search text   Advanced Search

This Article Abstract Full Text (PDF) P3Rs: Submit a response Alert me when this article is cited Alert me when P3Rs are posted Alert me if a correction is posted Citation Map Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Keren, R. Articles by Bhutani, V. K. Search for Related Content PubMed PubMed Citation Articles by Keren, R. Articles by Bhutani, V. K. Related Collections Premature & Newborn

A Comparison of Alternative Risk-Assessment Strategies for Predicting Significant Neonatal Hyperbilirubinemia in Term and Near-Term Infants

^a Division of General Pediatrics, Center for Pediatric Clinical Effectiveness
^c Division of Biostatistics and Epidemiology, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
^b Department of Pediatrics, University of Pennsylvania, Philadelphia, Pennsylvania
^d Department of Neonatal and Developmental Medicine, Lucile Packard Children's Hospital, Stanford University, Palo Alto, California

	ABSTRACT

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
OBJECTIVE. The purpose of this work was to compare the predictive accuracy of alternative risk-assessment strategies used to screen for the risk of significant neonatal hyperbilirubinemia.

PATIENTS AND METHODS. We conducted a prospective cohort study of 823 term and near-term newborns admitted to the well-infant nursery at the Hospital of the University of Pennsylvania. Maternal, infant, and delivery risk factors for significant hyperbilirubinemia were obtained from chart review, structured interviews with parents, and nurse assessments before discharge. Transcutaneous bilirubin measurement was performed daily until discharge and once by a visiting home nurse between 3 and 8 days of life. We used the c statistic to compare the predictive accuracy of 3 risk-assessment strategies for estimating the risk of significant neonatal hyperbilirubinemia, defined as a bilirubin level that at any time after birth exceeded or was within 1 mg/dL (17 µmol/L) of the hour-specific phototherapy treatment threshold recommended by the American Academy of Pediatrics in 2004. The compared strategies included those that use (1) a predischarge bilirubin level (obtained before 52 hours) expressed as a risk zone on an hour-specific bilirubin nomogram, (2) clinical risk factors other than the predischarge bilirubin level, and (3) a combination of the predischarge bilirubin risk zone and additional clinical risk factors.

RESULTS. Forty-eight patients (6%) developed significant neonatal hyperbilirubinemia. The predischarge (<52 hours) bilirubin level expressed as a risk zone on the bilirubin nomogram and a prediction model that combined multiple other clinical risk factors had similar accuracy for predicting significant hyperbilirubinemia. The only clinical risk factor that could be added to the predischarge risk zone to improve overall predictive accuracy was gestational age. The predischarge bilirubin risk zone and gestational age could be used to stratify patients into a large group (n = 523 [70%]) of infants with a very low (0.2%) risk of developing significant hyperbilirubinemia, a small group of infants (n = 127 [17%]) with a low (4%) risk of developing significant hyperbilirubinemia, and an even smaller group of infants (n = 100 [13%]) with a high (42%) risk of developing significant hyperbilirubinemia.

CONCLUSIONS. An infant's risk of developing significant hyperbilirubinemia can be simply and accurately assessed by using just the infant's predischarge bilirubin level and gestational age.

Key Words: neonatal hyperbilirubinemia • jaundice • prediction model • risk assessment • newborn screening

Abbreviations: AAP—American Academy of Pediatrics • GA—gestational age • HUP—Hospital of the University of Pennsylvania • TcB—transcutaneous bilirubin • TSB—total serum bilirubin • G6PD—glucose-6-phosphate-dehydrogenase • ROC—receiver operating characteristic • LR—likelihood ratio • IQR—interquartile range • OR—odds ratio • CI—confidence interval

The American Academy of Pediatrics (AAP) clinical practice guideline on the management of neonatal hyperbilirubinemia recommends that all newborn infants be assessed before discharge for the risk of developing significant neonatal hyperbilirubinemia.¹ The guideline recommends 2 risk-assessment options, used individually or in combination: predischarge measurement of the bilirubin level and assessment of clinical risk factors. Bilirubin values obtained before discharge can be plotted on an hour-specific bilirubin nomogram to calculate an infant's bilirubin percentile with respect to age in hours. The predischarge bilirubin "risk zone" (0–40th, 41st to 75th, 76th to 95th, and >95th percentile corresponding with low, low-intermediate, high-intermediate, and high risk zones on the hour-specific bilirubin nomogram) has been shown to be a strong predictor of subsequent risk of hyperbilirubinemia.^2–5 Clinical risk factor assessment involves consideration of an array of maternal, infant, and delivery factors known to be associated with neonatal hyperbilirubinemia. A few studies have formalized the clinical risk factor assessment process by developing scoring systems that incorporate the weighted contribution of multiple risk factors at once.^{3, 4, 6}

Previous research suggests that the predischarge bilirubin expressed as a risk zone on an hour-specific bilirubin nomogram is more accurate than formal clinical risk factor assessment for determining the risk of developing significant neonatal hyperbilirubinemia.^{3, 4} One study has suggested that the addition of clinical risk factors, such as gestational age (GA), can significantly increase the predictive accuracy of the predischarge bilirubin.⁴ However, the accuracy of clinical risk factor assessment in these retrospective studies may have been attenuated by omission of important clinical predictors (eg, breastfeeding, severity of jaundice on physical examination, and history of a previous sibling with jaundice or phototherapy). The retrospective nature of these studies may have also resulted in inaccurate risk factor assessment, as well as incomplete and possibly biased sampling of bilirubin measurements.

In this study, we prospectively enrolled a large cohort of newborn infants, performed exhaustive review of clinical risk factors, and systematically measured transcutaneous bilirubin values on nearly all of the enrolled infants throughout the first week of life. Recognizing that many clinicians rely solely on clinical risk factors for assessing the risk of hyperbilirubinemia, we sought to provide comparative data about the predictive properties of the predischarge bilirubin for clinicians considering whether to adopt predischarge bilirubin screening. Cognizant of the fact that clinicians often use both laboratory and history/physical examination data in making clinical decisions, our secondary goal was to describe the incremental predictive value of a risk-assessment strategy that incorporates both clinical risk factors and a predischarge bilirubin value.

	METHODS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Design, Setting, and Subjects
This prospective cohort study was conducted in the well-infant nursery of the Hospital of the University of Pennsylvania (HUP), an urban tertiary care hospital with 3800 deliveries per year. Infants were eligible for enrollment if they were managed exclusively in the well infant nursery and if they were 36 weeks' GA and weighed 2000 g or if they were 35 weeks' GA and weighed 2500 g. The birth weight criteria were used to exclude preterm (<35 week GA) infants who might otherwise be included because of inaccuracies in the estimation of GA. Infants who were transferred to the intensive care nursery for any reason and those who received >48 hours of intravenous antibiotics for concern for sepsis were excluded from the study.

Recruitment and Sampling Scheme
Because the patient population at HUP is predominantly black, we used a stratified sampling scheme to oversample nonblack infants to assemble a study sample from which results would be more generalizable. Each weekday morning a research assistant compiled a list of infants born in the previous 1 to 2 days stratified by maternal race and used a random number generator to randomize the order of the infants within racial strata. The research assistant then approached parents of infants in the order designated on the list, starting with mothers from race categories of the lowest prevalence at the HUP nursery ("other" category, which included American Indian, Alaskan Native, Native Hawaiian, and Pacific Islander), proceeding to race categories of increasing prevalence (Asian, Hispanic or Latino, white, and black). A maximum of 6 infants were enrolled each day.

Study Protocol
The HUP well-infant nursery adopted universal transcutaneous bilirubin (TcB) screening in June 2004, 3 months before the start of our study. Our study protocol differed only slightly from routine birth hospitalization care in that participants had TcB measurements obtained daily until discharge instead of only 1 routine measurement the night before discharge. TcBs were measured using BiliCheck (Respironics, Murrysville, PA), a Food and Drug Administration-approved noninvasive bilirubinometer. Conforming with the standard of care in the nursery, infants whose TcB measurement exceeded the 75th percentile on an hour-specific bilirubin nomogram^{7, 8} had a total serum bilirubin (TSB) measurement obtained. Because the BiliCheck is less accurate at TcB levels >13 mg/dL (222 µmol/L),^{8, 9} our protocol also dictated that all TcB values 12 mg/dL be confirmed with a TSB. Additional bilirubin measurements and therapeutic interventions (phototherapy or exchange transfusion) were performed during the birth hospitalization at the discretion of the health care team. No decision support was provided as part of the protocol. Also, no additional bilirubin measurements were obtained as part of the study protocol after infants were started on phototherapy.

Bilirubin measurements generally peak at age 3 to 5 days, and so we attempted to obtain bilirubin measurements on all of the study infants in this time frame. If an infant's birth hospitalization was >72 hours, we obtained at least 1 TcB measurement before discharge between 72 and 120 hours of life. For infants who were discharged from the hospital before 72 hours of life, a study nurse equipped with a BiliCheck device attempted to perform a home visit to obtain a TcB between 72 and 120 hours of life. Similar to the birth hospitalization protocol, the study nurse obtained a heel stick TSB measurement if the TcB recorded at the home visit was 12 mg/dL (205 µmol/L). Decisions to obtain additional bilirubin measurements and to initiate phototherapy after discharge were made by the primary care physicians, who were informed by nurses of all of the TSB values that exceeded the 75th percentile on the hour-specific bilirubin nomogram.

If a high TcB triggered the performance of a TSB (either because the TcB exceeded the 75th percentile before discharge or was 12 mg/dL at any time), we used the TSB instead of the TcB for analyses. This ensured that all of the bilirubin values (TcB when TcB was <12 mg/dL and TSB when TcB was 12 mg/dL) were valid measures of TSB.

Predictor Variables
Predischarge bilirubin values used for prediction were those obtained before 52 hours of age. We used an hour-specific bilirubin nomogram⁷ to convert the predischarge predictor bilirubin values into risk zones (low, 0-40th percentile; low-intermediate, 41st to 75th percentile; high-intermediate, 76th to 95th percentile; and high, >95th percentile). For infants who had >1 bilirubin value obtained before 52 hours of age, we selected the bilirubin value corresponding with the highest risk zone to serve as the predictor value. If the predischarge (<52 hours) bilirubin corresponding with the highest risk zone also met our criteria for significant hyperbilirubinemia, then that bilirubin served as both a predictor and an outcome value (in which case the predictor predicted the outcome perfectly).

Information on maternal, infant, and delivery characteristics was abstracted from chart review by 2 research assistants. Maternal race, intended method of feeding after discharge (breast, bottle, or both), and history of a previous infant with jaundice or requiring phototherapy were reported by the mother during structured interviews before discharge. GA was abstracted from the obstetric record and was generally estimated from fetal ultrasound or last menstrual period. For the purposes of analysis we rounded GA to the nearest whole week (eg, 37 weeks = 38 weeks) and categorized the variable into 3 levels (<38 weeks, 38–39 weeks, and 40 weeks). We used 38 weeks as the first cutoff because we anticipated a discontinuity in the development of significant hyperbilirubinemia at 38 weeks, where the threshold defining significant hyperbilirubinemia jumps to a higher curve. Jaundice assessments to estimate the extent of progression of jaundice were performed on the day after the second night of hospitalization (or the day of discharge if the infant was discharged earlier) to simulate the timing of predischarge decision-making during a typical vaginal birth hospitalization stay. Jaundice assessments were performed by well-infant nursery nurses with 2 years of nursery experience who were not caring for the infant being evaluated and were blinded to the infant's bilirubin values. Nurses used the homunculus shown in Fig 1 ¹⁰ to assess the maximum extent of jaundice on a 5-point scale. The presence of 1 of 5 glucose-6-phosphate dehydrogenase (G6PD) mutations was determined from review of the newborn screen results (Pediatrix, Bridgeville, PA) after hospital discharge. The G6PD mutation analysis included in the Pediatrix screen identifies 5 mutations known to account for >90% of all affected infants.¹¹ With the exception of the G6PD mutation result (which is not available until 2 weeks after delivery), all of the information about predictor variables was collected before development of the outcome.

View larger version (22K): [in this window] [in a new window]   FIGURE 1 Schema for grading the extent of the progression of jaundice.a10 a Nurses blinded to bilirubin values were instructed to grade the extent of jaundice by assigning a number corresponding with the maximum extent of jaundice progression. For example, an infant found to have jaundice extending to the upper arms (grade 4) and upper legs (grade 3) was categorized as having grade 4 jaundice.

We imported the published phototherapy treatment guidelines into Microsoft Visio (Redmond, WA) and used Visio's coordinate plotting function to estimate the hour-specific phototherapy treatment threshold values for all 3 of the treatment threshold curves. We followed the AAP guideline in using specific risk factors for bilirubin neurotoxicity to decide which phototherapy treatment threshold curve to use in determining whether a study infant's bilirubin exceeded the threshold. Because infants with asphyxia, lethargy, temperature instability, sepsis, and acidosis were excluded from the study and G6PD status and albumin level were not known at the time of discharge, the only risk factors that we used to select the appropriate phototherapy treatment threshold curve were GA and Coombs positivity.

Analysis
The primary outcome for this study was the predictive accuracy of alternative strategies for assessing an infant's risk of developing significant hyperbilirubinemia. We compared the predictive accuracy of strategies that use the following: (1) a predischarge bilirubin (obtained at <52 hours) expressed as a risk zone; (2) multiple clinical risk factors other than the predischarge bilirubin; and (3) a combination of the predischarge bilirubin risk zone and additional clinical risk factors. Overall predictive accuracy was measured using the c-statistic,¹² which is a measure of discrimination—the ability of a test or predictive model to distinguish between patients with and without the outcome of interest (in this case, the development of significant hyperbilirubinemia). The c-statistic is mathematically equivalent to the area under the receiver operating characteristic (ROC) curve¹³ for the risk-assessment strategy.

The association between individual risk factors (predictors) and the outcome of interest was estimated using ² tests and logistic regression. Three logistic regression models were developed to predict the outcome of significant hyperbilirubinemia: (1) a model that included predischarge (<52 hours) bilirubin risk zone alone; (2) a model that included clinical risk factors other than predischarge bilirubin risk zone; and (3) a model that combined the predischarge bilirubin risk zone and other clinical risk factors. To develop models 2 and 3, we added clinical risk factors individually that were likely to explain variation in the outcome based on their role in the various mechanisms of bilirubin accumulation (increased production, decreased metabolism, and increased enterohepatic circulation) and that were associated with the outcome at a P value <.05 in bivariate analyses. To remain in the model, the added variables had to continue being significant at a P value <.05 level. Model calibration was tested using the Hosmer-Lemeshow goodness-of-fit statistic. We plotted ROC curves and calculated and compared the area under the ROC curves for each of the 3 models using an algorithm described by DeLong et al.¹⁴ To estimate the predictive accuracy lost by categorizing the predischarge bilirubin variable into 4 discrete risk zones, rather than treating it as a continuous variable, we used methods suggested by Newman et al⁴ to generate a z score associated with each predischarge bilirubin value and calculated the discrimination of the predischarge bilirubin z score for predicting significant hyperbilirubinemia. Likelihood ratios (LRs) were calculated for multilevel risk strata derived from the combination model (model 3).

All of the data were collected by research assistants and entered directly into a Microsoft Access database. Analyses were performed using SAS 9.1 (SAS Institute, Cary, NC) and Stata 8 (Stata Corp, College Station, TX). Informed consent was obtained from mothers of all of the infants in the study. The institutional review boards from the University of Pennsylvania School of Medicine and the Children's Hospital of Philadelphia approved the study protocol.

	RESULTS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Subject Characteristics
Between September 2004 and October 2005, 1749 infants were born on days when recruitment occurred. A total of 1079 families (62%) were approached for enrollment in the study and, of those, a total of 823 (76%) consented to participate. Eleven patients were excluded from the study after they developed exclusion criteria (7 transferred to an intensive care nursery, and 4 received antibiotics for >48 hours), leaving a total of 812 infants. All but 1 of these infants had a bilirubin value obtained before 52 hours (predictor bilirubin value obtained at a mean age of 27 hours; interquartile range [IQR]: 20–32 hours). We were able to ascertain the development of significant hyperbilirubinemia in 751 infants (92%): 47 (6%) who received phototherapy during the birth hospitalization (mean age at time of initiation: 43 hours; IQR: 27–51 hours) and 704 (87%) who did not receive phototherapy during the birth hospitalization and had a bilirubin value obtained between 72 and 192 hours (mean age: 97 hours; IQR: 82–120 hours). Sixty-one infants (7%) were lost to follow-up and did not have birth hospitalization phototherapy or a bilirubin obtained after 72 hours (Fig 2).

View larger version (31K): [in this window] [in a new window]   FIGURE 2 Enrollment and follow-up of study subjects. a Twenty three of the 703 infants classified as not developing the outcome of significant hyperbilirubinemia received phototherapy without a bilirubin level that met our criterion for significant hyperbilirubinemia.

Infant, maternal, and pregnancy/delivery characteristics from Table 1 that were associated with the development of significant hyperbilirubinemia at a P value level <.05 are listed in Table 2. The factor most strongly associated with development of significant hyperbilirubinemia was the predischarge bilirubin risk zone. None of the 218 infants with predischarge bilirubins in the low-risk zone developed significant hyperbilirubinemia, and only 2 (1%) of the infants with predischarge bilirubins in the low-intermediate risk zone developed significant hyperbilirubinemia. Compared with infants who had predischarge bilirubins in the low or low-intermediate risk zones, the odds ratio (OR) for developing significant hyperbilirubinemia among infants with predischarge bilirubin levels in the high-intermediate and high-risk zones was 21 (95% confidence interval [CI]: 4.9–93.0) and 147 (95% CI: 34–639), respectively. GA <38 weeks (OR: 9.2; 95% CI: 4.4–19) and breastfeeding were also strongly associated with development of significant hyperbilirubinemia. Mother's stated intention to give both breast and bottle feeds (OR: 3.7; 95% CI: 1.6–8.6) was more strongly associated with the outcome than a stated plan to breastfeed exclusively (OR: 2.2; 95% CI: 1.0–4.5). Grade 4 or higher jaundice was also strongly associated with the development of significant hyperbilirubinemia (OR: 6.0; 95% CI: 2.1–17.0). In analyses stratified by race, the association between grade 4 or higher jaundice and development of significant hyperbilirubinemia was present only for nonblack infants (OR: 6.9; 95% CI: 2.2–22.0). Black infants did not demonstrate this association: 7 of 353 black infants with grade 0 to 3 jaundice developed significant hyperbilirubinemia, whereas 0 of 10 with grade 4 or higher jaundice did (OR: not calculable; P = .81). Interestingly, of the 140 infants (both black and nonblack) who had no jaundice noted (jaundice score = 0), only 2 developed significant hyperbilirubinemia (negative predictive value = 98.6%), both of whom were white and had a predischarge bilirubin in the high-intermediate (n = 1) and high-risk (n = 1) zones. A positive G6PD DNA analysis was not associated with the development of significant hyperbilirubinemia in either the entire sample (OR: 0.99; 95% CI: 0.23–4.3) or in the subgroups of male infants, black infants, or black male infants. Factors associated with a decreased risk of significant hyperbilirubinemia included black race (OR: 0.43; 95% CI: 0.23–0.80) and maternal history of smoking during the pregnancy.

View larger version (17K): [in this window] [in a new window]   FIGURE 3 ROC curves for risk-assessment strategies using predischarge bilirubin level, other clinical risk factors, or both. a The clinical risk factors model included GA, method of feeding, black race, extent of jaundice, and gender. b The combination model included predischarge bilirubin risk zone, GA, and percentage of weight loss per day over the first 2 days.

Although all of the factors in the combination model were statistically significant, we found that the only added variable that increased the model's predictive accuracy above that derived from the predischarge bilirubin risk zone alone was GA (c-statistic = 0.952 vs 0.954 for the full model). To illustrate how GA contributed to the predictive performance of the predischarge bilirubin, we calculated the proportion of infants who developed significant hyperbilirubinemia stratified by these 2 parameters (Fig 4). The figure shows that, for children with predischarge bilirubins in the high-intermediate or high-risk category, GA <38 weeks increases the risk of significant hyperbilirubinemia several fold. Figure 5 demonstrates how information on the predischarge bilirubin risk zone and GA can be used to classify infants into 1 of 3 distinct groups in terms of their risk of developing significant neonatal hyperbilirubinemia (very low risk: 0.2%, LR: 0.028; low risk: 4.0%, LR: 0.600; and high risk: 42.0%, LR: 10.600).

View larger version (13K): [in this window] [in a new window]   FIGURE 4 Probability of significant hyperbilirubinemia given predischarge bilirubin and GA.

View larger version (21K): [in this window] [in a new window]   FIGURE 5 Risk of significant hyperbilirubinemia according to predischarge bilirubin percentile and GA.

	DISCUSSION

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

Our analyses suggest that knowledge of just these 2 easily obtained and objectively determined factors can help a clinician to stratify patients into 3 distinct risk categories. Infants in the "very low" risk category can be followed clinically and may be spared follow-up bilirubin measurements, assuming the physical examination and history do not change dramatically to warrant a repeat bilirubin measurement. Infants in the "low" risk category can also be managed expectantly, but the threshold for performing a follow-up bilirubin may be a little lower than for the "very low" risk group. Finally, the infants who are categorized as "high" risk, depending on the height of their predischarge bilirubin, may warrant initiation of phototherapy according to AAP treatment guidelines, a follow-up bilirubin in 24 hours, or delayed discharge from the birth hospitalization if follow-up cannot be guaranteed.

Our results are fairly consistent with those reported by other investigators. Although defined differently across studies, significant hyperbilirubinemia has repeatedly been shown to be associated with the predischarge bilirubin risk zone,^{3–5, 7} GA <38 weeks,^{2–4, 6, 15} and breastfeeding.^{2–6, 16} Similar to the study by Newman et al⁶ on clinical risk factors for significant hyperbilirubinemia, we also found black race to be protective, but, unlike in that study, female gender was associated with an increased, rather than decreased, risk of hyperbilirubinemia in our subjects. Consistent with previous studies,^{3, 4} we found that risk-assessment strategies that combine multiple clinical risk factors (other than the predischarge bilirubin) have similar (but not better) accuracy than a strategy that relies on the predischarge bilirubin value alone. Similar to Newman et al,⁶ we also found that, after the predischarge bilirubin risk zone, GA was the strongest predictor of significant hyperbilirubinemia and could be used in combination with the predischarge risk zone to stratify infants into distinct risk categories. Interestingly, some of the factors listed as major risk factors in the AAP Practice Guideline on Management of Neonatal Hyperbilirubinemia were not found to be significant predictors of significant hyperbilirubinemia in our study, including the presence of cephalohematoma (OR: 1.8; 95% CI: 0.63–5.40) or bruising (OR: 1.5; 95% CI: 0.44–5.10) and a history of a sibling who received phototherapy (OR: 1.1; 95% CI: 0.46–2.70), although the OR point estimates and 95% CIs are consistent with those reported previously for these variables.

Our study is the first to prospectively ascertain clinical risk factors and to compare alternative strategies for assessing the risk of significant hyperbilirubinemia. Its prospective nature allowed us to perform an exhaustive review of potential clinical risk factors, including historical factors, such as a history of a previous child with jaundice or requiring phototherapy and self-reported maternal race. The availability of a study nurse to visit infants at home after discharge enabled us to perform bilirubin measurements during the time that bilirubin values peak on nearly all of the study infants, and, thus, avoided the potential for verification bias that was present in previous retrospective studies. Our study is also the first to incorporate a formal assessment of the extent of jaundice before discharge as a predictor of significant hyperbilirubinemia. Although the extent of jaundice did not add to the predictive accuracy of the predischarge bilirubin risk zone, it did remain significant in the multivariable clinical risk factor model. The fact that grade 4 or higher jaundice was predictive of subsequent significant hyperbilirubinemia only in nonblack infants supports the notion that the presence and degree of jaundice may be more difficult to discern in infants with darker skin pigmentation.^{17, 18} Interestingly, the complete absence of jaundice (grade 0) had high negative predictive value (98%) for subsequent significant hyperbilirubinemia in both black and white infants.

Finally, our study is the first to evaluate the relationship between the results of DNA-based screening for G6PD deficiency and the development of significant hyperbilirubinemia. Our results suggest that even if the DNA-based test result could be made available before discharge, it may not be helpful in predicting the risk of hyperbilirubinemia requiring phototherapy, even in subgroups of male, black, or black male infants. Scenarios that may compromise the predictive accuracy of DNA-based G6PD deficiency testing include detected mutations associated with adequate enzyme levels, detected mutations that result in low enzyme activity levels but not hyperbilirubinemia, and undetected mutations that result in low enzyme activity and hyperbilirubinemia.¹¹ Further investigation is needed to determine the role of G6PD mutation analysis, as well as alternative (G6PD enzyme level measurement^{19, 20}) and supplementary (UDPGT1 polymorphism²¹) testing for identifying infants who need close monitoring for the risk of hemolysis, severe hyperbilirubinemia, and possibly kernicterus related to G6PD deficiency.

Our study had a few limitations. Our sample size allowed us to detect only risk factors that were strongly associated with the outcome of interest or that were prevalent. We were reassured, however, that the only factor that added significantly to the predictive accuracy of the predischarge bilirubin risk zone, GA, is the one that is most strongly associated with the predicted outcome. Second, some infants were started on phototherapy before meeting our outcome criterion, which prevented us from determining whether they would have developed the outcome of interest. However, sensitivity analyses demonstrated no major differences in the relative accuracy of the 3 evaluated strategies when these infants were excluded or treated as having developed the outcome. Finally, despite our attempts to oversample infants of nonblack race, approximately half of the study infants were born to black mothers. If, as suggested by our data, black race is associated with a lower risk of significant hyperbilirubinemia, then patient populations with fewer black infants may have higher rates of significant hyperbilirubinemia, both overall and in various risk strata, than reported in our study.

	CONCLUSIONS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Strategies that use either clinical risk factors or predischarge bilirubin values to assess the risk of significant hyperbilirubinemia have similar predictive accuracy, although studies are needed to prospectively validate the accuracy and reliability of specific clinical risk factor prediction rules. The most accurate risk-assessment strategy incorporates information about both predischarge bilirubin values and clinical risk factors, specifically GA. Knowledge of just these 2 factors provides a simple and accurate way to assess an infant's risk of developing significant hyperbilirubinemia.

Future research on risk-assessment strategies for significant hyperbilirubinemia will need to identify the actions that should be taken given the predicted risk and quantify the "cost" of the "errors" in risk assessment (both false-positives and false-negatives) in terms of unnecessary testing and treatment, delay in discharge, and the occurrence of extreme hyperbilirubinemia and kernicterus. Given the low incidence of kernicterus, economic evaluation of risk-assessment strategies will have to rely on mathematical models in addition to empiric studies. The implementation of strategies that use universal bilirubin screening is associated with a reduced risk of extreme hyperbilirubinemia.²² Fine tuning the appropriate response to a predischarge bilirubin value will help ensure that universal bilirubin screening will also be cost-effective and minimally invasive.

	ACKNOWLEDGMENTS

 
Dr Keren was supported by grant K23 HD043179 from the National Institute of Child Health and Human Development (Bethesda, MD). This study was also supported by Respironics (Murrysville, PA), which loaned us a BiliCheck device for bilirubin measurements during home visits and donated a limited supply of BiliCheck disposable tips.

We thank research assistants Sasha Waring and Patrick Maloney, study nurse Tajuanna Harris, and all of the nurses and nurse practitioners in the Hospital of the University of Pennsylvania Well Infant Nursery for making this study possible.

	FOOTNOTES

 
Accepted Jun 24, 2007.

Address correspondence to Ron Keren, MD, MPH, Children's Hospital of Philadelphia, 3535 Market St, Room 1524, Philadelphia, PA 19104. E-mail: keren{at}email.chop.edu

The authors have indicated they have no financial relationships relevant to this article to disclose.

	REFERENCES

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 

1. American Academy of Pediatrics. Management of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation. Pediatrics. 2004;114 :297 –316[Abstract/Free Full Text]
2. Chou SC, Palmer RH, Ezhuthachan S, et al. Management of hyperbilirubinemia in newborns: measuring performance by using a benchmarking model. Pediatrics. 2003;112 :1264 –1273[Abstract/Free Full Text]
3. Keren R, Bhutani VK, Luan X, Nihtianova S, Cnaan A, Schwartz JS. Identifying newborns at risk of significant hyperbilirubinaemia: a comparison of two recommended approaches. Arch Dis Child. 2005;90 :415 –421[Abstract/Free Full Text]
4. Newman TB, Liljestrand P, Escobar GJ. Combining clinical risk factors with serum bilirubin levels to predict hyperbilirubinemia in newborns. Arch Pediatr Adolesc Med. 2005;159 :113 –119[Abstract/Free Full Text]
5. Stevenson DK, Fanaroff AA, Maisels MJ, et al. Prediction of hyperbilirubinemia in near-term and term infants. Pediatrics. 2001;108 :31 –39[Abstract/Free Full Text]
6. Newman TB, Xiong B, Gonzales VM, Escobar GJ. Prediction and prevention of extreme neonatal hyperbilirubinemia in a mature health maintenance organization. Arch Pediatr Adolesc Med. 2000;154 :1140 –1147[Abstract/Free Full Text]
7. Bhutani VK, Johnson L, Sivieri EM. Predictive ability of a predischarge hour-specific serum bilirubin for subsequent significant hyperbilirubinemia in healthy term and near-term newborns. Pediatrics. 1999;103 :6 –14[Abstract/Free Full Text]
8. Bhutani VK, Gourley GR, Adler S, Kreamer B, Dalin C, Johnson LH. Noninvasive measurement of total serum bilirubin in a multiracial predischarge newborn population to assess the risk of severe hyperbilirubinemia. Pediatrics. 2000;106(2) . Available at: www.pediatrics.org/cgi/content/full/106/2/e17
9. Rubaltelli FF, Gourley GR, Loskamp N, et al. Transcutaneous bilirubin measurement: a multicenter evaluation of a new device [see comment]. Pediatrics. 2001;107 :1264 –1271[Abstract/Free Full Text]
10. Kramer LI. Advancement of dermal icterus in the jaundiced newborn. Am J Dis Child. 1969;118 :454 –458[Medline]
11. Lin Z, Fontaine JM, Freer DE, Naylor EW. Alternative DNA-based newborn screening for glucose-6-phosphate dehydrogenase deficiency. Mol Genet Metab. 2005;86 :212 –219[CrossRef][ISI][Medline]
12. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15 :361 –387[CrossRef][ISI][Medline]
13. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143 :29 –36[Abstract/Free Full Text]
14. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44 :837 –845[CrossRef][ISI][Medline]
15. Sarici SU, Serdar MA, Korkmaz A, et al. Incidence, course, and prediction of hyperbilirubinemia in near-term and term newborns. Pediatrics. 2004;113 :775 –780[Abstract/Free Full Text]
16. Seidman DS, Ergaz Z, Paz I, et al. Predicting the risk of jaundice in full-term healthy newborns: a prospective population-based study. J Perinatol. 1999;19 :564 –567[CrossRef][Medline]
17. Moyer VA, Ahn C, Sneed S. Accuracy of clinical judgment in neonatal jaundice. Arch Pediatr Adolesc Med. 2000;154 :391 –394[Abstract/Free Full Text]
18. Tayaba R, Gribetz D, Gribetz I, Holzman IR. Noninvasive estimation of serum bilirubin. Pediatrics. 1998;102(3) . Available at: www.pediatrics.org/cgi/content/full/102/3/e28
19. Kaplan M, Hammerman C, Feldman R, Brisk R. Predischarge bilirubin screening in glucose-6-phosphate dehydrogenase-deficient neonates. Pediatrics. 2000;105 :533 –537[Abstract/Free Full Text]
20. Kaplan M, Herschel M, Hammerman C, Hoyer JD, Stevenson DK. Hyperbilirubinemia among African American, glucose-6-phosphate dehydrogenase-deficient neonates. Pediatrics. 2004;114(2) . Available at: www.pediatrics.org/cgi/content/full/114/2/e213
21. Kaplan M, Renbaum P, Levy-Lahad E, Hammerman C, Lahad A, Beutler E. Gilbert syndrome and glucose-6-phosphate dehydrogenase deficiency: a dose-dependent genetic interaction crucial to neonatal hyperbilirubinemia. Proc Natl Acad Sci USA. 1997;94 :12128 –12132[Abstract/Free Full Text]
22. Eggert LD, Wiedmeier SE, Wilson J, Christensen RD. The effect of instituting a prehospital-discharge newborn bilirubin screening program in an 18-hospital health system. Pediatrics. 2006;117(5) . Available at: www.pediatrics.org/cgi/content/full/117/5/e855

This article has been cited by other articles:

				Predicting Hyperbilirubinemia in Newborns Journal Watch (General), January 17, 2008; 2008(117): 4 - 4. [Full Text]

This Article Abstract Full Text (PDF) P3Rs: Submit a response Alert me when this article is cited Alert me when P3Rs are posted Alert me if a correction is posted Citation Map Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Keren, R. Articles by Bhutani, V. K. Search for Related Content PubMed PubMed Citation Articles by Keren, R. Articles by Bhutani, V. K. Related Collections Premature & Newborn

	Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2008 American Academy of Pediatrics. All rights reserved.