Hemolysis in newborn blood samples can lead to falsely decreased direct bilirubin test results, a common problem that many healthcare institutions deal with by rejecting hemolyzed neonatal samples outright and having them redrawn. Though necessary for the accurate diagnosis and management of neonatal liver conditions, these repeat blood collections are a serious stressor for infants. With the aim of reducing them, researchers recently developed an equation to predict the concentration of direct bilirubin in hemolyzed samples. Their findings are examined in this issue of Strategies.

Most hospitals screen all newborns for total bilirubin to prevent the severe neurological damage that can occur if, following birth, this byproduct of hemoglobin breakdown exceeds a certain concentration. If an infant tests positive for hyperbilirubinemia—or unusually high total bilirubin—then a direct bilirubin test may also be performed to determine if the cause is cholestatic liver disease, a potentially fatal condition that affects one in 2,500 newborns.

For newborns with the cholestatic liver disease extrahepatic biliary atresia, liver surgery at 60 days or younger is the number one predictor of good prognosis, making early detection critical to survival. A common obstacle when diagnosing this condition, though, is that newborn samples tend to have high levels of hemolysis, mostly due to the increased fragility of red blood cells during the first 2 weeks of life. Clinicians then have to redraw these samples, which delays diagnosis and raises the risk of complications from capillary heel stick collection such as structural foot damage or collapse of veins.

While the effect of hemolysis on total bilirubin measurement has been studied extensively, few have investigated the impact of hemolysis on direct bilirubin test results. Researcher Dina Greene, PhD, became aware of this when her lab switched from measuring bilirubin using Vitros analyzers—an instrument not affected by hemolysis—to using the Beckman AU680 analyzer. Greene is a scientific director at Kaiser Permanente's Northern California Regional Lab in Berkeley, Calif., and lead author of a paper examining this overlooked problem (Clin Chim Acta 2014;429:194–7).

"One of the big questions that we needed to answer before we let our physicians know that we were changing analyzers is what was going to happen with hemolyzed specimens from babies," said Greene. Her lab did not want to start rejecting high rates of newborn samples, particularly from sick infants who have to undergo multiple direct bilirubin tests for disease monitoring. "Say they had just treated the baby. And now the concentration of direct bilirubin should be going down, but it's a hemolyzed specimen, and they're thinking, 'this poor baby just had liver surgery; do I have to redraw the sample again?' Our thought was, 'Let's see if we can do some sort of experiment to try to mitigate rejection criteria of specimens that are hemolyzed that come in for direct bilirubin.'"

To do this, Greene's team took 32 non-hemolyzed adult serum samples with varying known concentrations of direct bilirubin and induced hemolysis in these samples by adding titrated amounts of hemoglobin. This created several series of six to seven samples with identical direct bilirubin concentrations but differing levels of hemolysis. Using the AU680, the researchers then measured direct bilirubin and the amount of hemolysis in each sample.

A comparison of the direct bilirubin values from the hemolyzed samples with the original values revealed that the degree of interference from hemolysis depends on both the initial direct bilirubin concentration and the hemoglobin concentration. Greene and her colleagues then determined that the true concentration of direct bilirubin in a hemolyzed sample can be predicted using the equation D0 = (0.334H0.334) D1.56H^(-0.14), where D0 is the corrected direct bilirubin concentration, H is the hemoglobin concentration, and D is the measured direct bilirubin concentration.

"What these authors did was very elegant," said Brad Karon, MD, PhD, director of hospital clinical laboratories at Mayo Clinic in Rochester, Minn. "I think it's going to be very valuable information for laboratories that have large neonatal and pediatric practices."

There are limitations to Greene's study, however, that could prevent other labs from easily applying this paper's findings to their individual practices. The main one is that the correction equation her team developed, as noted in their paper, is unlikely to work with instruments other than the AU680 because assay platforms aren't harmonized. Labs using other analyzers could repeat this experiment to develop their own equations, but Karon pointed out that this would be "a tremendous amount of work."

"And secondly, I think a lot of laboratorians have a lot of concerns about how robust any correction equation can be," added Karon. "Clinicians may view a change of 1.4 to 1.3 as a change in the patient's status, but it just may be the limitation of how good the correction equation is."

Greene and her co-authors addressed this potential shortcoming by suggesting that labs use the equation to calculate 95% confidence intervals for predicted direct bilirubin concentrations. Labs can then use these confidence intervals as a guide when advising clinicians on whether or not to redraw a blood sample.

As an example of how this process is used in Greene's labs, for a neonatal direct bilirubin specimen with a hemoglobin concentration of about 62 mg/dL and a measured direct bilirubin concentration of 1.62 mg/dL, the 95% confidence interval for predicted direct bilirubin (as calculated using the equation) would be 1.52–1.65 mg/dL. This is approaching 2.0 mg/dL, the 99th percentile for direct bilirubin in the first 2 weeks of life, above which a direct bilirubin result would be flagged as pathological. When the lab returns this result to the physician, it would therefore include an alert stating that the specimen was mildly hemolyzed and a redraw should be considered if clinically indicated.

Karon believes that this is the most valuable aspect of Greene's study—that it serves as an example of how to make and implement 95% confidence limits for direct bilirubin. "What they ended up recommending in practice is probably a more practical approach than for labs to develop their own correction equations," he said, "Other labs wouldn't be able to use the correction equation itself or the 95 percent confidence limits, but they could define a set of samples with some relatively small amounts of direct bilirubin, let's say less than 2 mg/dL, and mild to moderate amounts of hemoglobin. From these, they could generate their own 95 percent confidence limits that would allow them to tell pediatricians, 'at most there's no more than 1–2 mg/dL direct bilirubin.' Generally, along with the total bilirubin value, that's enough information to exclude a potentially pathologic cause for bilirubinemia and I think that would avoid the majority of redraws, on infants anyway."

Overall, Karon emphasized that even if labs do not adopt this study's approach to reducing redraws, it is critical that they find some way to tackle the problem of rejecting high numbers of direct bilirubin measurements from infants. "Pediatricians and parents of these infants do consider a repeat blood draw a fairly major event and an undesirable event, and we should in the lab try to be better and not have to redraw these specimens if we don't absolutely need to. Labs should find some way to address this issue, and I think the solution that Dr. Greene and her colleagues have proposed is a very reasonable one."