Diagnostic accuracy and patient outcomes cannot be understood in isolation when care is delivered through varied hospital pathways with different referral patterns, test availability and follow-up capacity. This methodological article develops a biostatistical framework for linking clinical pathways, test accuracy and outcomes across African hospitals without requiring new primary data collection. It argues that the appropriate estimand is pathway-conditional accuracy and pathway-mediated outcome risk, rather than a single pooled sensitivity, specificity or survival probability. The proposed approach combines hierarchical models for clustered binary and time-to-event outcomes, bivariate characterisation of sensitivity and specificity, explicit treatment of verification and missing-data mechanisms, and internal-external validation for transportability. The framework specifies what assumptions identify each quantity, what routine hospital records would permit estimation, and how reporting standards can make heterogeneity interpretable rather than averaged away.