Background: In recent years, many non-invasive instruments have been used to examine the correlation between transcutaneous bilirubin (TcB) measurements and serum concentrations of bilirubin (SB) in newborns with a view to reducing the need for blood samples. However, their exact role has not been univocal. The aim of the present study was to determine whether non-invasive measurement of skin colour combined with appropriate clinico-statistical methodology can be used to better quantify SB in the first days of life.
Methods: The study group consisted of 49 healthy term breast-fed newborns. Skin colour was measured with a Minolta CR-300 colorimeter and measurements were made in the first 12 h of life, at 48 h and in 13 babies at 96 h. To determine changes in SB from variations in skin colour between birth and 48 h (and also 96 h in 13 cases), a multivariate linear regression model was designed. To test the generalisation power of the regression model, we examined its predictive power for SB a third time (96 h; 13/49 cases) with respect to the previous two times.
Results: The model made it possible to recognise increases in SB accurately with a mean absolute error 0.99 mg/dL with a good generalisation power.
Conclusions: The methodological model proposed here made it possible to accurately recognise increases in SB with respect to an initial value. Provided the methodological protocol is observed, TcB can therefore be used in the screening and follow-up of neonatal jaundice.