Sixth-Hour Trancutaneous Bilirubin and Need for Phototherapy in DAT Positive Newborns

This was a consecutively sampled, retrospective study performed at the Naval Medical Center, Portsmouth (NMCP) between May 2013 and March 2017. The NMCP institutional protocol was to type and antibody screen all pregnant mothers. All newborns of blood type O mothers had cord blood assessed for infant blood type and DAT status. DAT positive type A or B infants of type O mothers with negative antibody screens were presumed to have ABO incompatibility. DAT positive infants to type O mothers with positive antibody screens had additional testing to identify possible antigens of concern. In Rh negative mothers with otherwise negative antibody screens, infants with DAT positive tests were presumed to be because of Rhogam. By a standard order set, infants with positive cord blood DAT received capillary blood draw testing at 6 hours for total serum bilirubin (TSB), hemoglobin, hematocrit, and reticulocyte count. For this study, clinicians were encouraged to additionally order the sixth-hour TcB in the standard order set. All DAT positive infants born ≥35 weeks’ gestation age admitted to the normal newborn nursery at NMCP were included with exclusions for DAT positive infants because of non-anti-A or anti-B antibodies or maternally administered RhoGAM, or if laboratory data were not available.

TcB measurements were completed with the JM-103 (Draeger medical, Telford, PA). The devices were calibrated daily. TcB was determined by taking the highest of 3 measurements obtained at the midsternum, because this has previously been shown to have the strongest correlation with TSB measurements.¹⁸ 

Data were extracted via sequential sampling from the inpatient electronic medical record by querying the records of any infant whose cord blood was DAT positive during the study period. Demographic (birth weight, delivery type, gestational age, sex, race, and breastfeeding status) and laboratory data (infant blood type, sixth-hour TcB and TSB, hemoglobin, hematocrit, and reticulocyte count) and hours at phototherapy initiation were extracted by an automated data pull. The clinical outcomes (hours at phototherapy initiation, NICU admission, IVIG, and exchange transfusion) were all determined by manual chart review. Ten percent of the bilirubin values from the electronic data pull were validated through manual chart review and found to be 96.3% accurate.

Initiation of phototherapy was at the discretion of the attending pediatrician in accordance with the AAP guidelines for the management of hyperbilirubinemia in the newborn infants.^4,13  Pediatricians were aware of ABO incompatible DAT positive newborns and, therefore, made phototherapy decisions accounting for this risk factor (ie, used the medium risk or lower threshold for initiating phototherapy).⁴  IVIG was given for immune-mediated hemolysis with bilirubin levels approaching exchange transfusion level. Dosing was 1gm/kg over 4 to 6 hours.

Descriptive statistics were used to characterize the clinical outcomes of this population. For analysis of the predictive ability of the sixth-hour TcB to detect neonates who need phototherapy, we restricted the sample to infants with paired TcB and blood test (TSB, reticulocyte count, hemoglobin, and hematocrit) that were assessed within +/− 1 hour of the sixth hour. By pairing these tests in the same infants at the same time, we controlled for demographic variables associated with hyperbilirubinemia risk and time (bilirubin rises steadily over the first 24 hours) in our comparative models. The primary outcome was the need for phototherapy ≤24 hours. We first assessed associations of predictor variables with phototherapy ≤24 hours using univariate logistic regression. Variables associated at a significance level of P < .2 were entered into multiple logistic regression models in backward and forward stepwise fashion to arrive at final models. Because we were comparing a noninvasive screening strategy (TcB) versus blood tests, we constructed 2 models; along with the demographic variables, one had the blood test results and no TcB and the other had TcB and no blood tests. From the final models, receiver operator characteristic (ROC) curves with areas under the curve (AUC) were derived as well as the test characteristics (sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and likelihood ratios [LR]) for various cutoffs. To assess validity of the models, we used cross validation. Goodness of fit was assessed using the Hosmer-Lemeshow test. This study was approved by the Naval Medical Center Portsmouth Institutional Review Board, protocol number NMCP20130031.

The study flow diagram is demonstrated in Fig 1. Of 10 942 newborns ≥35 weeks gestational age delivered during the study period, 5167 (47.2%) were born to mothers with type O blood and had cord blood testing. The DAT was positive in 829 (7.6%) infants; 51 (0.5%) were excluded because of other antibodies; 1 infant was blood type O; and 5 had missing data or duplicate records. The demographic and clinical characteristics of the remaining 772 ABO incompatible DAT positive infants are shown in Table 1. Phototherapy was required in 281 (36.4%) infants; 156 (20.2%) ≤24 hours. There were 10 (1.3%) NICU admissions attributable to hyperbilirubinemia; each received IVIG and none required an exchange transfusion.

Paired testing (TcB and blood test within +/− 1 hour of the sixth hour) was available for a subsample of 346 of 772 newborns who contributed to the analysis of the predictive ability of the sixth-hour TcB. The paired sample (n = 346) did not differ statistically from the nonpaired sample (n = 426) on any of the characteristics in Table 1 or on phototherapy initiation ≤24 hours (21% vs 22%, respectively, P = .71). NICU admission was more likely in the paired sample (2% vs 0.5%, P = .05). The analyses for the phototherapy ≤24 hours outcome are presented in Table 2. Birth weight, infant blood type B, TSB, reticulocyte count, and TcB were significantly associated with receipt of phototherapy ≤24 hours on univariate analysis. A multivariate analysis reveals that TSB and reticulocyte count were the only independent predictors when TcB was excluded from the model; and conversely, when blood test results were excluded instead, TcB alone was the best predictor. No demographic variables were independently predictive of phototherapy ≤24 hours. The ROC curves for the 2 models are shown in Fig 2, and the AUCs for the 2 models were 0.96 and 0.90, respectively. The AUCs did not change before and after cross-validation. The Hosmer-Lemeshow test showed both models were good fits to the data, with P values of 0.65 and 0.38, respectively.

Various cutoffs and test characteristics are presented in Table 3 for phototherapy ≤24 hours. A cutoff of <3.0 mg/dL yielded an NPV of 98% (95% confidence interval [CI], 96–99) and negative LR of 0.09; 210 of 346 (60.7%) newborns were in this category, and only 5 (2.4%) below this cutoff needed phototherapy ≤24 hours. A cutoff of ≥5.3 mg/dL yielded a PPV of 85% (95% CI, 73–96); a positive LR of 20.93 to 39 of 346 (11.3%) newborns were in this category, and 33 (84.6%) needed phototherapy ≤24 hours.

Paired TSB and TcB values were compared and the sixth-hour TcB was slightly lower (median [interquartile range (IQR)] = 1.2 [0.3–1.8] mg/dL) than the TSB and highly correlated (r = 0.8, P < .01).

To our knowledge, this study is the largest published report (n = 772) to evaluate the need for phototherapy in ABO incompatible DAT positive newborns, and the results reinforce the findings of others demonstrating these newborns are a high-risk group; over one third (36.3%) needed phototherapy because of hyperbilirubinemia, as defined by the 2004 AAP guideline,⁴  which is much higher than the general newborn population rate of 6% to 9%.^19–21  The results of other significant studies of ABO incompatible DAT positive newborns revealed phototherapy rates of 48.7% (Kaplan et al, n = 164)¹  in an Israeli population and 12.9% (Schutzman, n = 240)¹⁰  in a primarily African American population using the AAP guideline recommendations for phototherapy.⁴  Olozek in Pittsburgh (n = 242)²²  and Sarici in Turkey (n = 6)¹²  observed significant hyperbilirubinemia rates of 13.7% and 100%, respectively, using defined serum bilirubin thresholds. Similar to Kaplan,¹  we found that maternal-infant blood type O-B DAT positive newborns were over twice as likely to need phototherapy than blood type O-A DAT positive infants. We did not detect any difference in race. Despite a rigorous routine at our institution of cord blood testing, a sixth-hour TSB measurement, and adherence to the AAP guideline, a small number of these infants required NICU admission and IVIG treatment.

ABO hemolytic disease is characterized by an early and rapid rise in bilirubin levels that is distinct from other causes of neonatal hyperbilirubinemia.³  In our study, the majority of infants (55.5%) who needed phototherapy required it in the first 24 hours, similar to Kaplan’s study (67%).¹  The 2004 AAP clinical practice guideline⁴  on management of hyperbilirubinemia in neonates, updated in 2009,¹³  recommends routine predischarge measurement of TSB or TcB plotted on the Bhutani hour-specific nomogram²³  for all infants, which is generally done after 24 hours. For newborns born to blood group O positive mothers, 2 options are suggested: (1) routine cord blood infant blood type and Coombs’ testing or (2) surveillance and risk assessment before discharge with selective testing for any clinically jaundiced neonate in the first 24 hours.⁴ 

Neither recommendation is ideal for identifying infants with ABO hemolytic disease. Regarding the first option, there are high costs contrasted with the relatively low incidence of severe ABO hemolytic disease.^11,16,24,25  Christensen et al, in a large regional study, found no evidence that routine cord blood testing reduced the incidence of severe hyperbilirubinemia (bilirubin ≥25 mg/dL), readmissions for hyperbilirubinemia, or kernicterus for ABO incompatible DAT positive newborns.¹¹  The second option is problematic because visual assessment is not as reliable as TSB and TcB testing,^14,15,26  and it is the rationale for screening all infants before discharge. Reliance on selective testing in the first 24 hours may miss the early presentation of some infants with ABO hemolytic disease. Studies comparing routine cord blood testing to selective testing have demonstrated selective testing decreased costs with no increase in adverse outcomes (prolonged hospital stay, NICU admission, IVIG, or exchange transfusion); however, given the relative infrequency of severe ABO hemolytic disease, the sample sizes are insufficient to draw a firm conclusion.^16,24 

Therefore, we assessed the predictive ability of a sixth hour TcB as a potential enhancement to the current AAP guideline recommendations, which might allow for earlier detection of high-risk ABO incompatible DAT positive infants. Of note, less than one-half of the 772 newborns had paired TcB and TSB for this analysis because clinicians had to add the TcB order to the standing order set. They were slow to incorporate the TcB order, given the apparent redundancy and longstanding practice of doing a TSB at the sixth hour. Although this smaller paired subsample allowed for the potential of selection bias, we saw no evidence of this in baseline characteristics or the phototherapy ≤24 hours outcome. By using logistic regression, we evaluated multiple demographic and laboratory factors associated with developing clinically significant jaundice (birth weight, delivery type, gestational age, breastfeeding status, sex, and race)^19,20,27  and determined, in ABO incompatible DAT positive newborns, a sixth-hour TcB alone, when no blood tests were done, was the best and only important predictor of phototherapy ≤24 hours. Additionally, the sixth-hour TcB was highly predictive and nearly as accurate as the combination of TSB and reticulocyte count (AUC 0.90 vs 0.96).

Furthermore, we risk-stratified the sixth-hour TcB values to triage hyperbilirubinemia monitoring. A low-risk cutoff of <3.0 mg/dL yielded a very high NPV (98%) and could eliminate about two-thirds of newborns from testing again unless clinically indicated or until the predischarge bilirubin test, as specified by the AAP guideline.^4,13  A high-risk cutoff of ≥5.3 mg/dL indicated a threshold above which 85% of infants need phototherapy ≤24 hours. For infants above this cutoff, early blood testing and assessment for hemolysis may allow for earlier initiation of phototherapy, shorter hospital stays, and less undesirable outcomes such as NICU admission, IVIG, or exchange transfusion. Newborns with TcB values of 3.0 to 5.2 mg/dL should have a repeat TcB in 6 to 12 hours to further assess the rate of rise and need for phototherapy.

To our knowledge, this study is the first published report to assess the predictive ability of a sixth-hour TcB in a large population of high-risk ABO incompatible DAT positive newborns. Many studies have demonstrated that TcB is highly correlated with TSB, especially at bilirubin levels below 13 mg/dL.^18,28,29  A few previous studies have evaluated sixth-hour TcB measurement in normal newborns,²⁷  ABO incompatible newborns with unknown DAT status,^30–32  and a small number of DAT positive infants.¹²  Our data support the observation of Stoniene et al,³¹  that TcB is strongly correlated with TSB at 6 hours, although the correlation is even stronger at 24 and 48 hours of life. The results of our study suggest that clinicians can reliably use the sixth-hour TcB to screen for hyperbilirubinemia risk in ABO incompatible DAT positive newborns, thereby avoiding blood draws for most of these infants, provided the infant blood type and DAT status are known.

Many newborn nurseries do not obtain cord blood testing of infants of blood type O mothers; therefore, clinicians at these nurseries will not know which infants are ABO incompatible and DAT positive. Given our results, there are 3 potential screening strategies that these newborn nurseries can employ: (1) continue the AAP guideline recommendation of clinical assessment with selective testing of clinically jaundiced newborns,^4,13  (2) incorporate early TcB for infants of blood type O mothers, or (3) adopt early TcB testing for all infants. The risks of option 1 were discussed earlier; using clinical assessments alone may result in missing infants with ABO hemolytic disease who tend to present in the first 24 hours. Option 2 would require nursing to be aware of the maternal blood type but would provide systematic screening not dependent on clinical assessment. Approximately 40% of mothers are blood type O positive,¹¹  and about one-third of their infants will have ABO incompatibility.^3,11  Despite this, only one-third of these infants with ABO incompatibility will be DAT positive,^3,11,22  and only one-third of this group, according to our data, will need phototherapy. This equates to 1% to 2% of all newborns born to type O positive mothers needing phototherapy from ABO incompatibility. Therefore, while focused screening of infants of O positive mothers would pick up newborns with ABO hemolytic disease, this high-risk group is blended into the overall newborn population phototherapy rate of 6% to 9%.^19–21  Option 3 would involve small increased TcB and nursing costs, and applying our high- and low-risk TcB cutoffs to a lower risk newborn population in which the DAT status is unknown could lead to increased blood testing. However, these costs might be offset by earlier detection and treatment of hyperbilirubinemia and lower NICU admissions. The high predictive ability of the sixth-hour TcB for phototherapy in a high-risk group of ABO incompatible DAT positive infants provides some reassurance that this strategy could also detect infants with other causes of severe neonatal hyperbilirubinemia, such as pyruvate kinase deficiency, hereditary spherocytosis, glucose-6-phosphate dehydrogenase, and Gilbert’s disease.^3,11  Further studies of costs are necessary before recommending a best screening strategy.

Our study has some limitations. The data are derived from a single institution whose practices may differ from others. The JM-103 TcB device may perform differently than other TcB devices at the sixth hour. We used initiation of phototherapy as an outcome measure, which is less precise than the bilirubin level crossing the recommended phototherapy threshold but reflects real-world practice. Clinicians at our institution followed the recommendations of the AAP guideline⁴  for starting phototherapy. Because all infants in this study had ABO incompatible DAT positive status, they all had a presumed risk of isoimmune hemolytic disease; therefore the medium- or high-risk curve was followed. Our suggested low-risk threshold of 3.0 mg/dL is slightly higher than the 95th percentile for sixth-hour bilirubin levels in a study of normal healthy newborns,²⁷  but the high-risk threshold of 5.3 mg/dL is slightly below the AAP guideline medium risk threshold to start phototherapy.⁴ 

Newborns who are ABO incompatible and DAT positive are at high risk for phototherapy often requiring treatment within 24 hours of birth. For this population, current screening recommendations are either not cost effective or depend on visual assessment, which may not identify jaundiced newborns in a timely manner. A sixth-hour TcB can enhance these recommendations because it is noninvasive, inexpensive, and accurate, reliably identifying infants at low and high risk of phototherapy in the first 24 hours. Newborn nurseries can use these findings to reassess their hyperbilirubinemia screening strategies.

AAP

American Academy of Pediatrics

AUC

area under the curve

CI

confidence interval

DAT

direct antiglobulin test

HOL

hours of life

IQR

interquartile range

IVIG

intravenous immunoglobulin

LR

likelihood ratio

NICU

neonatal intensive care unit

NMCP

Naval Medical Center Portsmouth

NPV

negative predictive value

OR

odds ratio

PPV

positive predictive value

ROC

receiver operator characteristic

TcB

transcutaneous bilirubin

TSB

total serum bilirubin