O16 Development of risk calculators for hand osteoarthritis and invasive treatment

Abstract Aim Develop a prediction model for incident radiographic IPJ osteoarthritis when multiple clinical risk factors are present. Methods This study used secondary analysis from the Chingford 1000 Women Study, a prospective cohort of women aged 45 to 64 years. At baseline, anthropometric, clinical, and lifestyle measured had been collected. Hand radiographs had been taken at baseline and 10 years, read using the Kellgren-Lawrence atlas. For the current study, participants with osteoarthritis (Kellgren-Lawrence ≥2) in any IPJ at baseline were excluded. Risk factors were selected from baseline based on biological plausibility, a published systematic review, and a Delphi study of Hand Surgeons. Incident osteoarthritis was diagnosed at 10 years if ≥ 1 IPJ was Kellgren-Lawrence ≥2. The model was built with logistic regression and elastic net penalisation, and performance assessed through discrimination (c-statistic) and calibration (c-slope). Complete case analysis was used. Results Of 1,003 participants, 459 participants were included in this study (median age: 51 years, 202 (44%) with IPJ osteoarthritis at follow-up). Manual occupation (P < 0.01), base of thumb osteoarthritis (P < 0.03), and older age (using a 3-knot spline) were the most important risk factors. C-statistic was 0.67 (0.62 to 0.72) and C-slope was 1.00 (0.68 to 1.34). Conclusions Osteoarthritis at the base of thumb and IPJs might be a continuum of the same disease. Knowledge of these modifiable and non-modifiable risk factors can inform prevention strategies.