Indications and limitations for a neonatal pulse oximetry screening of critical congenital heart disease

Abstract

Aims: Critical congenital cardiovascular malformations (CCVMs) require surgical correction during the first month of life, physical examination is unable to detect >50% of affected infants. An oximetry screening has been previously proposed. Our aim was to verify the usefulness and consistency of a pulse oximetry screening for early detection of CCVMs in a small size nursery.

Methods: A single determination of SpO2 was performed on 5292 consecutive apparently healthy newborns, discharged from nursery at a median age of 72 h during the period May 1, 2000 and November 30, 2004. Infants showing signs of congenital heart disease before the screening and those with a prenatal diagnosis were excluded. Cardiac ultrasound was performed on all infants with SpO2< or =95% at >24 h. The accuracy of the screening in identifying CCVMs was assessed by receiver-operating characteristic (ROC) curves analysis.

Results: We found 2 (0.038%) true positives, 1 (0.019%) false negative, 1 (0.019%) false positive, and 5288 (99.92%) true negatives. Prevalence of critical CCVMs was 1 in 1764. Clinical follow-up showed no evidence of CCVMs in the negative cases. A pulse-oximetry cut-off value of < or =95% showed 66.7% sensitivity (95% CI: 11.6-94.5), 100% specificity (95% CI: 99.9-100.0), 50% positive predictive value, 100% negative predictive value and AUC of 0.833 (standard error: 0.145) (95% CI: 0.823 to 0.843) in identifying CCVMs.

Conclusions: Our findings indicate that pulse oximetry is a non-invasive and specific screening tool for an early detection of CCVMs, and is easily applicable to a small size nursery.