{ "background": "The adoption of district hospital systems is a critical component of health system strengthening in Kenya, yet robust evidence on the effectiveness of interventions remains limited. Quasi-experimental designs offer a pragmatic approach for evaluating adoption in real-world settings where randomised controlled trials are not feasible.", "purpose and objectives": "This systematic review aims to critically appraise the application of quasi-experimental methodologies in studies evaluating district hospital system adoption in Kenya, assessing their design, analytical rigour, and the validity of their findings.", "methodology": "A systematic search of multiple electronic databases was conducted following PRISMA guidelines. Eligible studies employed quasi-experimental designs (e.g., difference-in-differences, interrupted time series, regression discontinuity) to evaluate the adoption of systems within Kenyan district hospitals. Study quality was assessed using the ROBINS-I tool for risk of bias.", "findings": "Of the 27 included studies, difference-in-differences was the most prevalent design (63%). A key methodological weakness was the frequent failure to test and account for parallel trends assumptions, with only 22% of relevant studies reporting formal tests. The pooled analysis suggests a positive, albeit heterogeneous, association between system adoption and improved supply chain metrics, with a meta-analysed odds ratio of 1.85 (95% CI: 1.42, 2.41). The relationship was modelled using a random-effects meta-regression: $\\log(OR{i}) = \\beta0 + \\beta1 X{1i} + ui + \\epsiloni$, where $u_i$ represents study-level heterogeneity.", "conclusion": "Quasi-experimental evaluations have generated important insights but are often methodologically suboptimal, limiting causal inference. Inconsistent reporting and inadequate handling of confounding and secular trends are major constraints on evidence quality.", "recommendations": "Future research must prioritise stronger quasi-experimental designs, transparent reporting, and the use of robustness checks, such as placebo tests and alternative model specifications with clustered standard