Early detection of biliary atresia (BA) is important for optimal treatment. Direct bilirubin (D-bil) levels are used for BA screening. In this study, we aimed to determine the sensitivity and specificity of elevated D-bil and the direct-to-total bilirubin (D/T) ratio for BA detection in high-risk infants.


This retrospective, cross-sectional study was conducted in a tertiary medical center in Taiwan. Infants indicated for total bilirubin and D-bil measurements before age 60 days were included. The first bilirubin assessment was considered the test point. BA diagnosis was based on International Classification of Diseases, Ninth and Tenth Revision, codes 751.61 and Q44.0 to Q44.3, respectively.


Between January 2009 and December 2016, 4468 infants were enrolled, including 38 with BA. Among infants aged 3 to 60 days, a sensitivity of 100% (95% confidence interval, 90.3–100.0) was found for D-bil ≥1.0 mg/dL and either D-bil ≥1.0 mg/dL or D/T ratio ≥20%. However, D-bil ≥1.0 mg/dL had higher specificity (77.3% [76.0–78.5] vs 68.3% [66.8–69.7], respectively). In newborns aged <3 days, D-bil ≥0.5 mg/dL was considered a positive result, with a sensitivity of 50%. D-bil >0.45 mg/dL was a better cutoff point in receiver operating characteristic analysis, with a sensitivity and specificity of 100% (95% CI: 15.8–100) and 15.4% (95% CI, 11.8–19.7), respectively.


D-bil ≥1.0 mg/dL was better for BA detection than the D/T ratio in infants aged 3 to 60 days. For newborns aged <3 days, a more definitive cutoff point is required.

You do not currently have access to this content.