Despite substantial advancements in antiretroviral therapy (ART), HIV-related mortality remains a pressing challenge, particularly in resource-limited settings. In countries like Ethiopia, there is a critical gap in tools to predict mortality risk among patients receiving ART, which limits timely clinical decision-making and intervention. Addressing this gap, the present study aimed to develop a validated mortality risk prediction model for adult HIV patients receiving ART at Felege Hiwot Comprehensive Specialized Hospital (FHCSH). A retrospective cohort study was conducted among 777 HIV-positive adults receiving ART at FHCSH. Participants were selected using the “rule of thumb” for predictive modeling. Secondary clinical data were extracted, cleaned, and analyzed using Epi Info and R software. Logistic regression was employed to identify the significant prognostic determinants of mortality. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC) and calibration plots. A nomogram was developed to visualize the risk estimates. Internal validation was performed using bootstrapping method, and decision curve analysis (DCA) was conducted to assess clinical utility. The overall incidence of mortality among ART patients was 12.6% (95% CI 10.2–15.1%). Six key prognostic factors—WHO clinical stage, functional status, baseline ART regimen, duration of treatment, adherence level, and presence of comorbidities—were incorporated into the final prediction model. The nomogram model demonstrated excellent discriminative ability, with an AUC of 0.968 (95% CI 0.951–0.983) for the original model and 0.967 (95% CI 0.947–0.987) for the reduced model. At the optimal cut-off point (0.1667), the model achieved 93.3% accuracy, 87.8% sensitivity, and 94.1% specificity. DCA indicated strong net clinical benefit across a wide range of threshold probabilities (0.1–0.8). This study developed a robust and well-performing mortality risk prediction model for HIV patients on ART, incorporating six easily obtainable clinical indicators. The model, presented as a nomogram, showed excellent discrimination and calibration, providing a practical tool for frontline clinicians to stratify mortality risk and tailor interventions. However, external validation through large-scale, multicenter prospective studies is recommended before routine clinical application.
Tesfa et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: