Routine transcutaneous bilirubin measurements combined with clinical risk factors improve the prediction of subsequent hyperbilirubinemia.

OBJECTIVE: To evaluate predischarge transcutaneous bilirubin (TcB) measurements combined with risk factors as predictors of the risk of a subsequent total serum bilirubin (TSB) >or=17 mg per 100 ml (291 micromol l(-1)).

STUDY DESIGN: Routine TcB measurements are obtained daily for all infants in our well baby nursery. We performed a nested case-control study comparing all 75 infants who had been readmitted with TSB >or=17 mg per 100 ml (291 micromol l(-1)) between 1 February 2005 and 28 February 2007 with randomly selected controls that had not been readmitted.

RESULT: Between 1 February 2005 and 28 February 2007, 11 456 infants were discharged from the well baby nursery. Seventy-five infants (0.65%) were readmitted at a mean age of 110+/-29.9 h with a TSB>or=17 mg per 100 ml (291 micromol l(-1)). All received phototherapy. Using logistic regression analysis, three variables were statistically significant for predicting cases: the maximum predischarge TcB percentile group (P<0.0001, adjusted odds ratio (AOR), >95th percentile 148; 95% confidence interval (CI) 21 to >999, AOR 76 to 95th percentile 15; 95% CI 3.1 to 70, AOR 50 to 75th percentile 6.1; 95% CI 1.3 to 28 compared with <50th percentile), exclusive breastfeeding (P<0.0001, AOR 11; 95% CI 3.7 to 34) and gestational age (P=0.0057, AOR 35 to 36 6/7 week 21; 95% CI 2.3 to 185, AOR 37 to 37 6/7 week 15; 95% CI 1.9 to 115, AOR 38 to 38 6/7 week 1.8; 95% CI 0.3 to 11, AOR 39 to 39 6/7 week 1.1; 95% CI 0.2 to 7 AOR >or=41 week 0.88; 95% CI 0.1 to 10 compared with 40 to 40 6/7 week infants). These three variables provided the best prediction of a case (c=0.885, area under the receiver operating characteristic curve) and this prediction was significantly better than the use of the clinical risk factors, gestation and exclusive breastfeeding, alone (c=0.770, P<0.001) or the TcB percentile grouping alone (c=0.766, P<0.001). Substituting the TcB rate of rise (c=0.903, P=0.316) or the last measured TcB (c=0.873, P=0.292) for the maximum TcB measurement did not significantly improve the predictors of a case.

CONCLUSION: Combining predischarge TcB levels with two clinical risk factors-gestational age and exclusive breastfeeding-significantly improves the prediction of subsequent hyperbilirubinemia.