Thirteen studies (229,421 babies) comprising 12 cohort designs and one case-control design were included in the review. Most studies scored well on the criteria considered (above) to represent good quality, with the exception of blinding of the index test which was carried out in only one study. One cohort study was excluded from the meta-analysis because it included no newborn babies with critical congenital heart defects.
Overall sensitivity for pulse oximetry was moderate at 76.5% (95% CI 67.7 to 83.5; Ι²=25%); overall specificity was high at 99.9% (95% CI 99.7 to 99.9; Ι²=98.5%, which indicated high heterogeneity). The false-positive rate was low (0.14%; 95% CI 0.06 to 0.33) and this was particularly the case (p=0.0017) when pulse oximetry was carried out after 24 hours (0.05%; 0.02 to 0.12) compared to within 24 hours from birth (0.50%; 0.29 to 0.86).
False-positive rates were significantly increased when newborn babies suspected to have congenital heart defects were included (0.73%; 95% CI 0.50 to 1.05; p<0.0001). There were no other significant differences in sensitivity between subgroups in terms of timing of the test, test method or antenatal screening.
Substantial publication bias was reported.