LAUSR

Abstract To establish a clinical prediction rule for acute bilirubin encephalopathy (ABE) in term/near-term neonates with extreme hyperbilirubinemia. A retrospective cohort study was conducted between January 2015 and December 2018. Six hundred seventy-three out of 26,369 consecutive neonates with extreme hyperbilirubinemia were enrolled in this study. Data included demographic characteristics, total serum bilirubin (TSB), albumin, bilirubin/albumin ratio (B/A), direct antiglobulin test, glucose-6-phosphate deficiency, asphyxia, sepsis, acidosis. ABE was defined as a bilirubin induced neurological dysfunction score of 4 to 9. We used stepwise logistic regression to select predictors of ABE and devised a prediction score. Of the 673 eligible infants, 10.8% suffered from ABE. Our prediction score consisted of 3 variables: TSB (as a continuous variable; odds ratio [OR] 1.16; 95% confidence interval [CI], 1.02–1.31; logistic coefficient 0.15), B/A (as a continuous variable; OR 1.88; 95% CI, 1.19–2.97; logistic coefficient 0.67), and sepsis (OR 3.78; 95% CI, 1.40–10.21; logistic coefficient 1.19). Multiplying the logistic coefficients by 10 and subtracting 75, resulted in the following equation for the score: Score = 12 × (if sepsis) + 1.5 × (TSB) + 7 × (B/A) − 75. The model performed well with an area under the curve of 0.871. The risk of ABE can be quantified according to TSB, B/A, and sepsis in term/near-term neonates with extreme hyperbilirubinemia.