Systematic review reveals methodological gaps in DiD applications for health systems adoption.
Wide confidence intervals in effect estimates indicate substantial uncertainty in findings.
Event-study plots commonly used but often lack accompanying formal statistical tests.
Stricter methodological reporting standards are needed for credible policy evaluation.
Abstract
{ "background": "The difference-in-differences (DiD) model is a prominent quasi-experimental technique for evaluating policy impacts in health systems research. Its application to assess the adoption of new systems, such as electronic health records or financing reforms, in Kenyan district hospitals requires methodological scrutiny to ensure validity and inform future evaluations.", "purpose and objectives": "This systematic review aims to critically appraise the application of the DiD model in studies evaluating health systems adoption in Kenyan district hospitals. It seeks to assess model specification, identification assumptions, and robustness checks within this specific context.", "methodology": "A systematic search of multiple academic databases was conducted following PRISMA guidelines. Peer-reviewed studies employing DiD to evaluate health systems adoption in the specified setting were included. Data were extracted on study design, model specification, and validation of the parallel trends assumption, a core DiD prerequisite expressed as $E[Y{0it} - Y{0is} | Di=1] = E[Y{0it} - Y{0is} | Di=0]$ for pre-adoption periods s