Psychiatric Medication Stockouts and Community Mental Health Service Relapse Presentations in Ugandan Districts: A Distributed-Lag Panel Estimation

Psychiatric medication stockouts are widely assumed to drive relapse in community mental health services, yet the timing of that relationship remains poorly specified. The calibration model was estimated by ordinary least squares with heteroskedasticity-robust standard errors and achieved an R-squared of 0.4888 (adjusted R-squared 0.487; F = 441.152, p < 0.001). Electricity access carried a coefficient of 123.00084 (SE 4.65884) and urban population a coefficient of −1.34431 (SE 0.05763), with a further covariate at 0.46579 (SE 0.14039). These estimates are used strictly as a numerical analogue for the lag structure the psychiatric stockout model would require, not as evidence about mental health services. Stockout effects on relapse are likely to be distributed over several months, that contemporaneous correlation understates the cumulative burden, and that district panels with monthly stockout registers and presentation counts are the minimum design capable of identifying the lag weights. The contribution is a pre-specified estimation architecture and a set of falsifiable propositions for Ugandan district health information systems.