Artificial Intelligence Causal Transportability of Community Health Worker Triage Decision Support Under Variable Supervision Record Granularity: A Doubly Robust Evaluation of Healthcare Escalation Safety

Community health worker (CHW) triage decision support is increasingly proposed as a means of extending safe healthcare escalation into underserved districts, yet the safety of such systems is typically evaluated in pilot settings whose supervision records are more complete than those available at scale. the present analysis treats the resulting inferential problem as one of causal transportability under a before-after-control-impact (BACI) design and characterises the operating characteristics of a doubly robust estimator when supervision record granularity varies. No observational data are used or claimed; the analysis is a design-based evaluation under a declared variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64. Power was estimated across 2000 replications per design cell for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. Fifty clusters were required to detect a 5 percentage point effect at 80 per cent power. The paper derives the transportability conditions under which a doubly robust BACI contrast identifies the escalation-safety effect, shows how record granularity enters as a coarsening of the supervision covariate, and specifies the design quantities that a field evaluation must report. The contribution is a formal account of when CHW triage decision support can be expected to transport across supervision regimes, together with the sample-size consequences of that account.