Investigating outcome inequalities in osteoarthritis management programmes: an analysis of registry data from the good life with osteoarthritis in Denmark (GLA:D®) programme using tapered balancing

Abstract

Purpose: The burden of osteoarthritis (OA) is often greater among disadvantaged people and communities, prompting calls for more attention to equity-focussed research and policy. A specific concern is whether healthcare interventions may inadvertently widen health inequalities. OA management programmes (OAMP) have emerged in the past decade in a major international effort to improve provision of core non-surgical care for people with OA. Recent studies have focussed on equity of access. We address a complementary issue: having gained access, do people from socially disadvantaged groups have poorer outcomes than their advantaged counterparts, and if so, what might determine this?

Methods: The study population was consecutive adults with knee OA attending the 8-week GLA:D® supervised exercise and education programme in Denmark between Oct 2014-Feb 2018. We defined a ‘multiple social disadvantage’ group based on primary/secondary school education and being either born outside Denmark or not having Danish citizenship. Their outcomes were compared with those of native Danish citizens with higher education. Outcomes of interest were pain intensity (0-100 VAS), KOOS Quality of Life subscale (QOL 0-100), EQ-5D-5L health utility (-0.624-1.0) at 3 and 12 months. Missing data were imputed using multiple imputation with chained equations. We used Coarsened Exact Matching (CEM) to restrict group comparisons to areas of common support, i.e. sufficient overlap on key prognostic factors (age, sex, body mass index, baseline value of the outcome measure of interest). We then used Entropy Balancing to sequentially control for differences between disadvantaged and advantaged groups in: (1) baseline value of the outcomes of interest (2) type of treatment centre, enrollment year (3) age, sex (4) BMI, previous knee injury, previous knee surgery (5) no. of selected comorbidities, no. of other non-knee pain sites (6) self-efficacy score, self-reported presence of depression (7) previous/current tailored exercise advice, weight loss counselling, analgesia/natural remedies (8) attendance at GLA:D® education and exercise sessions. Mean differences in outcomes between disadvantaged and advantaged groups were then estimated by weighted linear regression without balancing and then with entropy balancing weights from steps 1-8.

Results: Of 18,448 eligible participants, 250 (1.4%) were classed as disadvantaged. Compared with advantaged participants, they were younger, less likely to have attended GLA:D® in a private physiotherapy clinic, reported more comorbidity, pain sites, depression, lower self-efficacy, and lower attendance on GLA:D® sessions. Both groups showed overall improvements over baseline in mean pain VAS, KOOS QOL and EQ5D scores at 3 months, typically maintained at 12 months. Before covariate balancing, disadvantaged participants had substantially worse scores than advantaged participants on each measure at both follow-up points (e.g. crude between-group mean differences (95%CI) in pain VAS at 12 months: 8.6 (4.5, 12.6) respectively: Table 1). Balancing for differences on baseline score, comorbidity, self-efficacy, and depression had the greatest effect on reducing differences in outcomes.

Conclusions: Both disadvantaged and advantaged adults with knee OA reported improvements in key outcomes up to 12 months after OAMP attendance. However, compared with more advantaged adults, disadvantaged adults typically began the OAMP with more severe pain and poorer quality of life. This gap in outcomes was not reduced following OAMP attendance: for generic health-related quality of life in particular the gap widened slightly. Low self-efficacy, depression and other comorbidities may be potential determinants. Our analysis is of observed outcomes in a relatively small group with multiple disadvantage attending one OAMP. We encourage further work in other settings and groups. If confirmed, our findings suggest that while improving access to OAMPs for socially disadvantaged people with OA is important, additional actions may be needed to reduce outcome inequalities.