Page 264 - 16Neonatal Jaundice_compressed

Appendix H: Evidence tables

Bibliographic details Study type & Patient characteristics Test, Reference Standard, Results Reviewers Comments
Evidence level Threshold for a positive test
phototherapy only, additional random cases, and predictive accuracy Breastfeed only at discharge: OR 6.9 (2.7–
sample of 30 babies with maximum compared by the c-statistic (equal 17.5)
TSB levels of 342 to to area under ROC curve) Asian race: OR 3.1 (1.5–6.3)
426 micromol/litre added to the Bruising: OR 3.5 (1.7–7.4)
control group Reference standard: Cephalohaematoma: OR 3.2 (1.1–9.2)
Significant hyperbilirubinaemia Maternal age ≥ 25 years: OR 2.6 (1.1–9.2)
Exclusion criteria: defined as maximum TSB
Not defined levels = 428 micromol/litre within Cases excluding early jaundice (n = 59)
the first 30 days after birth.
GA (per wk): OR 0.6 (0.4–0.7)
Breastfeed only at discharge: 5.7 (2.1–15.5)
Asian race: OR 3.5 (1.7–7.4)
Bruising: OR 4.0 (1.8–8.8)
Cephalohaematoma: OR 3.3 (1.1–10)
Maternal age ≥ 25 years: OR 3.1 (1.2–8.1)

Risk Index scoring

6 points each for exclusive breastfeeding
and family HISTORY OF jaundice in a
newborn,
4 points each for bruising and Asian race,
3 points each for cephalhematoma and
maternal age ≥ 25 years,
1 point for male sex, -2 points for black
race, and 2(40-GA)

Accuracy of Risk Index score in predicting
significant hyperbilirubinaemia

Overall c-statistic 0.85

Risk index score < 10
+LR: 0.2

Risk index score > 10
+LR: 2.2

Risk index score > 20
+LR: 18.2

Newman TB; Study Type: 1) Study 1: Study 1: Study 1: Retrospective cohort study
Nested case- control Cohort of all infants with Risk index score developed by Comparison of 1995–96 cohort (n = 51 387) Unselected population but
Year: 2005 study BW = 2000 g and GA = 36 weeks exclusion criteria not defined
2) Retrospective born alive at 11 hospitals of a health assigning points equal to the OR with 1997–98 cohort (n = 53 997) Confounding variables
Country: USA cohort maintenance organisation during a for risk factors significant in the No difference regarding % of babies with controlled for during
logistic regression model (not
two year period (n = 53 997) multivariate analysis
35 Evidence Level: II including family history of TSB level ≥ 342 micromol/litre, Test & Reference test described
jaundice) with the exclusion of
TSB ≥ 428 micromol/litre,
205