandelen van de juiste patiënten, op het juiste moment met de
juiste interventies tegen de laagst mogelijke prijs.

Samenvatting |161
Dankwoord
‘Een sprokkelpromotie? Daar heb ik echt nog nooit van gehoord. Wat is dat?’ Deze zinnen heb ik
de afgelopen zes jaar vaak gehoord uit monden van onderzoekers, beleidsmakers en
fysiotherapeuten. Voor NIVEL-lers is het een bekende constructie: je participeert in een
lopend onderzoeksproject en gebruikt data die verzameld zijn binnen dit project als basis voor
je proefschrift. Waar nodig ‘sprokkel’ je nog data bij andere projecten of je voert een
aanvullend onderzoek uit om je vraagstellingen te kunnen beantwoorden. Het klinkt misschien
wat armoedig, maar ik heb ervaren dat een sprokkelpromotie ook veel voordelen kent. Drie
voorbeelden. (1) Géén subsidie voor drie of vier jaar waarin je je onderzoek moet opzetten,
uitvoeren, analyseren én publiceren. Met drie kinderen geboren in vier jaar tijd was dat best
een opgave geworden. (2) Géén totale focus op één onderwerp. Doordat ik primair werkte
aan een veelzijdig project als ‘NIVEL Zorgregistraties eerste lijn’ heb ik me kunnen ontwikkelen
op veel meer vlakken dan alleen het doen van onderzoek. (3) Géén strijd tegen de
eenzaamheid van het promovendus-bestaan. Perioden van focus op het proefschrift wisselden
vaak met hectische perioden rond projectdeadlines.
Natuurlijk ben ik best wel eens jaloers geweest op collega-onderzoekers die zich vol konden
storten op hun vierjarige promotie-onderzoek. Ik bewonder de gedrevenheid waarmee deze
mensen zich vastbijten in één onderwerp en na vier jaar hard werken hun levenswerk
opleveren in een mooi boekje. Na het promotiefeestje sluiten ze een fase in hun leven af en
begint een volgende fase waarin het sociale leven weer een plaats kan krijgen.

Wat dat betreft sta ik er echt wel anders in. Promoveren heeft nooit mijn leven beheerst in
positieve of negatieve zin. Het was een natuurlijk onderdeel van mijn dagelijks bestaan, net als
mijn gezin en sociale leven. Dat neemt niet weg dat ik het als een groot voorrecht beschouw
dat ik de kans heb gekregen om via het promotie-onderzoek te doen waar ik energie van krijg:
bijdragen aan de innovatie van de (beweeg)zorg ten dienste van de individuele patiënt.
Gedurende het promotie-traject heb ik me kunnen ontwikkelen als onderzoeker, heb ik veel
verschillende mensen leren kennen in heel verschillende rollen en organisaties en heb ik
plaatsen van de wereld gezien die ik nog niet eerder gezien had. De waarde van promoveren
zit voor mij dus zeker niet alleen in het uiteindelijke boekje, maar veel meer in het traject daar
naartoe.

Tijdens dit traject is een groot aantal mensen belangrijk geweest, waarvan ik er een aantal in
het bijzonder wil noemen.

In de eerste plaats de leden van het promotieteam. Graag had ik willen beginnen met het
bedanken van Dinny de Bakker. Dinny heeft mij aangenomen als junior onderzoeker op het
project LiPZ, de voorganger van NIVEL Zorgregistraties, en werd later mijn promotor. Als
hoofd van de onderzoeksafdeling had hij met een groot aantal verschillende projecten en
onderzoekers te maken, maar door zijn enorme geheugen én persoonlijke betrokkenheid kon
hij altijd terughalen waar je mee bezig was. Een gesprek met Dinny voelde dan ook altijd
vertrouwd. Helaas is Dinny op de laatste dag van 2016 overleden aan de gevolgen van kanker.

Dankwoord |165
Beste Joost, heel erg bedankt dat je relatief laat in het traject toch bereid was om in te stromen
in het promotieteam. Jouw enorme kennis en ervaring op het gebied van artrose, revalidatie
én het doen van onderzoek hebben de laatste twee artikelen en de in- en uitleiding zeker naar
een hoger niveau gebracht. Ik heb veel geleerd van je concrete feedback en vind het
bewonderenswaardig dat je vrijwel altijd binnen 24 uur mijn mails beantwoordde, waar ter
wereld je ook was.
Lieve Cindy, jij bent de enige die als co-auteur aan alle vijf de artikelen van dit proefschrift
verbonden bent. Dat zegt eigenlijk al wel iets over onze band. Na mijn stage bij het NIVEL
verontschuldigde je je dat er op dat moment geen vacature was binnen het Paramed team.
Nog geen maand later belde je met het verzoek te sollicteren op een juniorfunctie binnen
LiPZ. Ik ben je enorm dankbaar voor deze kans. Mede dankzij jou kwam ik terecht in een
warm nest met gedreven onderzoekers dat echt als een team functioneerde. Je bent een echte
verbinder, zowel als persoon binnen het team als in je rol als leidinggevende tussen onderzoek,
beleid en beroepspraktijk. Je bent een voorbeeld voor me als het gaat om het enthousiasmeren
van mensen, kansen zien en het combineren van werk en gezin. Dank je wel voor je
betrokkenheid en support en ik vind het echt heel fijn om opnieuw lid te zijn van je
onderzoeksgroep in een andere setting.
Lieve Ilse, ook jou beschouw ik als één van de belangrijkste personen van mijn loopbaan tot
nu toe. Ik kan me mijn eerste dagen als LiPZ-onderzoeker nog goed herinneren. Na het rondje
door het gebouw en het verwerken van minimaal 30 nieuwe namen en gezichten doken we
samen achter de computer. Je liet me alle ins-en-outs van de STATA do-files zien die nodig
waren om het jaarbestand voor LiPZ te genereren. Dat was wel wat anders dan het registreren
van parameters van COPD-patiënten tijdens hun revalidatie… Maar door je enthousiasme en
schat aan ervaring op het dossieronderzoek voelde ik me al heel snel thuis op mijn nieuwe
plek. Doordat je alles top georganiseerd had, kreeg ik al snel ruimte om naast reguliere LiPZ-
werkzaamheden met innovatie aan de slag te gaan. Je nam me mee naar softwareleveranciers,
koepelorganisaties en het ministerie. Je hebt laten zien dat onderzoek veel meer (en leuker!)
is dan het analyseren van data en schrijven van publicaties. Ook toen je rol binnen het NIVEL
veranderde was je altijd bereid om even te helpen. Hoe vaak je niet gevraagd hebt of je nog
iets voor me kon doen met het oog op je rol als co-promotor kan ik niet zeggen. Je was er
altijd voor me. Dank je wel daarvoor! Ik heb echt wel moed moeten verzamelen om te
vertellen dat ik het NIVEL ging verlaten. Gelukkig zien en appen we elkaar nog regelmatig en
ik hoop ook echt dat dat zo blijft!

Met het afronden van mijn proefschrift komt er dan toch echt een einde aan mijn NIVEL-tijd.
Ik ben al even weg, maar ik voel me nog altijd verbonden met deze fijne plek. Lange tijd deelde
ik kamer 3.19 met Jacqueline en Margit. Wat was het een leuke tijd! Alledrie LiPZ-
onderzoekers en alledrie bezig met een sprokkelpromotie. Jacqueline is al even klaar en ik
hoop dat Margit ook nog gaat volgen. In mijn laatste NIVEL jaar was de samenstelling van 3.19
volop in beweging. Dat past ook wel bij een organisatie als het NIVEL en heb dat als positief
ervaren. Leontien, we konden lekker sparren over babykwaaltjes en peuteravonturen, want

166
onze jongens zijn ongeveer even oud. Ook de andere kamergenootjes wil ik bedanken voor
de gezellige tijd: Marijn, Daphne, Esther en Sabine.
Mijn eerste NIVEL-jaren maakte LiPZ onderdeel uit van het Paramed team. Dat het een hecht
en echt team was, blijkt wel uit de activiteiten die we ver na het uiteenvallen van Paramed nog
met z’n allen ondernomen hebben. Met als hoogtepunt toch wel de curlingworkshop in
Zoetermeer. Daniël, Karin, Wil, Chantal, Cindy, Corelien, Ilse, Jacqueline, Linda, Lisa en Margit
dank jullie wel voor de fijne tijd!
In 2013 werd LiPZ onderdeel van de NIVEL Zorgregistraties eerste lijn. Een heel nieuwe groep
mensen die echt wel aan elkaar en elkaars’ werkwijze moest wennen, maar het kernteam
vormde zeker wel een mooi geheel! Robert, als programmaleider had je de moeilijke taak om
een stel eigenwijze onderzoekers met verschillende achtergronden bij elkaar te brengen en er
één geheel van te smeden. Logisch dat het in dat proces nog wel eens botste, maar je liet je
niet uit het veld slaan. Dat waardeer ik enorm aan je. Ik ben je dankbaar voor het vertrouwen
dat je me gaf als persoon en je steun voor het relatief kleine onderdeel ‘fysio-/oefentherapie’
in het grote geheel van de Zorgregistraties. Lando, als projectleider en verantwoordelijke voor
de paramedische zorg binnen de Zorgregistraties hebben we de laatste jaren intensief
samengewerkt. Je zorgvuldigheid en precisie hebben zeker bijgedragen aan de kwaliteit van
onze publicaties. Je afscheidsspeech zal ik nooit vergeten. Dank je wel! Mark, Marieke, Karin
en Lisa, we zaten in hetzelfde schuitje: het was onze taak om gegevens binnen te halen uit de
deelnemende praktijken en via de juiste analyses iets te zeggen over zorg en gezondheid in de
eerste lijn. Dat de disciplines niet gelijk waren, was niet zo belangrijk. Dank jullie wel voor de
vele uurtjes LINEL-klein, LINEL-groot, website-overleg, peer-feedback en natuurlijk het spuien
over allerlei randzaken.
NIVEL Zorgregistraties zou echt niet kunnen bestaan uit alleen onderzoekers. Zowel
voorafgaand aan het onderzoekers-traject als aansluitend hieraan, zijn veel mensen onmisbaar.
Jan Gravestein, als expert én ervaringsdeskundige op het gebied van (LiPZ)-Beweegzorg heb
je me veel geleerd. Ik wist niets van datamodellen, entiteiten en csv bestanden maar al doende
konden we goed met elkaar communiceren. Dank je wel voor al je werk voor LiPZ en NIVEL
Zorgregistraties, je enthousiaste verhalen over je koor en je interesse voor mijn leven. Bram,
jij vormde echt de schakel tussen de ICT’ers en de onderzoekers. Waar de ICT-taal mij soms
echt boven de pet ging, kon jij de vertaalslag maken zodat ik toch weer aanhaakte. Onmisbaar
wat mij betreft. Marcus, vele uurtjes hebben we samen achter een scherm gezeten om de
websites van de Zorgregistraties en vergelijkbare projecten te vullen. Wat een gedoe soms
om alles precies goed te krijgen. Jij verloor nooit je geduld en daardoor kwam het altijd goed.

Toen ik in september 2016 het NIVEL inwisselde voor de Hogeschool Utrecht, was dat
inhoudelijk een grote verandering. Bij het NIVEL maakte ik onderdeel uit van een geoliede
machine, bij het lectoraat was alleen de bouwtekening nog maar gemaakt. In het begin was
deze vrijheid even wennen, maar het kwam goed van pas. Naast het opbouwen van een
netwerk en het initieren van onderzoeksprojecten kon ik tijd besteden aan de afronding van
mijn proefschrift. Corelien, je was daarin echt een droomvoorbeeld, maar je maakte me ook
wel zenuwachtig. Je had alles voor je promotie zo strak geregeld naast al je andere

Dankwoord |167
werkzaamheden en loopactiviteiten. Echt super knap! Ik ben heel blij dat ik eerst jouw paranimf
mag zijn en dat we daarna de rollen omdraaien. Andere HU-collega’s van de
kenniscentrumruimte, inhoudelijk stond mijn proefschrift al wel toen ik bij de HU begon, maar
in het proces heb ik veel steun van jullie gehad. De terugkerende vraag ‘heb je al een datum?’
heeft wel geholpen om vaart te maken. De lijst met namen is te lang om op te noemen, maar
ik hoop op nog heel veel gezellige uren in onze ietwat lawaaige ruimte, mét de imiddels al
traditionele dropjes op de vrijdagmiddag.

Éen persoon uit de bredere onderzoeksgroep ‘Fysiotherapiewetenschap’, waar het lectoraat

onderdeel van uitmaakt, wil ik nog in het bijzonder bedanken. Dat is Martijn. Martijn, jij was
begeleider van een werkgroep bij de master Fysiotherapiewetenschap en bezig met de
afronding van je promotie bij het NIVEL. Jij vormde de schakel naar mijn afstudeerstage bij het
NIVEL en legde zo de basis voor mijn wetenschappelijke loopbaan. Je vertrok bij het NIVEL
toen ik in dienst kwam, dus NIVEL-collega’s zijn we eigenlijk nooit geweest. Hoe grappig is het
dan dat we dat nu wel zijn binnen de mooie onderzoeksgroep van Cindy aan de HU/UMC.

Zoals ik aan het begin van dit dankwoord al schetste vormde dit promotie-traject een
natuurlijk onderdeel van mijn dagelijks leven. Familie en vrienden weten uiteraard dat ik
onderzoek doe op het thema Beweegzorg, maar niet persé dat dit deels verpakt is in een
promotie-onderzoek. Het voelt dan ook een beetje vreemd om mensen te bedanken voor
hun steun en betrokkenheid gedurende de afgelopen zes jaar. Ik ben hen los van dit proefschrift
dankbaar voor wie ze zijn en wat ze doen.

Een paar mensen wil ik toch apart noemen. Opa en oma, voor jullie is deze promotie een heel
bijzondere gebeurtenis. Het zou prachtig zijn geweest om dit moment met jullie allebei te
delen, maar helaas kan oma niet naar Utrecht komen. Pap, mam, jullie hebben Michel, Ninke
en mij van jongs af aan gestimuleerd om ons te ontwikkelen en door te zetten. Daar is deze
promotie denk ik wel een mooi voorbeeld van. Ninke, als ‘kleine zusje’ heb je altijd een
bijzondere plaats. Ik vind het super leuk dat ik jouw getuige mocht zijn toen je trouwde met
Bram en dat jij nu een speciale rol als paranimf hebt bij mijn promotie. Lieve Guus, Koen en
Stef, jullie hebben nog geen idee wat ‘promoveren’ is. Wat mij betreft blijft dat voorlopig ook
zo. Doen wat je leuk vindt en waar je anderen mee kan helpen heeft prioriteit. Als dat uitmondt
in het volgen van een studie of het doen van promotie-onderzoek is dat helemaal goed. Lieve
Jaap, jij hebt uiteraard het meest meegekregen van dit hele traject. Soms voelde ik me wel een
beetje schuldig als jij een keer lekker op de bank een filmpje zat te kijken terwijl ik aan het
typen was aan de keukentafel. Maar ‘gelukkig’ hebben we het laatste jaar best veel tijd samen
aan die keukentafel doorgebracht. Volgend jaar rond jij je parttime studie af, die je naast je
bijna fulltime baan in de ‘tropenjaren’ van ons leven gestart bent. Met die tropenjaren gaan we
nog even door, want we leven toe naar het hoogtepunt van het jaar: de geboorte van een
nieuw lid van ons prachtige gezin!

168
About the author
Di-Janne Barten was born on May 13th 1985 in Wageningen, the Netherlands and moved to
Opheusden, the Netherlands at the age of three years. She completed secondary school
(VWO) in 2003 at the Hendrik Pierson College in Zetten. Subsequently, she joined the
bachelor program Physical Therapy at the Utrecht University of Applied Sciences (2003-2007).
Her senior internships, respectively at a primary care practice in Wageningen and at a hospital
in Tiel, resulted in her first jobs at the domain of physical therapy. Shortly afterwards, these
activities were extended by therapeutic guidance of patients suffering from Chronic
Obstructive Pulmonay Disease at a primary care practice in Buren, Gelderland. Next to these
jobs, she completed the pre-master and master program Clinical Health Sciences, particularly
Physical Therapy Sciences at the Utrecht University Medical Center. During her internship at
the Netherlands Institute for Health Services Research (NIVEL) in Utrecht, the ‘contamination’
with osteoarthritis-research has happened. By conducting a systematic review on
psychometric properties of personalised measurement tools and validating a personalised
measurement tool in patients with hip/knee osteoarthritis, a sound basis for her PhD thesis
has been provided. After her graduation in 2010, she started as junior researcher at NIVEL on
the long-term project ‘NIVEL Primary Care Database’. By this project, she gained lots of
experience with developing data-extraction modules, projectmanagement, analyzing large
datasets, and writing (inter)national scientific publications both on the domain of health
services research as well as physical therapy. A small part of the data gathered by this project
has been used in her PhD-thesis. Additional data were gathered during a temporary assignment
at St. Maartenskliniek Nijmegen, a hospital specialized in orthopaedic care. In 2016, her
activities in clinical practice ended. Unfortnuately, the desired combination of applying
research, performing clinical care and ‘managing’ a family was not feasible any longer.

Since the autumn of 2016, Di-Janne has joined the researchgroup ‘Innovation of Human
Movement Care’ at the Utrecht Univerisity of Applied Sciences. Her research focuses on the
development and implementation of innovative physical therapy pathways, especially in
primary care settings, communities, and neighbourhoods. Interprofessional collaboration and
co-creation with stakeholders from a citizen’s or patient’s perspective, are important aspects
in her current research projects. In that respect, her current focus of research meets the
personalized focus of her PhD thesis in patients with hip/knee osteoarthritis.

About the author |171

List of publications
International publications

Barten DJA, Knoop J, Swinkels ICS, Peter WF, de Bakker DH, Veenhof C, Dekker J. One
size fits all in physiotherapy management of knee osteoarthritis? A cross-sectional, clinical
vignette study. Physiotherapy (under review)

Lankhorst NE, Barten JA, Meerhof R, Bierma-Zeinstra SMA, van Middelkoop M.

Characteristics of patients with knee and ankle symptoms accessing physiotherapy: self-
referral versus a general practitioner's referral. Physiotherapy. Submitted

Barten DJ, Smink A, Swinkels IC, Veenhof C, Schers HJ, Vliet Vlieland T, de Bakker DH,
Dekker J, van den Ende CH. Factors associated with referral to secondary care in patients
with osteoarthritis of the hip or knee after implementation of a stepped-care strategy. Arthritis
Care Res (Hoboken). 2016 May 9.

Kooijman MK, Barten DJ, Swinkels IC, Kuijpers T, de Bakker D, Koes BW, Veenhof C. Pain
intensity, neck pain and longer duration of complaints predict poorer outcome in patients with
shoulder pain - a systematic review. BMC Musculoskelet Disord. 2015 Oct 9;16(1):288

Barten DJ, Swinkels IC, Dorsman SA, Dekker J, Veenhof C, de Bakker DH. Treatment of
hip/knee osteoarthritis in Dutch general practice and physical therapy practice: an
observational study. BMC Fam Pract. 2015 Jun 27;16:75

Swinkels ICS, Leemrijse CJ, Barten JA, Veenhof C. Direktzugang zur Physiotherapie in den
Niederlanden = Accès direct à la physiothérapie aux Pays-Bas. Physioactive, 2014, nr. 6

Barten JA, Pisters MF, Huisman PA, Veenhof C, Takken T. Measurement properties of
patient-specific instruments measuring physical function. J Clin Epidemiol. 2012 Jun;65(6):590‐
601.

Barten JA, Pisters MF, Takken T, Veenhof C. Validity and responsiveness of the Dutch
McMaster Toronto Arthritis patient preference questionnaire (MACTAR) in patients with
osteoarthritis of the hip or knee. J Rheumatol. 2012 May;39(5):1064‐73.

Veenhof C, Huisman PA, Barten JA, Takken T, Pisters MF. Factors associated with physical
activity in patients with osteoarthritis of the hip or knee: a systematic review. Osteoarthritis
Cartilage. 2012 Jan;20(1):6-12.

List of publications |175

National publications

Nagelmaeker J, Molenaars R, Barten D. Fysiotherapeutische zorg voor patienten met

schouderklachten. Hoe functioneert het SchouderNetwerk. Fysiopraxis 2017 Mei 26(4):24-25

Barten JA, Koppes LJ. Zorg door de fysiotherapeut; jaarcijfers 2015 en trendcijfers 2011-
2015. Utrecht, NIVEL, 2016.

Verberne LDM, Barten JA, Koppes LJ. Zorg door de fysiotherapeut; jaarcijfers 2014 en
trendcijfers 2010-2014. Utrecht, NIVEL, 2015.

Barten JA, Verberne LDM, Koppes LJ. Zorg door de fysiotherapeut; jaarcijfers 2013 en
trendcijfers 2009-2013. Utrecht, NIVEL, 2015.

Kooijman MK, Verberne LDM, Barten JA, Leemrijse CJ, Veenhof C, Swinkels ICS. Jaarcijfers
2012 en trendcijfers 2008-2012: fysiotherapie. Utrecht: NIVEL, 2013.

Verberne LDM, Kooijman MK, Barten JA, Swinkels ICS. Jaarcijfers 2011 en trendcijfers 2007-
2011: fysiotherapie. Utrecht: NIVEL, 2012.

Barten JA, Swinkels ICS, Veenhof C. Patiëntkarakteristieken, het gezondheidsprobleem en

het fysiotherapeutische behandelproces bij patiënten met hoofd-, hals‐ en nekklachten.
Jaarboek Fysiotherapie & Kinesietherapie, 2012.

Kooijman MK, Barten JA, Swinkels ICS, Veenhof C. Jaarcijfers 2010 en trendcijfers 2006-
2010: fysiotherapie. Utrecht: NIVEL, 2011.

Barten JA, Swinkels ICS, Kooijman MK, Veenhof C. Hoe uiten klachten waarmee patiënten
bij de fysiotherapeut komen zich? Utrecht: NIVEL, 2011.

Kooijman MK, Swinkels ICS, Barten JA, Veenhof C. Fysiotherapeutisch zorggebruik door
patiënten met een chronische aandoening in de periode 2006-2009. Utrecht: NIVEL, 2011.

Barten D, Pisters M. Patiënt-specifiek evalueren van fysiek functioneren: MACTAR.

Fysiopraxis; 3, 2011.

Barten D, van Bloemendaal M, van Dijk S, van den Dool J, van der Giessen R, Harmelink K,
Hombergen S, van Huffelen S, Huisman P, Kokkeler A, Lap M, Oosting E, Prinsen E, Ruigrok
C, Sickler C, Slootweg L, van der Torre P, van der Veen R, Westeneng J, Wind E, Wondergem
R, Zagers C, Zinger D, Zoethout C, Speksnijder C. Functionele bekostiging: kansen en
bedreigingen voor de fysiotherapie. Fysiopraxis aug 2010: 33

176
Conference abstracts

Barten D, Koppes L, Verheij RA. Self-referral has become the most common mode of access
to Dutch primary care physical therapy. World Congress Physical Therapy: Cape Town, July
4, 2017 (Oral presentation).

Lankhorst N, Barten D, Meerhof R, Bierma-Zeinstra S, van Middelkoop M. Physical therapy

for knee and ankle complaints: direct access compared to a referral from general practice.
World Congress Physical Therapy: Singapore, May 2, 2015 (Oral presentation).

Barten D, Smink A, Swinkels I, Veenhof C, Schers , de Bakker D, Dekker J, van den Ende C.
Factors associated with the setting and content of care in patients with osteoarthritis of the
hip or knee. World Congress Physical Therapy: Singapore, May 4, 2015 (Poster presentation).

Barten D, Lanson S, Lucas C, Veenhof C. Demographics and clinical characteristics in patients

with hip/knee osteoarthritis treated in primary versus secondary care considering their
physical function. World Congress Physical Therapy: Singapore, May 3, 2015 (Poster
presentation).

Barten D, Leemrijse CJ, Swinkels ICS. Direct access to physical therapy in the Netherlands.
Journées Francophones de Kinésithérapie (JFK): Lille, February, 2015 (Oral presentation).

Barten D, Smink A, Swinkels ICS, Veenhof C, Schers HJ, de Bakker DH, Dekker J, van den
Ende CH. Factors associated with the setting and content of care in patients with osteoarthritis
of the hip or knee. European League Against Rheumatism (EULAR): Paris, June, 2014 (Poster
presentation).

Barten D, Swinkels ICS, Dorsman SA, Veenhof C. Heup-/ knieartrose bij de huisarts en de
fysiotherapeut. Koninklijk Nederlands Genootschap voor Fysiotherapie Congres: Maastricht,
November 2, 2012 (Oral presentation).

Barten D, Swinkels ICS, Dorsman SA, Veenhof C. Osteoarthritis of the hip and/or knee in
Dutch general practice and physiotherapy practice. Osteoarthritis Reseach Society
International (OARSI): Barcelona, April, 2012 (Poster presentation).

Barten D, Pisters M, Takken T, Veenhof C. Validity and responsiveness of the Dutch

McMaster Toronto Arthritis Patient Preference Questionnaire (MACTAR) in patients with
osteoarthritis. World Congress Physical Therapy: Amsterdam, June 21, 2011 (Poster
presentation).

List of publications |177

Barten D, Pisters M, Takken T, Veenhof C. Validity and responsiveness of the Dutch
McMaster Toronto Arthritis Patient Preference Questionnaire (MACTAR) in patients with
osteoarthritis. European League Against Rheumatism (EULAR): London, May, 2011 (Poster
presentation).

178