Latent transitions of ego depletion during pharmacological therapy in newly diagnosed rheumatoid arthritis patients: implications for self-management behaviors

Objective To delineate latent transition patterns of ego depletion during csDMARD induction in newly diagnosed rheumatoid arthritis (RA) and their association with self-management in this observational cohort. Methods In this single-centre cohort, 255 newly diagnosed RA patients were enrolled and 230 completed assessments at baseline (T0), 3 (T1) and 6 months (T2). Ego depletion (SEDS) and self-management (RASMS) were evaluated. Latent profile analysis was conducted at each occasion and latent transition analysis modelled T1 → T2 transitions. Classification uncertainty was propagated with the Bolck–Croon–Hagenaars three-step method, and modal assignment served as a sensitivity analysis. Associations with T2 self-management were estimated with lagged models adjusted for the baseline score and for time-varying treatment and disease-activity covariates. Results Two latent classes were identified at T0 (low depletion 58.0%, moderate 42.0%) and three at T1 and T2 (low 25.2% and 32.2%, moderate 47.0% and 45.2%, high 27.8% and 22.6%). Entropy was 0.873–0.892; average posterior probabilities were 0.897–0.938, with classification error of 9.7% at T1 and 8.7% at T2. From T1 to T2 the moderate class was the most stable (88.89%) and the high class the least stable (65.63%); 13.9% of patients moved to a lower and 4.3% to a higher depletion class. Three transition cells contained 2–4 participants and were pooled. The T1 high depletion class was associated with a lower T2 total self-management score ( β = −14.32, 95% CI: −17.86 to −10.78, P < 0.001). Stable high depletion was associated with low-level self-management (OR = 4.106, 95% CI: 1.724–9.778) and poor medication adherence (OR = 4.482, 95% CI: 1.926–10.432); exposure and outcome were concurrent, so these are not temporal predictions. Conclusions In this exploratory, single-centre cohort, ego depletion was heterogeneous and partly reversible during csDMARD induction. Persistent high depletion was associated with poorer self-management and medication adherence, but the observational and partly concurrent design precludes causal inference. Because only the T1–T2 interval was modelled, we cannot conclude that 3–6 months is uniquely a critical intervention window. External validation and intervention trials are required before targeted monitoring or psychological empowerment can be recommended.