"background": "District hospitals are critical nodes in Nigeria's healthcare system, yet robust evidence on the effectiveness of systemic interventions remains sparse. Quasi-experimental designs (QEDs) offer a pragmatic approach for evaluating clinical outcomes in such complex, real-world settings where randomised controlled trials are often infeasible. ", "purpose and objectives": "This systematic review aims to methodologically evaluate the application of quasi-experimental designs in studies measuring clinical outcomes within Nigerian district hospital systems, assessing their rigour, appropriateness, and reporting standards. ", "methodology": "A systematic search of multiple electronic databases was conducted following a pre-registered protocol. Peer-reviewed studies employing QEDs (e. g. , interrupted time series, difference-in-differences, regression discontinuity) to evaluate hospital-level interventions and report clinical outcomes were included. Study quality was appraised using the ROBINS-I tool. A meta-regression model, Yi = \0 + \1X{1i + \2X2i + ui, was specified to explore associations between design features and reported effect sizes, where Yi is the standardised effect, X1i is a rigour score, and X2i indicates the use of propensity score matching. ", "findings": "Of the screened records, 27 studies met the inclusion criteria. The methodological rigour was highly variable; a predominant theme was the inadequate handling of confounding and secular trends. Only a minority (approximately 26%) of studies employed robust methods like instrumental variables to address endogeneity. The meta-regression indicated a positive association between methodological rigour and more conservative effect estimates (95% CI: -0. 41 to -0. 12). ", "conclusion": "The application of quasi-experimental designs in this context is increasing but often methodologically suboptimal, limiting the reliability of causal inferences about hospital system interventions on clinical outcomes. ", "recommendations": "Future research should prioritise the use of more
Okonkwo et al. (Fri,) studied this question.