Community Mental Health Service Diagnostic Coding Drift and Antipsychotic Dose Escalation in Ghanaian District Clinics: An Interrupted Time-Series Estimation of Record-Keeping Exposure

Diagnostic coding drift in community mental health services is widely treated as a documentation nuisance rather than a clinical exposure. The framework specifies an interrupted time-series design in which coding drift is operationalised as the cumulative divergence between recorded and audited diagnoses, and dose escalation as the monthly change in defined daily doses per patient. Because no primary clinic-level data are available for this analysis, the paper does not report empirical estimates; instead it formalises the estimand, derives the identification conditions under which record-keeping exposure can be separated from secular prescribing trends, and specifies the sensitivity analyses required before the design is fielded. The framework is calibrated against the distribution of family planning messages at community meetings across fifteen African countries (N = 15, mean = 29.733, SD = 17.495, range 6.60 in Mali to 73.40 in Malawi), which illustrates the between-country heterogeneity that any multi-site exposure measure must accommodate. The contribution is a pre-specified, falsifiable estimation strategy that treats documentation quality as a modifiable service exposure rather than a background covariate.