Artificial Intelligence Causal Inference of Community Clinic Postpartum Haemorrhage Risk Under Unmeasured Misoprostol Tablet Quality Degradation: A Healthcare Instrumental Variable Safety Evaluation

Abstract

Community clinic protocols for postpartum haemorrhage (PPH) prophylaxis rely on misoprostol tablets whose active-ingredient content is unobserved at the point of care, so that the protective effect of the drug is confounded by an unmeasured quality gradient that varies across batches, districts and procurement cycles. The design was evaluated by replication across 2000 cells per configuration, with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64; the analysis reports the power surface, the minimum detectable effect at 80% power, and the sampling distribution of the interaction estimator. At an intraclass correlation of 0.64, 56 clusters are required to detect a 5-percentage-point effect at 80% power, and the minimum detectable effect falls steeply as cluster count rises from 8 to 32. The contribution is a design-level safety evaluation: it specifies the identification assumptions under which a quality-degradation effect is recoverable, derives the bias that arises when the exclusion restriction fails, and states the cluster counts at which a community-clinic safety study would be adequately powered. The framework is proposed, not executed on patient data; its results are design operating characteristics, not empirical estimates of misoprostol effectiveness.