Covariate-stratified analysis of prostate cancer and competing risk mortality in Ghana: A bootstrapped survival analysis approach

Prostate cancer remains a major health concern in sub-Saharan Africa, where late-stage presentation and comorbidities complicate survival estimation and treatment planning. This study evaluated 300 men aged 25–84 years diagnosed with prostate cancer drawn from a major tertiary care center in the region (2015–2020), to determine how competing health risks influence cancer-specific mortality. Using cumulative incidence functions (CIFs) derived from the complement of Kaplan–Meier (1-KM), Kaplan–Meier-based CIF (KM-CIF), and Aalen–Johansen estimators, we compared standard survival models with competing risk models. Bootstrapping (1000 replications) was applied to refine 95 % confidence intervals, and interpolation techniques identified a critical age range (59.9–69.3 years) where competing risks most strongly influenced mortality estimates. Results showed that standard Kaplan–Meier methods consistently overestimated prostate cancer-specific mortality, particularly among older patients, those with stage 3 disease, larger tumors (>6 cm), elevated Gleason scores (≥7), and high prostate-specific antigen levels. In contrast, Fine–Gray subdistribution hazard models provided more accurate and clinically meaningful estimates by accounting for deaths from other causes. Treatment-specific analyses revealed that Cox proportional hazards models overestimated the effects of chemotherapy and hormonal therapy relative to Fine–Gray models, underscoring the limitations of conventional approaches in settings with high competing mortality. These findings highlight the need to integrate competing risk frameworks into survival analyses to avoid biased risk predictions and inappropriate treatment decisions. By identifying an age-specific window where competing events substantially distort conventional survival estimates, this study provides actionable guidance for tailored screening and treatment strategies. The adoption of robust competing risk methodologies, especially in resource-limited settings, can improve prognostic accuracy, guide patient counseling, and enhance the allocation of scarce healthcare resources.