Predictors of prolonged hospitalization among children admitted with severe pneumonia in Southeast Ethiopia: a retrospective follow-up study

Background Severe pneumonia remains a primary driver of pediatric morbidity in Sub-Saharan Africa. Prolonged hospital stay (PHS) escalates healthcare expenditure and strains referral hospital resources. This study aimed to evaluate the magnitude, clinical determinants, and model performance of factors associated with PHS among hospitalized pediatric severe pneumonia cases in Southeast Ethiopia. Methods A hospital-based retrospective follow-up study was conducted at Asella Referral and Teaching Hospital, analyzing 395 systematically sampled pediatric cases (aged 3 months to 14 years) admitted with severe pneumonia. Multivariable binary logistic regression was executed with model diagnostics, including Variance Inflation Factors (VIF), Hosmer-Lemeshow goodness-of-fit, and Receiver Operating Characteristic (ROC) curve analysis. Results The magnitude of prolonged hospital stay (> 7 days) was 42.0% (95% CI 37.2–46.8). In the multivariable model, independent clinical predictors included presence of comorbidities (Adjusted OR [aOR] = 4.02; 95% CI 2.42–6.67; p < 0.001), care-seeking delay > 48 hours (aOR = 3.89; 95% CI 2.35–6.42; p < 0.001), history of previous pneumonia (aOR = 3.23; 95% CI 1.81–5.74; p < 0.001), partial vaccination status (aOR = 3.03; 95% CI 1.73–5.32; p < 0.001), and underlying malnutrition (aOR = 2.44; 95% CI 1.15–5.18; p = 0.021). The multivariable explanatory model demonstrated good discriminative performance (AUC = 0.804, 95% CI 0.759–0.848) and good model calibration (Hosmer-Lemeshow p = 0.612). Conclusions Over four in ten pediatric severe pneumonia admissions experience prolonged hospitalization. Early triage, prompt care-seeking, complete routine immunization, and nutritional interventions may reduce prolonged hospitalization and should be prioritized in similar referral settings.