Page 86 - 16Neonatal Jaundice_compressed

Early prediction of serious hyperbilirubinaemia

Recommendations
See the end of Section 4.1.

Pre-discharge risk assessment
Description of included studies

Seven studies have been included in this section, six from the USA 9;12;14;34-36 and one from
37
Italy. Four cohort studies 14;34;36;37 were conducted prospectively and two retrospectively, 12;35
36
9
while one study was a nested case–control study. Apart from one study with EL I, the studies
are of EL II. Two main strategies were employed in these studies to predict subsequent
hyperbilirubinaemia: pre-discharge bilirubin or early bilirubin measurement combined with
clinical risk factors. Two studies 34;37 evaluated the predictive accuracy of pre-discharge serum
bilirubin plotted on an hour-specific nomogram, while one study assessed pre-discharge
36
transcutaneous bilirubin measurements using BiliChek. Clinical risk factors were evaluated in
four studies, 9;12;14;35 either alone or in combination with pre-discharge bilirubin measurement. In
9
one nested case–control study, a risk index model was assessed; in two retrospective cohort
studies 12;35 the risk index was compared with pre-discharge serum bilirubin plotted in risk zones,
and in one prospective study the predictive value of multiple risk factors was first compared
14
with pre-discharge bilirubin (transcutaneous bilirubin or serum bilirubin) levels, and later their
combined accuracy was assessed.

Review findings
34
The first study, conducted in the USA, evaluated the predictive ability of an hour-specific pre-
discharge serum bilirubin measurement. The study population included 13 003 term and near-
term appropriate for gestational age babies admitted to the well-baby nursery of a tertiary
hospital over a 5 year period. Pre-discharge (18–72 hours) serum bilirubin was measured as part
of routine metabolic screening. Babies admitted to the intensive care unit, those with a positive
DAT, and those who started phototherapy before serum bilirubin measurement were excluded.
After discharge, the babies were followed up by home care nurses, who could request
laboratory serum bilirubin if they had clinical concerns. Based on the pre- and post-discharge
serum bilirubin measurements in 2840 eligible babies (recorded in epochs of 4 hours for the
first 48 hours of age, 12 hours for 48–96 hours of age, and 24 hours for age 5–7 days), an hour-
specific serum bilirubin nomogram was constructed. This was divided into zones: high risk
(≥ 95th centile), high intermediate risk (between the 75th and 95th centile), low intermediate
risk (between the 75th and 40th centile) and low risk (below the 40th centile). The nomogram
was used as the reference standard to determine the ability of pre-discharge serum bilirubin
(measured between 18 and 72 hours of age) to predict subsequent severe hyperbilirubinaemia,
which was defined as serum bilirubin level in the high-risk zone (≥ 95th centile). For 8.1%
(230 of 2840 babies), serum bilirubin fell within this zone at some time. In 58 babies (2.0%),
this occurred after discharge. Among 172 of 2840 babies with pre-discharge serum bilirubin
≥ 95th centile, 68 had subsequent hyperbilirubinaemia, giving pre-discharge serum bilirubin
≥ 95th centile a sensitivity of 54.0% and a specificity of 96.2% in predicting
hyperbilirubinaemia. Pre-discharge serum bilirubin ≥ 75th centile showed a sensitivity of
90.5% and a specificity of 84.7%. None of the 126 babies with pre-discharge serum bilirubin
< 40th centile developed subsequent hyperbilirubinaemia. The predictive accuracy of each risk
zone was also calculated in terms of the likelihood ratio (LR) for predicting serum bilirubin
≥ 95th centile. The LR was 14.1 for the high-risk zone (and 54% of babies continued in the
same zone), 3.2 for the high intermediate risk zone (12.9% moved up to the high-risk zone), 0.5
for the low intermediate risk zone (2.2% moved up to the high-risk zone), and 0 for the low-risk
zone (none moved into the high-risk zone). [EL II]
The second study, from Italy, was conducted in two phases. In the first phase, serum bilirubin
37
curves were developed from blood samples obtained at 6 hours of age and then every 4–
6 hours during the day and every 6–12 hours during the night. 438 full term appropriate for
gestational age babies without ‘asphyxia’ and without Rhesus or ABO incompatibility were
included. Serum bilirubin curves for babies with levels > 12 mg/dl (205 micromol/litre) and
those with serum bilirubin > 15 mg/dl (255 micromol/litre) were devised, and their percentile

57