Modelling recurrent complications among children with advanced HIV disease using conditional survival analysis models

Abstract

Infants and young children infected with HIV face higher morbidity & mortality rates. Children under five with HIV are considered to have advanced HIV disease. Common causes of death in these patients include TB, severe bacterial infections, diarrheal diseases, and severe acute malnutrition. Studying recurrent complications, time-to-event patterns, and risk factors among children would enhance interventions among health professionals that aim at better management of AHD and thus an improved quality of life. Conditional survival models analyze and predict the time until multiple events of interest occur, accounting for interdependencies among events, aiding in improved patient outcomes. This study aimed to identify the common recurrent complications in children with AHD and analyse the time-to-event patterns of their reoccurrence. It also assessed the effectiveness of different conditional modelling approaches in accounting for the interdependencies between sequential analyze complications. Data from Baylor College of Medicine Children’s Foundation, Uganda was used, involving HIV-positive children aged 0-18, those with AHD were enrolled and followed up in a retrospective longitudinal cohort study design (3 years from the first complication). The sample included 1000 patients and 100% power was calculated. The analysis employed descriptive statistics, the log-rank test, Kaplan Meier, cox proportional hazard model, and the conditional survival analysis models (AG, PWP, WLW) with Stata programming software used for computations, the lowest AIC was used for model selection and model diagnostics was done for model evaluation. Key findings revealed a high incidence of recurrent complications with notable variations based on demographic and clinical characteristics. The analysis demonstrated that certain factors, including adherence to ART (HR=0.12; 95% CI: 0.015-0.934), significantly influenced the recurrence rates. The study concluded that malnutrition, TB, and pneumonia are the most common complications in children with AHD. Factors such as age, treatment regimen, adherence level, and clinical stage of HIV predict the intensity of recurrence. The WLW model was found to be the best fit for analyzing the recurrence of these complications.