Page 102 - 16Neonatal Jaundice_compressed

Recognition

jaundice reached zone 3 on the Kramer scale, transcutaneous bilirubin measurements were
made from the sternum with the Minolta JM-102 and from the forehead and sternum with the
BiliChek. Simultaneously, blood was collected for serum bilirubin estimation and analysed
within 30 minutes. Apart from analysing the linear correlation between the three tests and
serum bilirubin levels, their diagnostic accuracy was evaluated by measuring the area under the
ROC curve for serum bilirubin > 250 micromol/litre in term babies and serum bilirubin
> 190 micromol/litre in preterm babies.
In term babies, transcutaneous bilirubin recordings using the Minolta JM-102 showed the best
results in terms of linear correlation and diagnostic accuracy (R² = 0.82; P < 0.01 and
AROC = 0.98). Clinical assessment showed variable results for the correlation coefficient among
the white and non-white babies (R² = 0.74 by nurse and 0.70 by investigator for white babies,
R² = 0.71 by nurse and 0.65 by investigator for non-white babies). The AROC for the Kramer
method was 0.88. It was also seen that a grading of jaundice below 2 on the Kramer scale
(determined by the nurses) had 100% NPV in ruling out serum bilirubin levels
> 250 micromol/litre. The second study done on healthy preterm babies showed similar
results: Minolta JM-102 showed the best performance with a AROC of 0.96 and squared
correlation coefficient R² = 0.76 (P < 0.001). The BiliChek performed worse than the Minolta
JM-102 but better than clinical assessment, with AROC of 0.88 and 0.89 at forehead and
sternum, respectively. Values for squared correlation coefficients and AROC for the Kramer
method were poor (0.22 and 0.73 for nurses’ observations, respectively, and 0.20 and 0.70 for
the principal investigator observations, respectively).

Evidence summary
Evidence from EL I and EL II studies shows that clinical estimation of the degree of jaundice by
experienced healthcare professionals and nursery staff is moderately correlated with actual
serum bilirubin levels. The value of the correlation coefficient was much less for the preterm
babies and babies with dark skin tones compared with babies with light skin tones and term
babies. In one study, parental assessment of cephalo-caudal progression showed better
correlation than assessment by nurses and paediatricians. Variable results were seen regarding
the diagnostic accuracy of clinical assessment in detecting severity of jaundice. In one study,
visible jaundice ‘caudal to nipple line’ had a sensitivity of 97% and a specificity of 19% in
detecting serum bilirubin levels > 205 micromol/litre, while the other study reported 76%
sensitivity with 60% specificity. Results from the EL 1 study show that visual assessment led to
more than 60% of babies being misclassified into the lower risk zones on the nomogram when
their serum bilirubin values were actually in the high-risk zones. Moreover, this study found
clinical assessment to have poor diagnostic accuracy in detecting jaundice in high-risk zones
when the observations were made before 36 hours of age and in babies born before 37 weeks
of gestational age.

Nevertheless, results from three studies show that if clinical examination carried out on the
second or third day indicates absence of jaundice, it has high NPV for ruling out the presence of
hyperbilirubinaemia (Table 5.1).
GDG translation of evidence

The experience of the GDG is that it is important to examine the naked baby in good light,
preferably natural light. Review of the evidence shows that, in most term babies, healthcare
professionals and parents are capable of recognising jaundice but not very good at assessing its
severity clinically.
GDG experience is that jaundice is more difficult to recognise in babies with dark skin tones.
The GDG recognised that international kernicterus registries and population studies of
hyperbilirubinaemia report over-representation of babies from ethnic groups with dark skin
tones. This difficulty may be ameliorated by a careful assessment of all infants including
examination of sclerae, gums and blanched skin.

Parents can recognise the head-to-toe progression of jaundice. In one study, parents’ recognition
of visible jaundice was better that that of clinical staff.

73