Title: A direct comparison of the reliability and validity of patientcentred measures of atwork productivit
1A direct comparison of the reliability and
validity of patient-centred measures of at-work
productivity loss in persons with inflammatory
and osteoarthritis.
- Dorcas Beaton, BScOT, PhD
- Scientist,
- St Michaels Hospital and the Institute for Work
Health - University of Toronto, Toronto, Ontario, Canada
2Co-investigators
- Claire Bombardier
- Diane Lacaille
- Monique Gignac
- Anusha Govinda-Raj
- Sherra Solway
3Work
4At work productivity loss
- But what is happening at work?
- Greatest portion of lost productivity costs due
to arthritis gtgt absenteeism costs (Li, 2005) - Understudied, inconsistently measured.
- Step 1 Measure it well
- Aim economic ( cost) and impact on QOL state
(state description)
5Review of existing literature
- 16 different scales
- 5 direct comparisons
- Different levels of work limitations (Lavigne,
2003) - Low correlations between measures
- Little information in arthritis
HLQ, Osterhaus (Lavigne, 2003)
WLQ25, WPSI (Ozminkowski, 2004)
Osterhaus, QQ (Brouwer, 1999)
WLQ25, SPS13 (Turpin, 2004)
QQ, HLQ (Meerding, 2005)
6Objectives
- To directly compare the measurement properties of
five different scales in persons with IA and OA - Design
- Cross sectional survey, n250 persons, 3 clinical
sites - 130 OA, 120 RA
- Working for pay in the past month
- Able to complete questionnaire in english
- Completed by either touch screen or paper
(randomly assigned to each disease group)
7Measures
- WIS-RA - Work instability scale (Gillworth, 2003)
- 23 items, one score, yes/no response,
difficulties encountered - WLQ-25 (Work Limitations Questionnaire Lerner,
2002) - 25 items, 4 domains, 5 response options, time in
difficulty - WALS (Work Activity Limitations Scale, Gignac,
2004) - 11 items, one domain, 4 response options,
difficulty - EWPS (Endicott Work Productivity Scale) (1997)
- 25 items, one domain, 5 response options,
frequency - SPS-6 (Stanford Presenteeism Scale-6)(Koopman,
2002) - 6 items, one domain, 5 response options, degree
of agreement with limitation
8Results
All
IA
OA
- N 250 130 120
- Age (mean (sd)) 50.4 (9.7) 54.0 (6.7) 46.2 (10.9)
- Gender (F) 82.7 80.5 85.3
- Duration 54.7 49.2 61.1
- ( gt 5 yrs)
- comorbidities 1.9 (1.1) 2.2 (1.1) 1.6 (0.9)
- HAQ-DI 0.76 (0-2.25) 0.76 (.58) 0.76 (.56)
9EWPS
21.5/100 (13.9)
WLQ-25
SPS-6
27.8/100 (18.7)
13.3/30 (5.1)
WALS
WIS
8.0/33 (5.1)
8.3/23 (6.4)
Note all scales transformed to 0-100, 100 more
limitation for graphs
10Scores IA versus OA (pgt0.05)
Raw scale scores
11Reliability Internal consistency
Cronbachs alpha or KR-20
12Construct validity
- Correlations with 5 global markers of work
ability (expected gt 0.75) - Spearman rho 0.5-0.7, WIS was 0.8
- Correlations slightly higher (0.1) in OA gt RA
- Correlations between scales (expected gt 0.75)
- Spearman rho 0.2 0.7
- Median 0.5
- Lower than expected
- WLQ-PD scale consistently low (0.3)
13Construct validity discrimination
Discrimination between those hindered (46) or
not hindered (54) at their work
Unpaired t-statistic (mean difference / se )
(All significant at plt0.05)
14Arthritis consumer preference
RA OA
Preferred scale ( of sample)
Order in survey ?
15Conclusion
- Reliability okay (SPS, WLQ-25 problematic)
- Validity lower, WIS ranked highest
- Slight differences OA/RA should not be of
concern. - More concerning
- Different scales ? different things.
- Problems with PD subscale WLQ ? reversed scale?
- ? Need to use 2 scales match patient
preference! - Economic analyses ? time
- Outcome state ? difficulty
16Future
- Longitudinal construct validity
- Responsiveness
- Same cohort ? one year
- Collaborations internationally
- On agenda at OMERACT 9 (May 2008)
- How to measure productivity loss
- How to blend off-work, at-work outcomes
- How to measure the contextual nature of work
17Thank you
- Acknowledgements
- This study was funded by the Canadian Arthritis
Network - Dr Beaton is supported by a New Investigators
award from CIHR - OMERACT is supporting collaboration on work
disability measurement - Thank you to the research staff, and the patients
who participated in this study