Community Mental Health Services Diagnostic Coding Inconsistency and Severe Mental Illness Prevalence Estimation in Ethiopian Primary Care: A Latent-Class Measurement Model

Diagnostic coding inconsistency in Ethiopian primary care threatens the validity of severe mental illness prevalence estimates derived from routine community mental health service records. A replication-based evaluation was conducted: 600 data sets of N = 500 were drawn from a process with intercept 12.0, exposure 0.42, group -3.1, severity 1.35 and error SD 4.2, using seed 2089151136. The estimator recovered the exposure coefficient with negligible bias (0.00049) and RMSE 0.01585, while the intercept showed bias -0.0147 and RMSE 0.88455; nominal 95% confidence interval coverage was 1.0 for both parameters. The results indicate that under the stated process, OLS estimation of the exposure–prevalence relationship is approximately unbiased and well calibrated, but that the intercept is sensitive to model specification. Latent-class correction of diagnostic coding error is a precondition for credible prevalence estimation, and that the measurement model, not the regression estimator, is the binding constraint on inference in Ethiopian primary care settings.