Exploring the relationship between breastfeeding and the incidence ... - BMC Public Health

Breastfeeding has not yet reached optimal prevalence in many countries, including Ireland [2, 5]. With a formula feeding rate of 43.1% in 2016, Ireland is a fertile population in which to study associated effects of infant feeding types on infant morbidity. The objective of this paper was to investigate the relationship between exclusive breastfeeding for at least 90 days and the incidence of infant illness while controlling for a broad range of potential confounders in an Irish cohort. We find that infants who were EBF for 90+ days were significantly less likely to be admitted to hospital, spent less nights in hospital, and were less likely to develop respiratory diseases including asthma, snuffles/common colds, chest infections, eczema, ear infections, wheezing and asthma, skin problems, vomiting, and colic indicating a protective effect of breastfeeding. Further outcomes such as current health of the infant at time of interview, feeding problems and sleeping problems were also statistically significantly negative, signalling the potential protective effect of EBF90days. However, these infants were also more likely to fail to gain weight and to develop nappy rash. It should be noted however that the implied differences with and without exclusive breast feeding for 90 plus days tend to be relatively small compared to the inherent variability of the outcome, less than 0.15 standard deviations in all cases (Fig. 1). The results of this research conforms to the international literature [24, 25]. But are striking in that they show that even in a high income country such as Ireland, breastfeeding is correlated with infant morbidity and health care utilisation.

Our estimates indicate that, given the population of Irish infants born in the year GUI was collected (2008) was 73,996 [26], there would have been 17,766 less nights spent in hospital if all infants were exclusively breastfed for at least 90 days versus had none of them been breastfed. Furthermore, there would have been 1644 fewer cases of respiratory illnesses, 7429 fewer cases of chest infection and 4320 fewer cases of ear infections. This data again points to the risks of early introduction of formula into the diet of a breastfed baby and supports exclusive breastfeeding for at least 90 days and termination of the practice of human milk substitute supplementation for reasons avoidable by proper planning, antenatal expression of colostrum [27, 28] and availability of donor milk from community milk banks [29]. These practices are envisaged in the recent Code of Marketing of Breast Milk Substitutes' policy adopted by the HSE [30] and WHO/UNICEF's 10 steps to successful breastfeeding [31].

The study reports a statistically significant increase in failure to grow among the cohort that were EBF for at least 90 days. In 2013, the Health Service Executive (HSE) in Ireland, introduced the use of new WHO growth charts, which were developed using natural infant weight gains in breastfeeding cohorts and thus showed slower weight gain trajectory in infants than in the previously used HSE growth charts developed using 'natural' weight gain data from formula fed cohorts. The use of the newer WHO growth charts is recommended in order to avoid mislabelling infants as underweight or failing to thrive [24] which led to unnecessary supplementation and cessation of breastfeeding. However, during the study period in 2008 Irish health care professionals were still using the old HSE growth charts which potentially led to erroneous beliefs among parents regarding failure to thrive/gain weight. Thus, parents of breastfed infants may have been more likely to report that their children exhibited 'Failure to gain weight or to grow'. Previous research analysing GUI data has shown a significant risk reduction for obesity development in BF infants, with this risk reduction being greater with increasing duration of breastfeeding [32].

Our analysis also showed that infants in the EBF90days group were more commonly reported to have persistent nappy rash (0.01 (CI: 0.00 to 0.02)) in comparison to the Non-BF cohort. A similar finding has also been observed in the 1997 Avon longitudinal dataset analysis of over 12,000 infants, which reported that breastfeeding is a risk factor for the development of nappy rash [33] in the first 4 weeks of life. This study also found frequent stooling, a phenomena associated with breastfeeding, to be an associated risk factor of diaper dermatitis. Despite this, studies show a protective effect of breastfeeding on nappy rash over the first year of life [34]. This finding in our study may relate to the presence of more frequent stooling, as is normal in exclusively breastfed infants and thus in the EBF90days cohort, who are similarly aged to those participants in the Avon study. This frequent stooling in exclusively fed infants is explained by the presence of Human Milk Oligosaccharides (HMOs), prebiotics for gastrointestinal microorganisms which, as indigestible by infants, leads to an osmotic laxative effect [35]. The whey dominant whey:casein ratio in breastmilk, in comparison to a casein dominant ratio in formula, also contributes to increased stooling in exclusively breastfed infants [36].

Another consideration is the culturally accepted use of commercial infant wipes to clean babies on napkin changing. There is evidence to show that frequent use of multi-ingredient baby wipes leads to increased incidence of napkin rash [37]. As the cohort of infants in the GUI is from 2008, participants would have used wipes containing multiple ingredients rather than those subsequently invented with only 2 ingredients which were shown to be associated with a lower incidence of napkin dermatitis [37]. Interestingly, a case control study, albeit with only 30 participants, also has shown statistically significant therapeutic impact of breastmilk on napkin rashes [38]. As napkin dermatitis results in significant difficulty for both parents and infants, this area would warrant further research considering that an effective free easily assessed therapeutic option, such as breastmilk, could help significantly reduce this suffering.

The findings from this study conform with the international data showing that the incidence of infectious illnesses increases with increasing exposure to human milk substitutes and that exclusive human milk feeding is most protective in terms of infectious illnesses [4, 5, 8, 11, 25]. Our results indicate that the protective effect against developing at least 1 of 4 infectious diseases examined, namely, chest infection, ear infection, common colds and meningitis; in a cohort of 100 infants EBF for 90, we estimate a reduction of 9.3 infants [CI 11.7 to 6.9] presenting with one of these infective illnesses when EBF for at least 90 days compared to a similar formula fed [Non-BF] cohort. Similar point estimates were found in the EBF180 vs NBF analysis, indicating beneficial effect of exclusive breastfeeding to 180 days as recommended by the WHO, albeit there was lower power to detect effects due to the reduced sample sizes for that analysis.

Furthermore, our study shows greater protection may be provided by EBF over non-EBF and Non-BF infants (see appendix Table A6) which leads us to consider that it is the risk of feeding human milk substitutes, rather than on the benefits of breastfeeding per se, that leads to differences in morbidity and health care utilisation, between the EBF, BF and Non-BF groups. This is not a new idea [39] but does emphasise the newer narrative that breastfeeding is the physiological normal, one which has driven mammalian evolution for millions of years [40] and that replacing human milk feeding with substitute human milks may infer risk. Our findings indicate an increased associated risk of morbidity with the early introduction of and substitution with human milk substitutes, with the risks greater among formula fed [Non-BF] infants.

There are some limitations to the present study. Most notably, GUI relies on recollection of whether a mother breastfed or not. However, there is evidence that maternal recollection of breastfeeding status tends to be a valid representation of breastfeeding status.⁴² We rely on self-reports of whether an infant was taken to the GP, Health Centre or Public Health Nurse or to Accident and Emergency for each of the conditions, which may be another source of misreporting bias. Linking the GUI data to administrative records was not possible in this study. Since GUI used a nationally representative sample, it is plausible that results would generalize to the population, although as we note in the appendix, results may be sensitive to the presence of unobserved confounders despite the rich set of covariates for which we controlled. Future work will explore the relationship between breastfeeding and the infants' health in later waves. We could also assess the potential cost savings attributable to optimised breastfeeding in Ireland. Finally, we could explore whether findings are similar in other observational child cohort studies such as Growing up in Scotland, Growing up in New Zealand and Growing up in Australia.