Survival and predictors of mortality among adults with type 2 diabetes at a teaching hospital in Ghana: a retrospective cohort study

Abstract Background Mortality rates remain high among individuals with type 2 diabetes. The rising cases of Diabetes Mellitus (DM) have negatively impacted society’s development, health and have impoverished individuals. T2DM among other types has the highest prevalence among all populations. The rising prevalence of diabetes in Ghana has become a serious public health concern. It demands extensive research to identify factors that need public health action to improve survival and to reduce suffering and complications and to inform policy decisions towards finding effective treatment and management of the disease. This study aimed to estimate the median survival time and determine the key predictors of mortality among diabetes patients in the Tamale Teaching Hospital. Methods This study employed a retrospective cohort study design. A total of 458 participants were included, of which 18.38% had died, after extracting the medical records of all diabetes patients admitted between 2021 and 2024 from the Lightwave Health Information Management System (LHIMS). Descriptive statistics were carried out to summarise socio-demographic and clinical variables. Kaplan-Meier curve was used to examine the cumulative, median and treatment regimen survival probabilities among patients. The Schoenfeld residual test was performed to establish Cox proportional assumption test and variables that satisfy the assumptions were selected with p-values ≤ 0.25 from bivariate Cox regression analysis into Multivariable Cox model. Multivariable Cox model was fitted to examine the adjusted association between significant variables and mortality. Results The overall survival was 93.05%, 81.5% and 73.7% for first, second and third years respectively. The median survival time was 1.52 years [(IQR = 0.87–2.09). The variables included in the adjusted Cox proportional hazard analysis were; education, age group, occupation, chronic kidney disease, stroke, treatment regimen, health insurance and BMI. In the adjusted Cox PH model, patients with basic education had higher risk of death (HR = 6.66, 95% CI: 2.55–17.37) from T2DM compared to those with no education. Self-employed participants had lower risk of death (HR = 0.41, 95% CI: 0.22–0.75). Oral and Insulin combination increased the risk of death (HR = 16.63, 95% CI = 2.28–121.35). Stroke elevated the risk of death (HR = 2.75, 95% CI: 1.41–5.36). Conclusion T2DM patients had a decreasing survival year-on-year, with a very short median survival time. The factors that predict mortality among T2DM patients are education, employment, treatment regimen and stroke. Hospitalised diabetics especially those with stroke should be given intensive care including oral anti-diabetic medications to ensure better survival outcomes.