Epidemiological characteristics and clinical predictors of neonatal jaundice in a multi- ethnic population in Xinjiang, China: a retrospective study

Background: Neonatal jaundice is one of the most common conditions leading to neonatal hospitalization and is a preventable cause of bilirubin-induced neurological dysfunction. Objectives: This study aimed to characterize the clinical profile of neonates with jaundice admitted to the People's Hospital of Atushi City, Xinjiang and to explore the clinical predictors of severe hyperbilirubinemia. Methods: We conducted a retrospective observational study of all neonates (≤ 28 days) admitted to the neonatal unit of the People's Hospital of Atushi City with jaundice. Demographic, perinatal, and clinical variables, including ethnicity, gestational age, birth weight, age at admission, feeding type, perinatal risk factors, total serum bilirubin (TSB) at admission and discharge, phototherapy and antibiotic use, and length of stay, were extracted from electronic medical records. Severe jaundice was defined as an admission TSB at or above the 75th percentile of the cohort distribution (348.7 μmol/L). Early-onset jaundice was defined as an age at admission of ≤ 48 h. Multivariable logistic regression was applied to identify independent predictors of severe jaundice. Results: Among 261 infants (57.1% male; 90.4% born at term), the mean gestational age was 39.2 ± 1.2 weeks, and the mean birth weight was 3.35 ± 0.52 kg. Uyghur infants comprised 82.8% of the cohort, and Han infants comprised 9.6%. The mean admission TSB was 298.1 ± 79.1 μmol/L; 22.6% of infants met the definition of severe jaundice. The proportion of severe jaundice was 27.9% among Uyghur infants and 20.0% among Han infants. In adjusted analysis, early-onset jaundice was associated with lower odds of severe hyperbilirubinemia (odds ratio [OR] 0.12, 95% confidence interval [CI] 0.01-0.98), whereas pre-term birth showed a non-significant trend toward increased risk (OR 4.03, 95% CI 0.85-19.23). Ethnicity was not an independent predictor. These findings should be interpreted with caution and validated in larger multicenter cohorts.