Page 95 - 16Neonatal Jaundice_compressed

Neonatal jaundice

hyperbilirubinaemia. At the threshold value of 1.8 ppm, it showed 86% sensitivity, 80%
specificity, 40% PPV and 97% NPV. [EL II]

Evidence summary
ETCOc levels were measured at various times and their accuracy evaluated with various
threshold values in two studies with EL II. While one study reported ETCOc to have a PPV of
40% and NPV of 97%, the other study reported 13% PPV and 96% NPV for subsequent
hyperbilirubinaemia. The second study also found no additional benefit from combining this
test with pre-discharge laboratory serum bilirubin levels.

GDG translation from evidence
Although ETCOc shows good negative predictive value, it is not routinely available and does
not accurately predict neonatal hyperbilirubinaemia.
Recommendations

See the end of Section 4.2.
Umbilical cord direct antiglobulin (Coombs’) test (DAT)

Description of included studies
41
38
40
One study with EL II and four EL III studies 38-40;42 from the USA 41;42 , Norway, Taiwan and
39
Turkey examining the predictive ability of the DAT have been included but no meta-analysis
was possible as the studies used different criteria for defining hyperbilirubinaemia.
Review findings
41
In the first study, from the USA, universal DAT was evaluated with reference to ETCOc, and its
accuracy in predicting hyperbilirubinaemia was then assessed. The study population included
660 babies (mean gestational age 38.9 ± 1.4 weeks, mean birthweight 3267 ± 480 g) admitted
consecutively to the postnatal ward of a tertiary hospital. In all cases, cord blood was collected
and DAT was conducted by the gel test. In positive cases, the baby was investigated for
haemolytic disease. The reference standard for haemolysis was ETCOc measured in all babies at
12 ± 6 hours and again at 24 ± 6 hours. Significant haemolysis was defined as ETCOc ≥ 95th
centile. Since maternal cigarette smoking was shown to influence ETCOc, measurement results
were given separately for babies of non-smoking and smoking mothers. Bilirubin measurement
(transcutaneous bilirubin in the majority with subsequent serum bilirubin if required) was
performed in all babies at the time of hospital discharge or earlier if clinically indicated.
Hyperbilirubinaemia was defined as a bilirubin reading ≥ 75th centile on the Bhutani
nomogram. Blinding of outcome assessors was not specified. More than 80% of the study
population was black. The DAT was positive in 3.5% of babies (23 of 659). In babies of non-
smoking mothers, DAT could predict haemolysis (ETCOc levels ≥ 3.2 microlitre/litre) with a
sensitivity of 38.5% and specificity of 98.5%, while in babies of all mothers it showed a
sensitivity of 8.5% and specificity of 97.6% in detecting haemolysis (ETCOc levels
≥ 2.5 microl/litre). The accuracy of DAT in predicting hyperbilirubinaemia was evaluated and
compared with that of high ETCOc levels. A positive DAT showed a sensitivity of 14.7% while
ETCOc showed 27.9% sensitivity in predicting subsequent hyperbilirubinaemia in babies of
non-smoking mothers. The specificity of DAT compared with ETCOc was 98.2% and 97.9%,
respectively. [EL II]
The second study, from the USA, evaluated selective DAT and cord blood bilirubin
42
measurement in predicting hyperbilirubinaemia. The study population included 91 ABO
incompatible babies in a state-sponsored neonatal programme; Rhesus incompatible babies
were excluded. Demographic information on gestational age, birthweight, gender and ethnicity
was not provided. Cord blood was obtained from all babies of group O mothers, and bilirubin
estimations were carried out at 12, 24, 36 and 48 hours of life in cases of ABO incompatibility.
The cord blood bilirubin threshold for a positive test was a measurement > 68 micromol/litre.
Babies with serum bilirubin levels > 273 micromol/litre between 12 and 36 hours were classed
as severely hyperbilirubinaemic. Blinding of outcome assessors was not specified. DAT was

66