Page 99 - 16Neonatal Jaundice_compressed

5 Recognition

Introduction
This chapter addresses the problem of recognition of jaundice and discusses visual assessment
and the measurement of jaundice. Although bilirubin causes yellow discolouration of the skin,
the whites of the eyes and the palate, detection of this discolouration can be surprisingly
difficult. Even babies with very pale skin can appear ‘suntanned’ rather than yellow, and
detection of jaundice in babies with dark skin tones can be almost impossible. In jaundice
caused by liver disease, the total bilirubin level is variable. Sometimes a baby may not be
obviously jaundiced yet have a serious, potentially lethal disease. In babies with liver disease,
the degree of jaundice does not correlate with the severity of the liver disease. Traditional
teaching on examination for jaundice has recommended ‘blanching’ a small area of skin (often
on the nose) by pressing it, and inspecting at the whites of the eyes and palate. Jaundice is
thought to spread from the head to the toes in a ‘cephalo-caudal’ progression. The ‘zones of
52
Kramer’ attempt to quantify this progression. This review of the evidence is a crucial part of the
guideline, because if babies are not recognised to be jaundiced in the first place they cannot
enter the care pathway.

Clinical question
What is the accuracy of various tests (clinical history and examination, urine/stool
examination, icterometer and transcutaneous bilirubin levels ) in recognising neonatal
jaundice or hyperbilirubinaemia?

For answering the question on diagnostic accuracy of various tests in the recognition of jaundice
or detection of its severity, these studies were reviewed against the following predefined criteria:
● prospective studies
● diagnostic accuracy of the test or its correlation evaluated against the reference standard
(serum bilirubin levels)
● test and the reference test performed within 1 hour of each other.
As in Chapter 4, the primary screening of 2840 titles and abstracts from the database led to the
retrieval of 148 papers.

A total of 30 studies have been included in this review. Except for four studies with quality EL I
(one on visual inspection and three on transcutaneous bilirubin measurement with BiliChek)
and six studies with EL III, the rest of the studies are of EL II, with the main reason for
downgrading their quality being the absence or non-reporting of blinding among the
test/reference test operators. Only one study was identified on the diagnostic accuracy of urine
or stool examination and limited evidence was available for the icterometer. As few diagnostic
accuracy studies had been carried out in preterm and dark-skinned babies, the selection criteria
were relaxed in studies related to these populations. Diagnostic accuracy of three devices used
for transcutaneous bilirubin measurements (Minolta JM-102, Minolta JM-103 and BiliChek) has
been reviewed.

Most of the studies have reported the correlation coefficient (r) of the test results with the serum
bilirubin values. This statistical measure indicates a degree of association between the two tests,
but it is largely dependent on the distribution of serum bilirubin values in the sample population
and does not adjust for various biases. Efforts were made to convert the unit of bilirubin
measurement from mg/dl to micromol/litre (1 mg/dl = 17.1 micromol/litre) and present the
diagnostic accuracy results in terms of sensitivity and specificity where the data were sufficient.
Meta-analysis was performed to calculate the diagnostic accuracy of the Minolta JM-102 and JM-
103 using the statistical programme Meta-DiSc (www.hrc.es/investigacion/metadisc_en.htm). As

70