Abstract
OBJECTIVE: Early discharge puts neonates at risk of delayed detection of jaundice and resulting neurological injury. In these neonates, we can use cord bilirubin to make predictions. In this meta-analysis, we assessed the diagnostic accuracy of cord bilirubin in predicting the need for phototherapy (AAP-2004 or NICE-2010 charts). METHODS: We searched the databases of PubMed, Embase, Cochrane Library, Google Scholar, and Index Medicus for Southeast Asian Region. We included all observational studies that assessed the diagnostic accuracy of cord bilirubin. A bivariate model was used to pool the data in prespecified range of cord bilirubin levels (<1.5 mg/dl, 1.5-2.0 mg/dl, 2.0-2.5 mg/dl, 2.5-3.0 mg/dl, and >3.0 mg/dl). Data were pooled separately for studies including all neonates (no risk stratification), high-risk neonates (Rh and/or ABO incompatibility only), and low-risk neonates (excluded Rh and ABO incompatibility). RESULTS: Of the 1990 unique records, we studied 153 full texts and included 54 studies in the meta-analysis. For all the three groups of studies, the highest diagnostic odds ratio was noted for a cord bilirubin cut-off of 2.5-3.0 mg/dl (all neonates: 22.5, 95% CI: 21.1, 22.9; high-risk neonates: 75.5, 95% CI: 63, 85.7; low-risk neonates: 91.9; 95% CI: 64, 134.14). Using the same cut-off, the studies including all neonates without risk stratification had a pooled sensitivity of 0.31 (95% CI: 0.18, 0.47) and a pooled specificity of 0.98 (0.96, 0.99) in predicting the need for phototherapy. In studies on high-risk neonates, the pooled sensitivity was 0.8 (0.39, 0.96) and pooled specificity was 0.95 (0.78, 0.99). In studies on low-risk neonates, the pooled sensitivity was 0.74 (0.39, 0.93) and pooled specificity of 0.97 (0.91, 0.99). We noted significant heterogeneity and a high risk of bias in the index test's conduct. CONCLUSION: A cord bilirubin cut-off of 2.5-3 mg/dl has good diagnostic accuracy in predicting the need for phototherapy in neonates. REGISTRATION NUMBER: CRD42020196216.