Page 87 - 16Neonatal Jaundice_compressed

Neonatal jaundice

values (for each hour of life) connected to form percentile tracks. Any serum bilirubin value
exceeding the 1st percentile track of babies with serum bilirubin > 12 mg/dl was ‘trend 12’,
and serum bilirubin value exceeding the 1st percentile track of babies with serum bilirubin
> 15 mg/dl was ‘trend 15’. Trend 12 and trend 15 were taken as indicative of
hyperbilirubinaemia.
In the second phase, the nomogram was validated in a prospective study carried out at two
hospitals (Hospital A, n = 1244; Hospital B, n = 498). The study population included term
babies who had serum bilirubin measured between 30 and 72 hours because of clinical
jaundice. Most of the babies had a single serum bilirubin measurement, but 514 of 1244 babies
in Hospital A and 175 of 498 babies in Hospital B had two serum bilirubin determinations
12 hours apart. The ability of serum bilirubin measurements exceeding trends 12 and 15 to
predict subsequent hyperbilirubinaemia was evaluated. In Hospital A, 18.5% babies had serum
bilirubin values > 12 mg/dl while 8.0% had serum bilirubin > 15 mg/dl. With a single serum
bilirubin measurement and trend 12 as the threshold, a sensitivity of 99% and a specificity of
49% were obtained, while applying trend 15 gave 100% sensitivity and a specificity of 75%. In
Hospital B, trend 12 gave similar results (98% sensitivity and 36% specificity) to Hospital A but
trend 15 was less accurate, with 88% sensitivity and 78% specificity. Two consecutive serum
bilirubin determinations accurately identified all babies reaching serum bilirubin levels
> 12 mg/dl in the two hospitals (100% sensitivity), and all but one baby reaching serum
bilirubin levels > 15 mg/dl in Hospital B. [EL II]
The third study, conducted in two tertiary hospitals in the USA, compared transcutaneous
36
bilirubin measurement with serum bilirubin for prediction of hyperbilirubinaemia in a
multiracial population. The study population comprised 490 healthy babies with gestational age
≥ 36 weeks and birthweight ≥ 2000 g, or gestational age ≥ 35 weeks and birthweight
≥ 2500 g, and included 59% white, 29.5% black, 3.5% Hispanic and 4.5% Asian babies. At
the time of routine metabolic screening (24–72 hours of age), transcutaneous bilirubin readings
were recorded from the forehead with a BiliChek device and simultaneously two blood samples
were taken for serum bilirubin estimation – one at the local laboratory and the other sent for
high-performance liquid chromatography (HPLC) assay. The laboratory technicians, clinicians
and investigators were all blinded to the transcutaneous bilirubin and serum bilirubin data.
Paired transcutaneous bilirubin and HPLC serum bilirubin values were then plotted on the hour-
34
specific nomogram developed by Bhutani et al. Hyperbilirubinaemia was defined as serum
bilirubin levels ≥ 95th centile on the nomogram (i.e. in the high-risk zone). Altogether, 30 of
490 (6.1%) babies had HPLC serum bilirubin values > 95th centile and only 1.1% had serum
bilirubin levels > 255 micromol/litre. The correlation between transcutaneous bilirubin and
serum bilirubin values was linear and statistically significant (r = 0.91; P < 0.001), and the
values for correlation coefficient were similar when the data were categorised by race. The
mean difference between paired serum bilirubin and transcutaneous bilirubin values was
8 micromol/litre (95% CI −38.9 to 54.9 micromol/litre). For predicting hyperbilirubinaemia,
pre-discharge transcutaneous bilirubin > 75th centile showed a sensitivity of 100%, a
specificity of 88% and a likelihood ratio of 8.4. None of the babies with serum bilirubin levels
in the high-risk zone had a transcutaneous bilirubin recording below the 75th centile on the
nomogram, while all babies with serum bilirubin levels below the 40th centile also had
transcutaneous bilirubin values below the 40th centile. No adverse events were reported using
the BiliChek device. [EL II]
A nested case–control study was carried out at 11 hospitals in a health maintenance
9
organisation in the USA to investigate predictors of hyperbilirubinaemia and evaluate the
predictive accuracy of a risk index model. This study has been described in Chapter 3 on risk
factors. Information on risk factors was collected by reviewing hospital records and interviewing
parents. Using bivariate analysis, several clinical and demographic variables were found to be
associated with an increased risk of hyperbilirubinaemia. They included maternal factors (race,
age, family history of jaundice in a newborn, vacuum delivery) and neonatal factors (male sex,
lower gestational age, early jaundice, cephalohaematoma, bruising, breastfeeding at time of
discharge). These variables then underwent multiple regression analysis to identify independent
predictors of hyperbilirubinaemia. This was done by including and later excluding cases of early
jaundice (n = 14) in order to predict hyperbilirubinaemia after initial hospital discharge. When

58