Page 93 - 16Neonatal Jaundice_compressed

Neonatal jaundice

predict serum bilirubin levels ≥ 200 micromol/litre was calculated. Blinding of the outcome
assessors was not specified in three studies.

Review findings
The prevalence of hyperbilirubinaemia (serum bilirubin ≥ 290 micromol/litre) varied between
2.9% and 9.5% in the three studies, while in the Danish study 20.3% of the babies had serum
bilirubin levels ≥ 200 micromol/litre. The sensitivity of cord blood bilirubin to predict serum
bilirubin levels ≥ 290–300 micromol/litre ranged from 22% to 100%, while the specificity
ranged from 41% to 95%. The pooled sensitivity was 79% (95% CI 68% to 87%) and the
pooled specificity 60% (95% CI 58% to 62%), but there was strong evidence of statistical
heterogeneity for both the pooled results with I² at 99.3% and 90.5%, respectively(Figures 4.3
and 4.4). The Danish study showed that cord blood bilirubin levels with threshold value
29
≥ 35 micromol/litre had a sensitivity of 71% and specificity of 68% in predicting serum
bilirubin ≥ 200 micromol/litre. [EL II]

Figure 4.3 Pooled specificity of umbilical cord blood bilirubin in predicting later hyperbili-
rubinaemia

Figure 4.4 Pooled sensitivity of umbilical cord blood bilirubin in predicting later hyperbili-
rubinaemia

Evidence summary

Results from three EL II studies indicate great variation in the ability of cord blood bilirubin to
predict hyperbilirubinaemia in healthy term and preterm babies. Sensitivity ranged from 22% to
100% and specificity from 41% to 95%. The pooled sensitivity and specificity were 79% and
60%, respectively, but the results were marred by strong evidence of statistical heterogeneity.
The remaining study had a sensitivity of 71% and specificity of 68% in predicting serum
bilirubin ≥ 200 micromol/litre.

64