Abstract 1666

Background: Management of jaundiced newborns often requires measurement of total serum bilirubin (TSB). Past studies of the inter- and intra-instrument and hospital variability of bilirubin analyses have shown significant discrepancies. Objective: To compare the accuracy and bias of various contemporary clinical laboratory clinical neonatal TSB methods against HPLC analysis, the "gold standard". Methods: 944 serum or plasma samples were obtained from newborns at PAH. A sample was aliquoted into an amber tube, frozen at -70°C and analyzed by HPLC and Ektachem (DT 60, Vitros NBIL slides) within two weeks at UW. TSB was measured at PAH by Hitachi 911 (automated, diazo reaction), American Optical or White (direct spectrophotometric methods) or Ektachem (DT 60-II, Vitros NBIL slides). All samples were analyzed with HPLC. Results: See table below. Correlation of each method with HPLC was similar (r≈0.98). The Root Mean Square Error (RMSE) was largest for the Ektachem NBIL. The UW and PAH NBIL RMSE values differed significantly (p<0.05), and were also larger than RMSE values of the other methods (p<0.05). The RMSE of the Hitachi, American Optical and White methods did not differ (*similar letters in the RMSE group column of the table indicate that group RMSE values did not differ significantly, p>0.05). Conclusion: Although current TSB methods in our laboratories show high correlation with HPLC, there are significant differences from HPLC and between methods. In assessing the accuracy of non-invasive methods to measure TSB, data must include HPLC and multiple clinical methods.

Table 1 No caption available
Full size table

Supported by SpectRx, Inc. Norcross, Georgia.