BackgroundObservational data in the United States suggests that those without health insurance have a higher mortality and worse health outcomes. A linkage between insurance coverage and outcomes in hospitalized patients has yet to be demonstrated in resource-poor settings.MethodsTo determine whether uninsured patients admitted to the public medical wards at a Kenyan referral hospital have any difference in in-hospital mortality rates compared to patients with insurance, we performed a retrospective observational study of all inpatients discharged from the public medical wards at Moi Teaching and Referral Hospital in Eldoret, Kenya, over a 3-month study period from October through December 2012. The primary outcome of interest was in-hospital death, and the primary explanatory variable of interest was health insurance status.ResultsDuring the study period, 201 (21.3%) of 956 patients discharged had insurance. The National Hospital Insurance Fund was the only insurance scheme noted. Overall, 211 patients (22.1%) died. The proportion who died was greater among the uninsured compared to the insured (24.7% vs. 11.4%, Chi-square=15.6, p<0.001). This equates to an absolute risk reduction of 13.3% (95% CI 7.9–18.7%) and a relative risk reduction of 53.8% (95% CI 30.8–69.2%) of in-hospital mortality with insurance. After adjusting for comorbid illness, employment status, age, HIV status, and gender, the association between insurance status and mortality remained statistically significant (adjusted odds ratio (AOR)=0.40, 95% CI 0.24–0.66) and similar in magnitude to the association between HIV status and mortality (AOR=2.45, 95% CI 1.56–3.86).ConclusionsAmong adult patients hospitalized in a public referral hospital in Kenya, insurance coverage was associated with decreased in-hospital mortality. This association was comparable to the relationship between HIV and mortality. Extension of insurance coverage may yield substantial benefits for population health.

Among adult patients on the medical wards, patients admitted on weekends had similar mortality rates to those admitted on weekdays. This similarity may reflect a stable level of care or a generalized shortage of resources and staffing that subsumes any impact of weekly variations. Future research examining optimal staffing and resource levels is needed in such settings.