In-hospital mortality and survival outcomes among admissions with diabetes mellitus or hypertension at a tertiary hospital in southwestern Nigeria: a five-year retrospective cohort study

Abstract

Abstract Background Diabetes mellitus and hypertension account for a growing share of hospital admissions in sub-Saharan Africa, yet admission counts alone provide limited insight into how mortality risk unfolds after hospitalization. This study compared in-hospital survival and mortality hazards among admissions with a primary diagnosis of diabetes mellitus or hypertension at a tertiary hospital in southwestern Nigeria. Methods A retrospective cohort study was conducted using inpatient records from the Obafemi Awolowo University Teaching Hospitals Complex, Ile-Ife, Nigeria, covering admissions between 2017 and 2021. Each record represented a unique patient contributing one admission. Eligible admissions had a single recorded primary diagnosis of diabetes mellitus or hypertension, forming mutually exclusive groups. The outcome was time to all-cause in-hospital death. Kaplan–Meier methods estimated survival, the log-rank test compared survival curves, and cause-specific Cox regression generated hazard ratios. Age was modelled per 10-year increase and sex was included as a covariate. Aalen–Johansen cumulative-incidence functions and Fine–Gray regression accounted for recorded discharge alive as a competing event. Nine zero-duration admissions were assigned 0.5 days; sensitivity analysis excluded them. Results The cohort comprised 630 admissions, including 242 with diabetes mellitus and 388 with hypertension. There were 83 deaths, with a case fatality rate of 13.2%, which was higher in diabetes mellitus than in hypertension, at 17.8% and 10.3%. Survival declined early in both groups and remained lower among admissions with diabetes mellitus. The log-rank test showed limited separation between the curves ( p = 0.098). After adjustment for age and sex, admissions with hypertension had a lower cause-specific hazard of in-hospital death than those with diabetes mellitus (adjusted hazard ratio 0.618, 95% CI 0.397 to 0.960; p = 0.032). The adjusted Fine–Gray subdistribution hazard ratio was 0.499 (95% CI 0.323 to 0.770; p = 0.002). Excluding the nine zero-duration admissions left 621 admissions and 80 deaths and produced an adjusted hazard ratio of 0.635 (95% CI 0.405 to 0.995; p = 0.047). Conclusions At this tertiary hospital in southwestern Nigeria, patients admitted with diabetes mellitus had poorer in-hospital survival than those admitted with hypertension, and 61.4% of deaths occurred within 30 days of admission.