Time to death and its predictors among under-five children with acute pneumonia: a Bayesian parametric survival analysis

Abstract

Pneumonia is one of the most common and deadly infectious diseases affecting under-five children, responsible for about 15% of all deaths in this age group worldwide. In Ethiopia, the prevalence ranges from 16% to 21%, contributing substantially to under-five mortality. Despite national child survival efforts, pneumonia-related deaths remain a major public health concern. Understanding the burden and identifying key risk factors are essential for effective prevention and timely intervention. This study aimed to estimate the time to death and identify its predictors among under-five children with acute pneumonia using Bayesian parametric survival analysis. A retrospective study was conducted with 451 under-five children diagnosed with acute pneumonia. Three survival analysis models were applied: the Cox proportional hazards model, the parametric accelerated failure time (AFT) model, and the Bayesian parametric survival model. In the Bayesian model, Markov Chain Monte Carlo (MCMC) methods such as Gibbs sampling and the Metropolis-Hastings algorithm were employed to obtain samples from the posterior distributions of the parameters. Each model was evaluated using appropriate model selection criteria to identify the best-fitting approach. The Bayesian Lognormal AFT model identified several significant predictors of time to death among under-five children with acute pneumonia. All model parameters showed good convergence, with Monte Carlo errors under 5% of their standard deviations. Female children had shorter survival times compared to males (AF = 0.46; 95% CI: 0.36–0.97). Children aged 1–11 months had better survival outcomes (AF = 0.10; 95% CI: 0.05–0.21) than those aged 48–59 months. Rural residence (AF = 1.48; 95% CI: 1.03–2.09), diagnosis during spring (AF = 0.73; 95% CI: 0.52–0.92) and summer (AF = 0.66; 95% CI: 0.49–0.84), comorbidities (AF = 1.26; 95% CI: 1.03–1.65), severe acute malnutrition (AF = 0.26; 95% CI: 0.13–0.43), anemia (AF = 0.88; 95% CI: 0.73–0.93), low weight (AF = 0.72; 95% CI: 0.55–0.90), and home delivery (AF = 0.75; 95% CI: 0.59–0.95) were all associated with reduced survival times. This study identified key predictors of mortality among under-five children with acute pneumonia using a Bayesian parametric survival model. Female, rural residence, severe acute malnutrition, comorbidity, anemia, and low weight were significantly associated with reduced survival times. Seasonal variation and place of delivery also influenced mortality, highlighting the impact of environmental and health system factors. These findings emphasize the need for targeted interventions focusing on early diagnosis, nutritional support, and tailored care for high-risk groups. Furthermore, the Federal Ministry of Health could enhance community awareness of early pneumonia detection and effective home management, particularly in rural areas where mortality risk is higher.