Community Mental Health Service Multimorbidity Cluster Randomisation and Psychotropic Polypharmacy Reduction in Ugandan District Clinics: A Stepped-Wedge Estimation of Prescribing Cascade Exposure

Psychotropic polypharmacy in Ugandan district mental health clinics is frequently attributed to prescribing cascades, in which a drug prescribed for an adverse effect of a first drug is counted as a new clinical need. Stepped-wedge cluster designs are increasingly proposed to evaluate cascade-reduction interventions, but their operating characteristics under the high intraclass correlation typical of prescriber-level clustering remain poorly characterised. Under a declared variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64, the analysis characterises power, minimum detectable effect and interval coverage. Fifty-two clusters are required to detect a five-percentage-point reduction at eighty per cent power. The results indicate that the design is feasible only for large effects at the cluster counts typically available to district health authorities, and that the intraclass correlation of 0.64 dominates the power calculation. Stepped-wedge evaluations of prescribing-cascade interventions should be planned around the intraclass correlation rather than the nominal sample size, and it specifies the design conditions under which a district-level trial can credibly detect polypharmacy reduction.