Sixty to 80% of healthy, full-term infants are expected to present with idiopathic neonatal jaundice in the initial postnatal period [1]. Idiopathic neonatal jaundice is attributed to an increased breakdown of heme, immature liver function, low amount of intestinal bacteria, increased enterohepatic circulation of bilirubin and inadequate intake [2]. Indicators of inadequate intake include 4 to 6 thoroughly wet diapers in 24 hours, and the passage of 3 to 4 stools per day by the fourth day. In addition, the stools of adequately breastfed infants should change from meconium to a mustard yellow, mushy stool by the third to fourth day. These assessments help to identify breastfed infants who are at risk of dehydration because of inadequate intake, however it is relatively subjective due to individual differences [3]. Compared to other methods, body weight loss (BWL) percentage is an objective and useful tool that may indicate when interventions such as supplemental feeding should be considered.

Previous studies have suggested that 7% to 10% BWL by day 3 in fully breastfed infants is abnormal neonatal BWL [4, 5]. However, there are conflicting opinions about what constitutes normal neonatal BWL, and about when supplemental feeding should be considered to prevent significant hyperbilirubinemia. The aim of this study was to analyze the optimum cutoff values of BWL percentages in the first three days after birth to predict neonatal hyperbilirubinemia 72 hours after birth.

During the study period, 343 neonates with a gestational age of more than 37 weeks and a birth BW above 2500 g were analyzed. The mean birth BW was 3119 ± 352 g, and the mean BWL percentage on day 3 was 7.07 ± 2.82%. The mean TSB level 72 hours after birth was 13.39 ± 3.12 mg/dL. In addition, 115 neonates (33.5%) had levels of TSB higher than 15 mg/dL 72 hours after birth.

Table 3 lists the upper and lower limits of the BWL percentage within the first 3 days in predicting significant hyperbilirubinemia 72 hours after birth by ROC analysis and the ROC curves are shown in Figure 1(A, B,C). The BWL percentage on day 3 had an AUC of more than 0.7. The neonates with BWL percentages less than 0.4% on day 1, less than 2.9% on day 2, and less than 0.6% on day 3 did not get significant hyperbilirubinemia 72 hours after birth. The infants with BWL percentages more than 10.2% on day 1, more than 10.9% by day 2, and more than 11.3% by day 3 had significant hyperbilirubinemia 72 hours after birth. In addition, a BWL percentage of 7.60% by day 2, and 8.15% by day 3 appeared to be the optimum cutoff points in predicting hyperbilirubinemia 72 hours after birth (Table 4). The correlation between body weight loss percentage within the first 3 days and total bilirubin levels at 72 hours after birth was performed by Pearson Correlation Coefficient and shown in Figure 2.

Day	BWL percentage	Sensitivity (%)	Specificity (%)	LR+	LR-	AUC	95% CI	P-value
Day 1	>0.431%	100.00	4.39	1.05	0.00	0.581	0.527-0.634	0.014
	>10.19%	0.00	100.00	-	1.00
Day 2	>2.97%	100.00	7.89	1.09	0.00	0.632	0.579-0.684	<0.001
	>10.93%	0.87	100.00	-	0.99
Day 3	>0.60%	100.00	2.63	1.03	0.00	0.703	0.652-0.751	<0.001
	>11.33%	10.43	100.00	-	0.90

BWL: bodyweight loss; LR: likelihood ratio; AUC: area under curve; CI: confidence interval.

Breastfeeding offers many benefits for neonates including host protection, better developmental outcomes, and decreased rates of infectious diseases, infant death syndrome, overweight and even asthma [8–12]. However, inadequate breastfeeding may cause excessive weight loss during the first postpartum days and may be associated with hyperbilirubinemia [13, 14]. Breastfeeding has been thought to be a risk factor for significant hyperbilirubinemia since the 1960s [15–17], and breastfed infants have been shown to experience a maximum BWL by day 3 with a BWL of 6.6% compared to 3.5% for formula fed neonates [18]. This finding is believed to be attributable to the different components of mixed feeding, and may be because infants who receive mixed feeding have a poorer intake than those who are breastfed. In the current study, the healthy and term neonates who presented with significant hyperbilirubinemia 72 hours after birth had a BWL percentage of 8.4 ± 2.54% by day 3 compared to 6.4 ± 2.73% in the non-significant hyperbilirubinemia group. This finding is similar to previous studies in which 33% of 86 neonates with hyperbilirubinemia suffered from severe BWL of more than 10% [19], and 25% of 874 newborns with hyperbilirubinemia suffered a BWL of 8.96 ± 1.99% [20].

Even though many reports support the idea that breastfeeding induces severe hyperbilirubinemia, we found that breastfeeding was not statistically correlated with significant hyperbilirubinemia, thereby supporting the use of breast milk. A previous study reported that a BWL of more than 7% was an independent risk factor for early neonatal jaundice, and that infants with a BWL of more than 7% had a 1.4-fold increased risk of jaundice [21]. In the current study, the BWL percentages in the first three days after birth were all statistically significantly correlated with significant hyperbilirubinemia 72 hours after birth. In particular, the BWL percentage on day 3 was a better predictive factor for significant hyperbilirubinemia 72 hours after birth compared to day 1 and day 2. Moreover, regardless of being exclusively breastfed or mixed fed, the BWL within the first three days seemed to be influenced by the BWL on the previous day. This not only emphasizes the importance of BWL percentage on day 2 and day 3, but also suggests that the infants with a high BWL percentage on day 1 can still avoid further significant hyperbilirubinemia with optimal intervention. In this study, the order of birth (first born vs. second or third) was not analyzed since although the experience with breast feeding may influence the timing of giving extra formula milk and even the percentage of weight loss, the BWL was still the most objective factor regardless of how many babies the mother had.

We determined two cutoff points for each BWL percentage from day 1 to day 3 in order to be more easily applicable in a clinical setting, and to provide a point as to when significant hyperbilirubinemia 72 hours after birth can be ruled in or out. Based on the results of ROC analysis, the neonates with a BWL percentage on day 1 of less than 0.43% did not progress to significant hyperbilirubinemia 72 hours after birth, while a BWL percentage of more than 10.23% on day 1 led to significant hyperbilirubinemia 72 hours after birth. In addition, the neonates with BWL percentages less than 2.97% on day 2 and less than 0.60% on day 3 did not progress to significant hyperbilirubinemia 72 hours after birth, while those with BWL percentages of more than 10.93% on day 2 and 11.33% on day 3 did progress to significant hyperbilirubinemia 72 hours after birth. Primary care pediatricians should pay more attention to the patients in the indeterminate zone because it is not easy to predict significant hyperbilirubinemia 72 hours after birth in such cases. In addition, the physicians should attempt to persuade family members to increase the oral intake for these neonates to prevent further dehydration-induced significant hyperbilirubinemia 72 hours after birth. Based on our results, BWL percentage cutoff values of 4.48% on day 1, 7.60% by day 2, and 8.15% by day 3 were useful in predicting significant hyperbilirubinemia 72 hours after birth. We believe that this may be a helpful clinical tool when evaluating neonates with physical dehydration on the first three days after birth.

For non-pathological hyperbilirubinemia infants, dehydration is thought to be the major factor [1–5]. However, there seemed to be low accuracy when measured by the AUC curve with BWL at 24, 48, and 72 hours. In addition, the positive and negative likelihood ratios were very close to 1, suggesting that they had low practical significance. The cutoff values we propose are based on higher specificity and higher positive likelihood ratio, as the purpose was to prevent further significant hyperbilirubinemia. Accordingly, although BWL may not be an ideal predictor for significant hyperbilirubinemia, the results may be useful for physicians and parents as an indication of when extra feeding should be added.

There are some limitations to the present study. First, we used two different methods of measuring total bilirubin. Although the accuracy of a transcutaneous device has been proven, there may have been some deviation between the two methods. Second, maternal and neonatal risk factors for hyperbilirubinemia were not addressed (time of cord clamping, diabetic mothers, information on the order of birth in each group, previous siblings with significant jaundice and preterm infants). Third, the difficulty in assessing the adequacy of intake in breastfeeding and the variable timing of shifting to mixed feeding made the line between two feeding groups more blurred. Fourth, the major limitation may have been that we didn’t enroll more patients in the study. Finally, the study was performed in Taiwan and the cutoff values may not necessarily be applicable to other regions of the world.

In conclusion, BWL on the first three days after birth may be a predisposing factor for neonatal hyperbilirubinemia, and may also serve as a helpful factor to prevent significant hyperbilirubinemia 72 hours after birth. The optimal BWL cutoff percentages on the first three days after birth presented in this study may predict hyperbilirubinemia and indicate the need for supplementary feeding.