Evaluation of Treatment Thresholds for Unconjugated Hyperbilirubinemia in Preterm Infants: Effects on Serum Bilirubin and on Hearing Loss?

Background: Severe unconjugated hyperbilirubinemia may cause deafness. In the Netherlands, 25% lower total serum bilirubin (TSB) treatment thresholds were recently implemented for preterm infants. Objective: To determine the rate of hearing loss in jaundiced preterms treated at high or at low TSB thresholds. Design/Methods: In this retrospective study conducted at two neonatal intensive care units in the Netherlands, we included preterms (gestational age <32 weeks) treated for unconjugated hyperbilirubinemia at high or low TSB thresholds. Infants with major congenital malformations, syndromes, chromosomal abnormalities or toxoplasmosis, rubella, cytomegalovirus, herpes, syphilis, and human immunodeficiency infections were excluded. We analyzed clinical characteristics and TSB levels during the first ten postnatal days. After two failed automated Auditory Brainstem Response (ABR) tests we used the results of the diagnostic ABR examination to define normal, unilateral, and bilateral hearing loss (>35 dB). Results: There were 479 patients in the high and 144 in the low threshold group. Both groups had similar gestational ages (29.5 weeks) and birth weights (1300 g). Mean and mean peak TSB levels were significantly lower after the implementation of the novel thresholds: 152 +/- 43 mu mol/L and 212 +/- 52 mu mol/L versus 131 +/- 37 mu mol/L and 188 +/- 46 mu mol/L for the high versus low thresholds, respectively (P<0.001). The incidence of hearing loss was 2.7% (13/479) in the high and 0.7% (1/144) in the low TSB threshold group (NNT = 50, 95% CI, 25-3302). Conclusions: Implementation of lower treatment thresholds resulted in reduced mean and peak TSB levels. The incidence of hearing impairment in preterms with a gestational age <32 weeks treated at low TSB thresholds was substantially lower compared to preterms treated at high TSB thresholds. Further research with larger sample sizes and power is needed to determine if this effect is statistically significant.