OBJECTIVES : The disability levels of osteoarthritis (OA) patients may fluctuate over time in response to disease activity, medical and personal circumstances. This study aims to generate annual transition probabilities for OA-related disability states using a 10-year prospective population-based Australian cohort.

METHODS : 304 participants with self-reported OA diagnosis from the Tasmanian Older Adult Cohort (TASOAC) aged 50-80 years, with longitudinal data on disability severity (collected at baseline, 2.5 years, 5 years and 10 years) were included. Disability severity was assessed using the Health Assessment Questionnaire Disability Index (HAQ-DI), annual transition probabilities were obtained between no disability (HAQ-DI level 0), mild (0<HAQ-DI ≤1), moderate (1<HAQ-DI ≤2) and severe (HAQ-DI>2) disability. A four-state continuous-time Markov model was used to describe the progression of disability and the analysis was performed using the ‘msm’ package in R. Separate transition probability matrices were also derived for each category of five binary covariates: age, sex, OA affected sites, body mass index (BMI) and number of comorbidities.

RESULTS : From no disability, 8.5% (95% confident interval [CI]: 6.9–10.7) progressed annually to mild disability and 0.2% (0.2–0.4) to moderate disability (0.0% transitioned from no disability to severe disability). From mild, 9.3% (7.3–11.9) improved to no disability annually, and 4.9% (3.5–6.6) worsened to moderate disability. From the moderate state, 12.6% (8.4–18.8) improved to mild disability annually and 0.7% (0.4–1.1) to no disability; and 0.2% (0.0–13.8) worsened to severe disability. Once in the severe state, the transition probability to improve was 0%. Baseline older age and >=2 comorbidities were related to higher probabilities of worsening disability. Probabilities did not considerably differ when broken down by BMI, sex and OA affected sites.

CONCLUSIONS : Our estimates suggest that disability progression in the OA sample was slow, and probabilities often differed between subgroups of participants. Our results can be helpful in projecting the disability progression in individuals with OA.