Digital self-management engagement and long-term patient-reported outcome trajectories in inflammatory arthritis: a real-world retrospective cohort study

Abstract Introduction/objectives Patient Global Assessment of Disease Activity (PtGADA), pain, and Routine Assessment of Patient Index Data 3 (RAPID3) are core patient-reported outcomes in inflammatory arthritis. We examined whether digital self-management engagement was associated with their long-term trajectories. Method This retrospective cohort included 3513 adults using a certified digital self-management application. Patients were grouped by completion of 0, 1–2, or ≥3 structured self-management plans. Longitudinal models with patient-level cluster-robust standard errors estimated post-baseline change trajectories. Sensitivity analyses addressed differential attrition, measured confounding, and reverse causality. Results Of 2895 patients with baseline PtGADA, 1026 (35.4%) contributed post-baseline observations. Patients completing ≥3 plans had more favourable PtGADA, pain, and RAPID3 trajectories than zero-plan patients (group-effect p =6.2×10 −5 , p =0.0055, and p =0.0044, respectively), without diverging slopes. At weeks 12–48, PtGADA improved by 11.1–12.1 points and pain by 7.2–7.3 points in the high-engagement group. After attrition weighting, PtGADA remained significant, RAPID3 retained a significant joint group effect, and pain was no longer significant. Conclusions Higher engagement was most robustly associated with more favourable long-term PtGADA trajectories; support for pain was directionally similar but statistically more limited, and RAPID3 findings were the least consistent of the three outcomes. These results describe an engagement-associated phenotype rather than a causal treatment effect and support its evaluation as a potential marker of sustained digital engagement in future prospective stratification studies. Key Points • In this real-world cohort of 3513 patients with inflammatory arthritis, higher digital self-management engagement was most robustly associated with more favourable PtGADA trajectories; the association with pain was directionally similar but less robust, while RAPID3 findings were less consistent. • Engagement-group trajectories were approximately parallel, and the findings remained directionally consistent in analyses examining sensitivity to differential follow-up, measured baseline differences, and reverse-causality concerns, although statistical significance varied by outcome and landmark. • The engagement-outcome association was directionally consistent across multiple sensitivity analyses addressing measured confounding and differential follow-up, although temporal exposure assignment could not be established and residual reverse-causality bias cannot be excluded. • Digital self-management engagement, operationalised by action-plan completion, represents a potential marker of sustained digital engagement for future prospective stratification studies, rather than a validated tool for baseline risk stratification.