This subject has been reviewed in detail (284,446,447) and addressed, most recently by Lo and associates (282). Although this is one of the most commonly performed
laboratory measurements in the newborn, the measurement of TSB concentration remains remarkably inaccurate. Repeated surveys over the last 3 decades have disclosed a high level of interlaboratory variation in the measurements of both total and direct-reacting serum bilirubin concentrations in neonatal sera. In a recent study, a sample with a known TSB concentration of 14.8 mg/dL (243 μmol/L) was analyzed in 14 different university hospital laboratories. The mean (± SD) measured TSB was 15.2 ± 2.5 mg/dL (coefficient of variation 16.4%) and the range was 12.1 to 18.5 mg/dL (207 to 316 μmol/L) (283). This remarkably wide range of values in skilled hands suggests that considerable caution is necessary before generalizing bilirubin levels from one institution to the wider universe of newborns. Within-laboratory variation generally is considered to be lower, but in the study of Vreman and associates (283), over time, the coefficient of variation for repeated measurements of the same sample was as high as 17.2% in one laboratory (283).

The most recent analysis of laboratory neonatal TSB measurements was conducted by Lo and associates (282) who examined data submitted by laboratories that participate in the College of American Pathologists (CAP) neonatal bilirubin and chemistry surveys. They compared the data from laboratories throughout the United States with the reference method for measuring TSB (448). They found an acceptable coefficient of variation ranging from 1.9% to 4.5%, in the four most commonly used laboratory instruments. When compared with the reference method, the mean values from these instruments differed from the reference method by —21.6% to 10.9%. The authors attributed these discrepancies to the use of different methods used (primarily with the Vitros instrument, Ortho-Clinical Diagnostics, Raritan, NJ) in the neonatal bilirubin versus the chemistry surveys of the CAP, as well as the presence of a nonhuman protein base and ditaurobilirubin in the survey specimens (115). Lo and associates concluded that it is impossible to evaluate the accuracy of TSB measurements from the CAP surveys because the standard specimens consist of bovine serum containing a mixture of unconjugated bilirubin and ditaurobilirubin. They recommend that survey specimens should consist of human serum enriched with unconjugated bilirubin. The CAP has adopted this recommendation and is now using human-serum-based specimens in their neonatal bilirubin and chemistry surveys (BT Doumas, personal communication, October, 2003).

These variations between laboratories might explain the frequent occurrence in clinical practice of an infant being admitted for treatment of hyperbilirubinemia because an outside laboratory found a high TSB level, but when the test is repeated in the hospital laboratory, the TSB level is 5 or 6 mg/dL (85 to 103 μmol/L) lower. Of course, there is no way of knowing which value is correct. A 16.4% coefficient of variation between laboratories means that if the true serum bilirubin value is 20 mg/dL (342 μmol/L), the 95% confidence limits of a repeat measurement at another laboratory could fall anywhere between 14.4 and 26.6 mg/dL (246 to 455 μmol/L). Because our followup, surveillance, and intervention in jaundiced infants are based on TSB values, spurious underestimation of the TSB concentration might lead to withholding of necessary therapy, and overestimation will produce unnecessary clinical intervention.

Similar concerns exist with regard to measurements of direct-reacting or conjugated bilirubin where laboratory measurements are considered, at best, “only an approximation to the true value” (449). In one study, measurements of direct-reacting bilirubin using the Dupont ACA (automatic clinical analyzer) or Coulter Dacos analyzer produced direct-reacting bilirubin levels in a population of full-term newborns at one hospital that were twice as high as those measured in a similar population at another institution, where a Kodak Ektachem 700 (now known as the Vitros) method was used (450). The Vitros method measures conjugated bilirubin, whereas the other methods measure direct-reacting bilirubin. Although often used synonymously, direct-reacting bilirubin is not the same as conjugated bilirubin. Direct-reacting bilirubin refers to the bilirubin that reacts directly with diazotized sulfanilic acid (i.e., without the addition of an accelerating agent), whereas conjugated bilirubin refers to bilirubin that has been made water soluble by binding with glucuronic acid in the liver.

Data regarding the differences in TSB levels when measured in capillary or venous samples are conflicting (451,452). It is useful to recall, however, that virtually all of the published data regarding the relationship of TSB levels to kernicterus or developmental outcome are based on capillary TSB levels. Thus, for the purposes of clinical decision making, capillary blood samples are the gold standard and there is no reason to delay initiation of treatment to obtain a venous blood sample to “confirm” an elevated capillary TSB level.