Abstract Background/Aims Work disability is a common early outcome of inflammatory arthritis (EIA), with around 10% of patients leaving employment within three months of diagnosis¹. Interventions exist but access is limited, creating a need for effective risk stratification. This proof-of-principle study tested whether routinely collected clinical and occupational data could inform a practical tool to identify patients at highest risk. Methods This cohort study used data from patients with EIA enrolled in the National Early Inflammatory Arthritis Audit between 2018 and 2025. Eligible participants were employed at diagnosis and had three-month work outcome data. Self-reported employment loss within three months of diagnosis was analysed using Poisson regression. Predictors were selected using penalised regression (LASSO/elastic net with 10-fold cross-validation) alongside clinical feasibility, considering demographics, disease characteristics, comorbidities and job demands. Candidate models were compared sequentially, with the final specification chosen for performance and routine applicability. Regression coefficients from the optimal model were converted into a simplified integer-based risk score, from which risk strata were defined. Results Of 11,894 patients, 1,662 were employed at baseline and had complete data; 168 (10.1%) reported work loss within three months. Age was the strongest predictor (IRR/decade 1.65, 95%CI 1.43-1.90), alongside manual work (1.54, 1.15-2.06), poor musculoskeletal health (1.55, 1.10-2.19) and anxiety/depression (1.45, 1.02-2.06). DAS28 5.1 was not independently associated with work loss. Penalised regression approached initially selected seven variables from the candidate set: physical job demands, age, sex, disease activity, musculoskeletal health, mental health and diagnosis type. Diagnosis type contributed no predictive value and was excluded from the final model specification. Standard logistic regression performed best (C-statistic:0.708), consistently supporting four key predictors: age, physical job demands, musculoskeletal health and mental health with good calibration and minimal overfitting (bootstrap-corrected:0.705). An 8-point risk score stratified patients into low, medium and high-risk groups, with a seven-fold gradient between extremes. Conclusion Employment loss in EIA is driven by occupational demand, age, musculoskeletal symptoms and mental health, while disease activity showed no independent effect. We developed a proof-of-principle pragmatic tool to stratify risk, enabling early identification of patients needing occupational support and targeted interventions for those at greatest risk. Disclosure E. Alveyn: None. J. Galloway: Grants/research support; Versus Arthritis, NIHR. M. Adas: None. P. Amlani-Hatcher: None. M. Dey: None. S. Gallagher: None. M. Gibson: None. B.E. Jones: None. D. Mehta: None. S. Norton: Grants/research support; NIHR. E.J. Price: None. M. Russell: None. K. Walker-Bone: None. E. MacPhie: None. K. Bechman: None.
Alveyn et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: