Clinical governance and staff-perceived healthcare quality in conflict-affected settings: a cross-sectional study of hospitals in Yemen

Background Evidence on clinical governance (CG) implementation and its relationship with service quality in fragile and conflict-affected settings is scarce. This analytical cross-sectional study assessed CG implementation, examined its association with staff-perceived healthcare quality (HCQ), and evaluated public–private differences in accredited, relatively stable hospitals in Sana’a, Yemen. Methods The study was conducted in eight hospitals (five public, three private) between June and October 2025. A total of 382 multidisciplinary healthcare workers, selected from 2024 payroll records by proportionate stratified random sampling, completed a context-specific, self-administered questionnaire measuring CG and HCQ. Structural validity was assessed using confirmatory factor analysis. Associations were examined using hierarchical multiple linear regression and two-level linear mixed-effects models with hospital-level random intercepts. Small-sample bias-corrected cluster-robust standard errors (CR2) were used as a sensitivity analysis. Results CG correlated strongly with HCQ ( $$r = 0.764$$ , $$p < 0.001$$ ), and its seven domains accounted for 59.7% of the variance in HCQ ( R² = 0.597 ; adjusted R² = 0.589 ). Between-hospital clustering was substantial ( ICC = 0.278 ). In fully adjusted mixed models, CG remained positively associated with HCQ ( $$B = 0.678$$ , SE = 0.039 , 95% CI: 0.606–0.758, $$p < 0.001$$ ). Information Technology Management ( $$B = 0.238$$ , $$p = 0.003$$ ) and Risk and Safety Management ( $$B = 0.124$$ , $$p = 0.018$$ ) showed significant independent associations with HCQ. Private hospitals had higher unadjusted CG and HCQ scores ( $$p < 0.001$$ for both), but the difference was not significant in mixed models ( $$B = 0.109$$ , SE = 0.125 , $$p = 0.412$$ ) or in the CR2 sensitivity analysis ( $$B = 0.064$$ , $$p = 0.395$$ ). Gender, qualification, and professional experience were not independently associated with HCQ. Conclusions CG was positively associated with HCQ in the participating hospitals, with Information Technology Management and Risk and Safety Management showing significant independent associations. Unadjusted sectoral differences favoring private hospitals were not significant after adjusting for CG and hospital-level clustering. Findings are cross-sectional, rely on self-reported perceptions, and may not be generalizable to remote or highly disrupted facilities in Yemen. Future research should incorporate objective quality indicators, broader facility sampling, and sample-size planning that accounts for hospital-level clustering.