Should chronic hepatitis B mothers breastfeed? a meta analysis

The hepatitis B virus (HBV) accounts for a significant morbidity and mortality worldwide. An estimated one-third of the world population has been exposed to HBV and 400 million people are chronic carriers [1]. Mother to infant transmission of HBV would occur among up to 90% of infants born of chronic hepatitis B (CHB) mothers who are positive for the infectivity markers, the HBeAg and/or the HBV DNA [2]. With the introduction of a safe and effective hepatitis B vaccine, infantile CHB infection has dramatically dropped to around 5% [3].

The advantages of breastfeeding over formula feeding have been well documented [4]. However, the presence of the HBV in breast milk firstly reported by Linnemann [5] in 1974 and confirmed by later studies [6, 7] raised the possibility of transmission of HBV through breastfeeding. But later studies found no association between breastfeeding and infantile CHB infection, both before and after the introduction of hepatitis B vaccine [8‐13]. In the absence of evidence that breastfeeding poses any additional risk of infection to infants born of CHB mothers, World Health Organization (WHO) recommends breastfeeding even for area where HBV infection is highly endemic and HBV vaccine is not available [14].

Scepticism about this recommendation remains among clinicians, however. About 25% of obstetricians in Australia [15] and 50% of physicians (most of them were hepatologists!) in Illinois [16] did not recommend CHB mothers to breastfeed their babies. It was noted that 5.4% of pregnant women in Shanghai were HBsAg positive and 30.3% of whom were HBeAg-positive (Tao, personal communication), which was reported to pose a significant risk of HBV transmission to infants, including those who had been vaccinated against the virus [17]. In the light of this and other studies [10], it is a common and recommended practice that CHB mothers with an abnormal alanine aminotransferase (ALT) level before delivery do not breastfeed.

In the present paper, we conduct a meta analysis to assess the risk of CHB infection of vaccinated infants through breastfeeding associating with CHB mothers who are and are not positive for the HBV infectivity markers.

Our results suggest that breastfeeding by CHB mothers does not pose a significant risk of infection by the virus, provided that the infants have been vaccinated against HBV at birth.

In this meta analysis, the overall transmission rate of HBV from CHB mothers to their infants who had completed routine hepatitis B vaccination was 4.32%, which was similar to those previously reported [3]. Mother-to-infant transmission of HBV can occur prenatally, at birth and postnatally [49], but most commonly infection occurs at the time of delivery. In-utero transmission is very uncommon, except in China and Japan, and often persists after birth, which results in the ineffectiveness of HBV vaccination, i.e., immunoprophylaxis [49, 50]. From this perspective, most mother-to-infant transmission occurs before deciding on infant feeding methods. But there is an exception: the neonate can be infected postnatally by household contact with CHB family members [51]. Apparently, breastfeeding would produce more frequent contacts between mother and infant than formula-feeding. Though separation of breastfed transmission of CHB from contact seems to be important in the clarification of the net effect of breastfeeding on infantile CHB infection, this contact transmission occurs only at the successful exchange of body fluid from infected mothers to their infants under some special situations, such as excoriation. Thus breastfed transmission of HBV, if existed, should be minimal.

Two previous meta analyses reported different associations of breastfeeding and infantile CHB infection with an odds ratio (OR) = 1.68 (95%CI: 1.16, 2.44) [52] and 0.67 (95%CI: 0.41, 1.12) [53] respectively based on 13 and 10 studies. The inconsistent results reflected difference between the studies with respect to their study aim, search strategies, criteria of inclusion and exclusion of the studies, and analysis strategies. One study included both vaccinated and not vaccinated infants [52] and the other included only those who had been given both passive and active immunization [53]. After careful evaluation on these two analyses, we identified their obvious shortcomings: (1) many eligible studies were not included in these analyses without any reason; (2) the analyses excluded those studies with zero outcome of infantile CHB infection in both groups of feeding methods; (3) neither of the studies evaluated the quality of the included studies and their influence on the results, and (4) neither of the studies considered the possible confounding of mothers' infectivity, which was important in mother-to-infant transmission of HBV and their decisions on how to feed their infants [17, 54]. In our meta analysis, reference search, data extraction and quality evaluation were done by two independent investigators and examined by the third one before a consensus was reached, thus the completeness and appropriateness of the studies included for analysis were ensured. We used RD instead of OR or risk ratio (RR) as effect measure, which ensured that a reasonable estimation of effect measurement remained untouched and the 2^nd shortcoming mentioned above was overcome simultaneously.

Our findings have important implications for current clinical practice about breastfeeding by CHB mothers, even with high infectivity. Routine hepatitis B vaccination allows the infant to enjoy breastfeeding, the nutritional, immunological and psychological advantages of which are well known[4]. And breastfeeding would also greatly benefit the mother[4], especially in weakening or eliminating her fear and guilt which occur often. Unfortunately, contrary to our findings that risk of breastfed transmission of infantile CHB infection simply does not exist, the current clinical practices seem discouraging: (1) any breastfeeding rate is about 30% lower for those mothers with than without CHB [54], and even lower (5.4% only) for those with high infectivity [55]; (2) there is still 25%-50% of the medical professionals did not recommend infantile breastfeeding by CHB mothers [15, 16]. Thus, health education of breastfeeding with correct knowledge, attitude and behaviour on the decision makers, such as medical professional and CHB mothers, seems to be urgent, especially in those epidemic countries, like China.

This meta analysis demonstrates no association between breastfeeding and risk of infantile CHB infection from a CHB mother to her infant. Two alternative explanations must be considered. First, a systematic bias may be present in follow-up studies arising from differential lost to follow-up, which has not been mentioned in most of the studies. Second, the studied population varies from study to study, and some unknown confounders may not be effectively evaluated. Further evaluation of the effect of breastfeeding from an HBV infected mother and infantile CHB infection while taking into consideration of potential biases can be feasibly achieved with more large-scale prospective follow-up studies.

Breast milk is infectious; yet, breastfeeding, even by mothers with high infectivity, is not associated with demonstrable risk of infection of infantile CHB. This may be partly because infection mainly occurs during delivery and the protection by vaccination. The finding concurs with that of individual studies and supports the WHO recommendation of BF, irrespective of the HBV status of the mothers.