Exclusive bottle feeding of either formula or breast milk is associated with greater infant weight gain than exclusive breastfeeding, but findings may not reflect a causal effect of bottle feeding
- Department of Pediatrics and of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Quebec, Canada
- Correspondence to Dr Michael S Kramer
Department of Pediatrics and of Epidemiology, Biostatistics, and Occupational Health, McGill University, 2300 Tupper Street, Montreal, Quebec, Canada H3H 1P3;
Commentary on: 
The paper by Li et al addresses a topic that has been much studied and debated over the last several decades: the relationship between type of infant feeding and growth in the first year of life. This study adds a new twist, however: the consideration of bottle feeding, even among infants who receive breast milk via the bottle, rather than formula. Although formula and other nonhuman milk can be provided only by bottle, breast milk can be provided either via the breast or the bottle. Many breastfeeding mothers, particularly those who choose to or are obligated to return to work during breastfeeding, will pump their milk to provide bottled breast milk for their infants during their absence.
This study uses a longitudinal (cohort) design beginning at birth. Mothers of infants born ≥35 weeks gestational age with birth weights ≥2250 g were mailed 10 questionnaires at approximately monthly intervals during the first year. Although 3033 mothers completed neonatal questionnaires, only 1899 mother–infant pairs provided valid weight and feeding data to be included in the analysis.
The authors found that bottle feeding itself, even with breast milk, was associated with a higher rate of weight gain over the first year of life (adjusted mean difference of weight gain velocity in infants who were fed breast milk by bottle versus those breastfed only=88.8 (95% CI 13.2 to 164.5 g/mo)).
The major strengths of this study include its longitudinal design and its sophisticated statistical analysis, which takes into account the repeated observations in the infants, the shape of the weight curve as a function of age, and a variety of potentially confounding covariates. The analytic model does not, however, adjust for regression to the mean: the tendency of infants who are heavier at the start of an interval to gain less weight during that interval.
Like most previous authors, Li et al assume that associations between infant feeding and growth reflect causal effects of the former on the latter. The fact that babies fed with bottles, even bottles containing breast milk, gained weight more rapidly than those receiving breast milk only from the breast does not necessarily mean, however, that the bottles caused the infants to gain more weight.
Infant feeding is a dynamic process determined largely by the infant's demand and the mother's supply. The larger the demand, the greater the potential supply, because the infant's suckling stimulates maternal milk production. Infants who are growing along a slower trajectory may well be satisfied with breastfeeding only, do not get hungry between breastfeedings, and thus are less likely to receive a bottle.1 In other words, the authors have not considered the reverse causality relationship: faster growth may lead to bottle feeding, rather than bottle feeding leading to faster growth. Indeed, we have shown that a randomised controlled trial (RCT) of a breastfeeding promotion intervention yields very different results for infant growth when analysed by intention to treat (as randomised) than when analysed observationally (as fed).2
It is difficult to study the receipt of breast milk via bottle versus breast using an RCT. Some mothers will be unwilling to bottle-feed their infants, others will be unwilling to proscribe the use of bottles if they seem needed, and many will be unwilling to accept randomisation. Until RCT evidence is produced, however, attempts to replicate these findings in other cohorts would be worthwhile. Animal studies could shed light on possible biological mechanisms, if faster growth is observed in those who receive bottle feeding.
Breastfeeding is superior to formula feeding with respect to risks of gastrointestinal and respiratory infection, atopic dermatitis in infancy, necrotising enterocolitis in preterm infants, and sudden infant death syndrome, as well as long-term neurocognitive development.3 ,4 Benefits for the mother include prolonged contraception and reduction in future risk of breast cancer and perhaps ovarian cancer.3 Which of these health benefits for the mother and child depend on breastfeeding or also accrue to infants who receive breast milk by bottle remains an open question.