Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
February 1999

Neonatal Jaundice and Diet

Author Affiliations

From the Departments of Pediatrics (Dr Gourley and Mr Kreamer), Statistics (Dr Kosorok), and Biostatistics and Medical Informatics (Dr Kosorok), University of Wisconsin School of Medicine and Waisman Center on Mental Retardation and Human Development, Madison, and Justus-Liebig Universität, Giessen, Germany (Ms Cohnen).


Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999

Arch Pediatr Adolesc Med. 1999;153(2):184-188. doi:10.1001/archpedi.153.2.184

Objective  To determine whether an earlier observation, that infants fed a casein-hydrolysate formula (Nutramigen) have lower neonatal jaundice levels than those fed standard formulas, would be repeated in a larger independent group of infants with more frequent measurements and more rigorous statistical analysis.

Design  Newborn infants were fed human milk, a standard whey-predominant formula (Enfamil), or Nutramigen (n=20 for each group) during the first 3 weeks of life. Transcutaneous jaundice index was measured daily for the first week of life and every 2 to 3 days thereafter, using a noninvasive jaundice meter. Linear regression models of the data were constructed, validated, and compared statistically.

Setting  General community hospital with subsequent home visitation.

Participants  Healthy, term newborn infants selected by convenience, based on time of birth.

Intervention  Infants were exclusively fed human milk, Enfamil, or Nutramigen. Formulas were randomly assigned.

Main Outcome Measure  Jaundice index, a transcutaneous measurement of jaundice.

Results  The jaundice index differed significantly among the 3 groups. Paired comparisons showed that the jaundice index of the Nutramigen group was significantly lower than that of the Enfamil group (on days 6-16) and the human milk group (on days 3-20). The jaundice index of the Enfamil-fed group was significantly lower than that of the human milk group on days 13 to 19.

Conclusions  Jaundice levels are lower in neonates fed Nutramigen rather than Enfamil and both these groups have lower jaundice levels than breast-fed infants.

HYPERBILIRUBINEMIA in the first days of life is observed more frequently in breast-fed infants than in formula-fed infants.1,2 Although the pathogenesis of this finding is poorly understood, many hypotheses exist.3 There are few data comparing jaundice levels among infants whose diet consists solely of infant formula. One study showed that serum bilirubin concentrations and the transcutaneous measurement of jaundice index were significantly different among infants who were exclusively fed infant formulas.4 These results showed that infants consuming a casein-hydrolysate formula, Nutramigen, had lower serum bilirubin levels and a lower jaundice index than infants consuming standard formulas.

Pediatricians support breast-feeding as the best choice for neonatal nutrition. Nonetheless, "temporary interruption of breast-feeding and substitution with formula" is advocated by the American Academy of Pediatrics as one therapeutic option for jaundice associated with breast-feeding.5 If breast-fed infants are going to be fed formula to decrease their jaundice, it would be useful to know if a particular formula is associated with lower jaundice levels. This knowledge could potentially allow infants to get back to breast-feeding faster. The purpose of the present study was to independently assess an earlier finding4 that neonatal jaundice differs among infants who are exclusively fed various infant formulas. In the present study we assessed jaundice index in a larger group of infants, with more frequent measurements and a more rigorous statistical analysis. These infants were exclusively fed breast milk or 1 of 2 commercial infant formulas.


Subjects were healthy, full-term, white, vaginally delivered neonates. Their parents were contacted for recruitment before birth (via newspaper advertisements) or within several hours of birth (before the passage of the first stool). Recruitment was attempted if infants meeting the above criteria were born between the approximate hours of 9 AM and 3 PM. No more than 3 infants were recruited into this study per week. All 3 feeding groups were recruited simultaneously. This research was approved by each subject's personal physician, the University of Wisconsin Human Subjects Committee, and the Meriter-Madison (Wis) General Hospital Institutional Review Board. Infants (n=20 in each group who successfully completed the study) were completely fed ad libitum with human breast milk or 1 of 2 infant formulas: a standard whey-predominant formula (Enfamil) or a casein-hydrolysate formula (Nutramigen) (both from Mead Johnson, Evansville, Ind). Breast-feeding was initiated within 30 to 60 minutes after birth and was subsequently offered on demand every 1 to 3 hours (10-20 minutes on each breast) with no supplementation of water or formula at any time. Formula feeding was initiated within 30 to 60 minutes after birth and was subsequently offered on demand every 1 to 3 hours. Enfamil or Nutramigen was randomly assigned to the formula-fed infants on the basis of a random assignment chart that was constructed prior to study initiation. Formula-fed infants received no other supplementation. Any deviation from the initial feeding plan or occurrence of serious illness prompted exclusion from the study. Phototherapy invalidates transcutaneous jaundice index measurements, so phototherapy also prompted exclusion from the study. Fourteen infants were initially enrolled but subsequently dropped out of the study: 7 breast-fed (4 required formula supplementation, 1 received phototherapy, and 2 missed the first stool collection), 2 Enfamil-fed (1 received phototherapy and 1 had diarrhea), and 5 Nutramigen-fed (3 due to maternal preference, 1 received formula supplementation, 1 had Hirschsprung disease). If a subject was dropped out of the study, none of the data from that subject was used for this study and this dropout subject was replaced with a subsequent subject who successfully completed the whole study. Infants were recruited between March 13, 1992, and October 13, 1994.

Neonatal jaundice was assessed daily by the determination of the transcutaneous "jaundice index" during the first 7 days of life and every 2 to 3 days during the residual course of the 3-week study. This was done by using a jaundice meter (Jaundice Meter 101; Minolta/Narco Scientific, Hatboro, Pa). This instrument was checked daily with "0" (white) and "20" (yellow) reading checkers, and was found to be within calibration throughout the study (00 or 01 refer to white checker; 20±1, yellow checker). Jaundice index has been shown to be highly correlated with serum bilirubin levels in many reports,6 including a previous study4 that showed a correlation of r=0.94 in 106 paired jaundice meter and neonatal serum bilirubin measurements in a local population similar to the one studied here.


To compare subjects at exactly the same time points, daily values for jaundice index were determined by constructing models of the data for each dietary group. These models were constructed based on data for which the exact date and time of each jaundice measurement was recorded to the nearest minute of each day. Each model was a polynomial regression fit (using up to 5 df) chosen to minimize the Akaike Information Criterion.7 Estimates of jaundice index pooled all data within dietary groups. However, to estimate the variance and P values, the repeated measurements of individual subjects were taken into account by using the robust variance estimator for repeated measurements (used in generalized estimating equations with the independence working correlation matrix).8 Models were validated by examining predicted vs residual values and in all cases no significant departures from the homogeneity of variance assumption were found.

P values for overall fit were obtained by using the results from the above analysis for each of the 3 dietary groups. We used the χ2 test of overall differences among the 3 groups as well as the paired comparisons between specific groups. Paired comparisons between groups were only made when significant overall χ2 test results were obtained. Paired comparisons were made every 24 hours from birth to 20 days of age. Since there were 3 comparisons at each time point, significance was only concluded if P<.017 (0.05/3) (Bonferroni adjustment).9 Adjusting for multiple comparisons across time was not necessary because of the initial overall model comparisons. All analyses were accomplished with S-PLUS statistical software (Statistical Sciences Inc, Seattle, Wash). Demographic differences among the 3 dietary groups were evaluated with a rank sum test for more than 2 groups (Kruskal-Wallis test),10 and a χ2 test.


Demographic characteristics of the neonates are shown in Table 1. There were no significant differences among the 3 dietary groups in sex, gestational age, parity, birth weight, incidence of vacuum extraction delivery, or previous siblings with jaundice. The maternal age of the breast-fed group was significantly greater than either formula-fed groups.

Table 1. 
Image not available
Demographic Information*

Figure 1 presents jaundice index data. All the raw data from each of the 3 dietary groups are presented along with an overlay of the model for each group. Figure 2 presents the 3 models in 1 graph, thus allowing for easier comparison. The statistics comparing the 3 models are presented in Table 2. The simultaneous overall comparison of the 3 groups indicates that the models generated are significantly different (P=1.2×10-15). Thus, we proceeded to make paired overall comparisons to determine which groups differed from which other groups. As indicated in the top line of Table 2, each of the 3 groups was significantly different from each of the other groups (overall comparison of days 0-20). To determine the specific days during which specific groups differed from each other, we made daily comparisons of the model means for all of the group pairs. These paired comparisons (Table 2) show that the mean model jaundice index was significantly higher (P<.017) in the human milk group than in the Enfamil group on days 13 to 19, in the Enfamil group than in the Nutramigen group on days 6 to 16, and in the human milk group than in the Nutramigen group on days 3 to 20.

Figure 1.
Image not available

Transcutaneous jaundice index in newborn infants fed human milk (top), standard whey-predominant formula (Enfamil) (middle), or casein-hydrolysate formula (Nutramigen) (bottom). These figures display all of the raw data in each of the 3 dietary groups. The model generated for each group is displayed as a solid line. Both Enfamil and Nutramigen are manufactured by Mead Johnson, Evansville, Ind.

Figure 2.
Image not available

Transcutaneous jaundice index in newborn infants fed human milk, standard whey-predominant formula (Enfamil), or casein-hydrolysate formula (Nutramigen). This figure displays the model generated for each of the 3 dietary groups. Both Enfamil and Nutramigen are manufactured by Mead Johnson, Evansville, Ind.

Table 2. 
Image not available
Jaundice Index Comparisons in Newborns Fed Human Milk (H), Enfamil (E), or Nutramigen (N)*

Results of this study independently confirm the findings of the prior study.4 All formulas are not equal in their effect on neonatal jaundice. While it is well established that formula-fed infants generally have lower jaundice levels than breast-fed infants,1,2,1113 we are unaware of other data comparing jaundice levels among infants fed various formula diets. This and the previous study4 produced very similar findings. In general, infants fed a casein-hydrolysate formula (Nutramigen) had significantly lower jaundice levels than infants fed a routine whey-predominant formula (Enfamil) during the first 3 weeks of life. It is of note that jaundice levels in the Enfamil-fed group were not significantly lower than those of the breast-fed group during the first week of life, as is usually reported. We suggest that this is because the breast-fed infants having the highest serum bilirubin levels received therapy (formula supplementation or phototherapy) and thus were excluded from the study. This biased the breast-fed group toward lower jaundice levels.

ARE THERE ANY implications to this finding? It is too early to be definitive. One consideration relates to the time when breast-fed infants are developing jaundice severe enough to consider hyperbilirubinemia therapy. The American Academy of Pediatrics has advocated temporary interruption of breast-feeding and substitution with formula as one therapy for jaundice associated with breast-feeding.5 When this situation arises, it is conceivable that an infant might receive more benefit from changing from a diet of human milk to Nutramigen rather than to Enfamil. In some infants, a simple formula change for a short time is all that is necessary. In these infants, the more expensive casein-hydrolysate might not be cost-effective. However, in other infants, a change to formula is just the step that precedes other therapy such as phototherapy or exchange transfusion. In these infants, it may be possible that change to the casein-hydrolysate might prevent or lessen the additional therapy. This could only be determined through controlled clinical trials. Hence, use of Nutramigen for this purpose is not recommended at this time. Alternatively, if this effect on jaundice is due to a specific component of casein hydrolysate, perhaps it might be possible to isolate and identify this component so that it could be given separately and thus sometimes avoid interruption of breast-feeding.

What ingredient in hydrolyzed casein could be responsible for causing this difference in jaundice levels between the 2 different formulas? At this time we are unable to do more than speculate. One possible reason for the elevated bilirubin levels in breast-fed infants is β-glucuronidase.14,15 In the fetus, intestinal β-glucuronidase is believed to play an important role in intestinal bilirubin absorption that allows fetal bilirubin to be cleared via the placenta.16 β-Glucuronidase cleaves glucuronic acid from bilirubin conjugates. The unconjugated bilirubin produced can be absorbed from the intestine more easily than the bilirubin glucuronides.17 However, following birth, an increased level of intestinal β-glucuronidase has long been suggested to potentiate the enterohepatic circulation of bilirubin, thus leading to neonatal jaundice.18 Previous studies had indicated that breast milk contains considerable amounts of β-glucuronidase, while the β-glucuronidase level in a standard infant formula is negligible.14,19 Nutramigen, a casein-hydrolysate formula, has been shown to contain an inhibitor of β-glucuronidase.20 High levels of β-glucuronidase in breast milk, feces, or serum have been shown to be associated with neonatal jaundice in some,14,15,19,21 but not all studies.22,23 Recently, we have shown that there is a significant β-glucuronidase inhibitor present in hydrolyzed casein and Nutramigen.20,24 This β-glucuronidase inhibitor remains active after passing through the neonatal intestinal tract. If β-glucuronidase is being inhibited in vivo, then there would be less hydrolysis of bilirubin glucuronides. Thus, less unconjugated bilirubin would be available for intestinal absorption. Since unconjugated bilirubin is absorbed more easily than bilirubin glucuronides,17 the expected result of β-glucuronidase inhibition would be to decrease the intestinal absorption of bilirubin and thus improve bilirubin excretion in the stool. This would fit with the observed order of jaundice in this study: breast milk (rich with β-glucuronidase14,15,21,25) is associated with highest jaundice levels, Enfamil (containing negligible β-glucuronidase14) is associated with intermediate jaundice levels, and Nutramigen (containing a β-glucuronidase inhibitor20,24) is associated with the lowest jaundice levels.

Alternatively, perhaps unhydrolyzed cow milk formula and human milk, to a greater degree, contain components other than β-glucuronidase that promote hyperbilirubinemia, eg, through enhancement of intestinal absorption of bilirubin. Hydrolysis may destroy this component and result in less jaundice with Nutramigen. Since enterohepatic circulation of various intestinal components, in addition to bilirubin, is a potentially significant biological event, there are implications that are theoretically harmful in providing a feeding that does not contain the factor promoting enterohepatic circulation of such components as thyroxine and other hormones. These considerations need investigation before making recommendations based on bilirubin data only.

Back to top
Article Information

Accepted for publication August 4, 1998.

Supported in part by grants HD28619 and HD29287 from the National Institutes of Health, Bethesda, Md; grant MO1 RR03186 from the General Clinical Research Center, University of Wisconsin–Madison; grant HD03352 from the Waisman Center, Madison; and Mead Johnson, Evansville, Ind. Minolta-Air Shields, Hatboro, Pa, provided some of the jaundice meters.

Presented at the American Pediatric Society–Society for Pediatric Research Annual Meeting, Washington, DC, May 2, 1997.26

The authors gratefully acknowledge the technical assistance of Carol Cowgill, RN, Nithya Hariharan, MD, Brenda Egan, Judy McDonald, RN, Deb Hamele, RN, Bridget Pribbenow, Kristin Gorton, Ruth Merrill, Sherrie Lama, Marla Mussotter, Sarah Thorsett, Adolfo Garcia, Kristin Schroederus, Kim Merkl, Lan Zeng, and the physicians and nurses of Meriter Hospital, Madison.

Reprints: Glenn R. Gourley, MD, University of Wisconsin Waisman Center, Room 609, 1500 Highland Ave, Madison, WI 53706 (e-mail:

Editor's Note: In this case, casein is in and whey is out, way out.—Catherine D. DeAngelis, MD

Osborn  LMReiff  MIBolus  R Jaundice in the full-term neonate. Pediatrics. 1984;73520- 525
Schneider  AP Breast milk jaundice in the newborn: a real entity. JAMA. 1986;2553270- 3274Article
Gourley  GR The pathophysiology of breast milk jaundice. Polin  RAFox  WWeds.Fetal and Neonatal Physiology. Philadelphia, Pa WB Saunders1992;1173- 1179
Gourley  GRKreamer  BArend  R Effect of diet on fecal excretion and jaundice during the first three weeks of life. Gastroenterology. 1992;103660- 667
American Academy of Pediatrics, Provisional Committee for Quality Improvement and Subcommittee on Hyperbilirubinemia. Practice parameter: management of hyperbilirubinemia in the healthy term newborn. Pediatrics. 1994;94558- 565
Schumacher  RE Noninvasive measurements of bilirubin in the newborn. Clin Perinatol. 1990;17417- 435
Akaike  H Information theory and an extension of the maximum likelihood principle. Petrov  BNCsaki  Feds.Second International Symposium on Information Theory. Budapest, Hungary Academia Kiado1973;267- 281
Liang  KYZeger  SL Longitudinal data analysis using generalized linear models. Biometrika. 1986;7313- 22Article
Milliken  GAJohnson  DE Analysis of Messy Data.  New York, NY Van Nostrand Reinhold1984;31- 33
Hollander  MWolfe  D Nonparametric Statistical Methods.  New York, NY John Wiley & Sons1973;114- 120
Saigal  SLunyk  OBennett  KJPatterson  MC Serum bilirubin levels in breast- and formula-fed infants in the first 5 days of life. CMAJ. 1982;127985- 989
Maisels  MJGifford  K Normal serum bilirubin levels in the newborn and the effect of breast-feeding. Pediatrics. 1986;78837- 843
Narayanan  IGupta  AMandal  RNChugh  RKSingh  S Infant feeding and early neonatal jaundice. Indian J Pediatr. 1987;54257- 260Article
Gourley  GRArend  RA β-Glucuronidase and hyperbilirubinemia in breast-fed and formula-fed babies. Lancet. 1986;1644- 646Article
Gaffney  PTButtenshaw  RLWard  MDiplock  RD Breast milk β-glucuronidase and neonatal jaundice. Lancet. 1986;11161- 1162Article
Odell  GB Bilirubin metabolism during fetal life. Neonatal Hyperbilirubinemia. New York, NY Grune & Stratton1980;27- 34
Lester  RSchmid  R Intestinal absorption of bile pigments, I: the enterohepatic circulation of bilirubin in the cat. J Clin Invest. 1963;42736- 746Article
Brodersen  RHermann  LS Intestinal reabsorption of unconjugated bilirubin: a possible contributing factor in neonatal jaundice. Lancet. 1963;11242Article
el-Kholy  MSHalim  HYMarzouk  AH Beta-glucuronidase and hyperbilirubinemia in breast-fed versus formula-fed babies. J Egypt Public Health Assoc. 1992;67237- 248
Gourley  GRKreamer  BLCohnen  M Inhibition of beta-glucuronidase by casein hydrolysate formula. J Pediatr Gastroenterol. 1997;25267- 272Article
Sirota  LFerrera  MLerer  NDulitzky  F Beta-glucuronidase and hyperbilirubinaemia in breast-fed infants of diabetic mothers. Arch Dis Child. 1992;67120- 121Article
Yau  KIChien  CHFong  LLChen  CL Breast milk beta-glucuronidase in breast milk jaundice. J Formos Med Assoc. 1992;91287- 293
Wilson  DCAfrasiabi  MReid  MM Breast milk beta-glucuronidase and exaggerated jaundice in the early neonatal period. Biol Neonate. 1992;61232- 234Article
Gourley  GRKreamer  BLCohnen  M Inhibition of beta-glucuronidase by casein hydrolysate formula [abstract]. Pediatr Res. 1997;4183A
Thomas  NCShirey  RSBlakemore  KKickler  TS A quantitative assay for subclassing IgG alloantibodies implicated in hemolytic disease of the newborn. Vox Sang. 1995;69120- 125Article
Gourley  GRKreamer  B Neonatal jaundice levels are lower in infants fed Nutramigen rather than Enfamil [abstract]. Pediatr Res. 1997;4183A