RESUMO
AIM: 1) To compare the clinical assessment of craniocaudal progression of jaundice and two transcutaneous bilirubinometers with serum bilirubin values in preterm neonates; 2) to identify factors affecting the difference between non-invasive bilirubin estimation and serum bilirubin. METHODS: Serum bilirubin was clinically estimated in healthy preterm newborn infants (34 to 36.9 gestational weeks) independently by a primary investigator and by nurses, and subsequently compared with separate measures of two transcutaneous bilirubinometers. RESULTS: A total of 107 measurements were performed on 69 infants. Minolta JM-102 showed the best performance, with ROC area under the curve of 0.96, followed by BiliCheck over the sternum (0.89) and over the forehead (0.88), clinical assessment by nurses (0.73) and by a physician (0.70). Serum bilirubin >190 micromol/l can be detected with 95% sensitivity with Minolta JM-102 > or =19 units, with BiliCheck > or =145 micromol/l over the sternum and > or =165 micromol/l over the forehead and with jaundice progression to the trunk or further (Kramer zone > or =2). Gestational age affects all non-invasive methods in the estimation of serum bilirubin, whereas skin colour affects both BiliCheck and clinical assessment. Ambient light affects only clinical assessment. CONCLUSION: Minolta JM-102 showed the best performance, closely followed by BiliCheck, with clinical assessment performing far worse than either transcutaneous method. None of the three methods are recommended as complete substitutes for serum bilirubin values in jaundiced preterm infants.