Validation of a Risk Prediction Nomogram for Virologic Failure Among Patients on First-Line ART After ONE-J Program Implementation in Northwest Ethiopia

IntroductionVirological failure is defined as a plasma viral load >1000 copies/mL on 2 consecutive tests after 3 months of adherence support and 6 months of ART. In Ethiopia, limited studies have developed risk prediction models, though such models are vital for guiding patient-specific interventions and improving HIV treatment outcomes.ObjectiveThis study aims to determine the incidence and prognostic factors of virologic failure among HIV patients on first-line HAART following the one-J program at Felege Hiwot Hospital, Northwest Ethiopia, in 2025.MethodsA retrospective cohort study was conducted using patient records from Felegehiwot Comprehensive Specialized Hospital. Data were analyzed with STATA 17 and R 4.5.0. Descriptive statistics and multivariable logistic regression (via LASSO-selected predictors) were applied to develop a simplified nomogram. Model performance was evaluated using discrimination, calibration, and decision curve analysis to assess clinical utility.ResultThe incidence of virologic failure was 20.2% (95% CI: 17.9-22.6). Key predictors included TB-HIV co-infection, INH prophylaxis, CPT, adverse drug reactions, disclosure, alcohol use, smoking, and CD4 level. The model showed good performance (AUC = 0.817; reduced AUC = 0.810), strong calibration (Brier = 0.117), and superior clinical benefit in decision curve analysis.ConclusionA nomogram integrating 8 predictors TB-HIV co-infection, INH prophylaxis, CPT, adverse drug reaction history, disclosure status, alcohol use, smoking status, and CD4 count showed excellent discrimination (AUC = 0.817) and good calibration, indicating strong potential for individualized risk prediction in HIV patient management.