The risk of cessation of breastfeeding was increased among women who gained excess weight during pregnancy for women of normal and obese pre-pregnancy BMI, and a similar, although non-significant trend was observed for women with overweight pre-pregnancy BMI. The association between these variables and time to exclusive breastfeeding cessation was attenuated in comparison. Overall, women who gained within the Institute of Medicine’s guidelines tended to demonstrate lower risk of breastfeeding cessation compared to women who gained outside of these recommendations in their pre-pregnancy BMI category.
These findings are similar to results from previous literature (13, 14) utilizing the Institute of Medicine’s 1990 guidelines; however, this is the first study that has found a significant effect of pregnancy weight gain since the 2009 guidelines were released. A study conducted in a US cohort, found no significant impact of pregnancy weight gain after adjustment for confounders (21). Two cohort studies evaluating the impact of pregnancy weight gain and pre-pregnancy BMI on breastfeeding outcomes conducted in Brazil and China (19, 20). The Brazilian study found no differences in any or exclusive breastfeeding duration by pregnancy weight gain category, and did not evaluate the interaction between pregnancy weight gain and pre-pregnancy BMI. The study conducted in China considered the interaction between these two factors, and similarly found no differences by pregnancy weight gain category. Importantly, this study’s BMI and pregnancy weight gain categorization was based on the Chinese classification system, which differs slightly from the Institute of Medicine’s recommendations. Differences in sample size as well as regional and global breastfeeding behavior, norms, and predictors may explain the differences in results observed in the current study.
The implications of these findings are an important factor for maternal and infant health extending well beyond the early postpartum period. Mothers with increased pre-pregnancy BMI and excess gestational weight gain are at increased risk of diabetes mellitus, higher BMI in later pregnancies, and greater risk of depression (22, 23). Breastfeeding reduces the risk of diabetes (24-28), is associated with less postpartum weight retention (29-36), and may be a protective factor for depression in the postpartum period (37-42). Women with above normal BMI also have increased risk of postmenopausal breast cancer (43-45) and ovarian cancer (45), both of which are negatively associated with breastfeeding (46). These associations highlight the important risk reduction that mothers at increased risk due to high BMI and excess weight gain could experience with improved breastfeeding outcomes.
Given the numerous maternal health benefits imparted by breastfeeding, especially given the increased risk state of mothers with above normal BMI or excess pregnancy weight gain, efforts to increase the proportion of infants who receive optimal nutrition in the first months of life is warranted, and should include information about appropriate weight gain during pregnancy. For mothers who prenatally intend to breastfeed, it may be helpful to provide them with information about improved breastfeeding duration among mothers who gain within the Institute of Medicine’s recommendations. For mothers who gain above recommendations, anticipatory guidance regarding common breastfeeding problems, additional breastfeeding support especially in the prenatal and early postpartum periods, and general education about the benefits of breastfeeding for mom and baby tailored to the risk factors most relevant to the individual mother may be helpful to improve breastfeeding outcomes. Individuals providing breastfeeding support should consider weight gain above the Institute of Medicine recommendations as a potential risk factor and increase the frequency of contact, assessment, and interventions, particularly in the hospital and immediate postpartum period.
Additional factors that could not be evaluated in this analysis include gestational diabetes, hypertensive disorders of pregnancy, postpartum hemorrhage, pre-eclampsia and C-section. These conditions are more common among women who gain excess weight during pregnancy and increase the risk of maternal intensive care unit admission and longer hospital stays, however, given that this increased risk is conferred at least partially from this excess weight gain, placing these important predictors of breastfeeding outcomes on the causal pathway between pregnancy weight gain and the outcomes evaluated. Including these conditions as covariates in multivariable models of breastfeeding outcomes would have resulted in underestimation of the association of interest. Due to a limited total sample size and low occurrence rates for these pregnancy complications in our sample it was not possible to evaluate these factors as potential mediators of the association between pregnancy weight gain and breastfeeding outcomes. Given the biological importance of these factors for successful lactation outside of maternal decision making (establishment of milk supply, timing of mature milk production, etc), future studies in large cohorts should evaluate whether a portion of the increased risk of poor breastfeeding outcomes among mothers who gain more than is recommended is due to these pregnancy and birth factors. Due to the small number of women with underweight pre-pregnancy BMIs, we were unable to evaluate the impact of pregnancy weight gain and weight status on breastfeeding outcomes in these women. Additionally, self-reported breastfeeding data were utilized for this study. As noted, the source of weight data on the birth certificate is consistent (prenatal/hospital records), however it is possible the some of the weights in these records were drawn from prior visit documentation rather than direct measurement.
While studies have demonstrated that PRAMS self-report measures of breastfeeding initiation show a high degree of agreement with the birth certificate (47), similar studies evaluating post-hospital breastfeeding outcomes are not available. We estimate that on average, breastfeeding duration may have been mildly overestimated due to social desirability bias, however, we do not believe this would be differential with respect to the exposure of interest, and therefore feel any bias in the effect estimates reported is likely in the direction of the null hypothesis. Additionally, given that that breastfeeding data for women who chose not respond to the survey, it is unclear if this sample is truly representative of the community, and is likely that responders were more likely to be breastfeeders. Importantly, the relationship between pregnancy weight gain and breastfeeding outcomes may be different between survey responders and non-responders. Utilizing data from sources less prone to response bias, social desirability and recall biases (e.g. the child’s pediatric record) would be a valuable addition to future studies.
Despite these limitations, this analysis has several key strengths. The sample of respondents included in this study are representative of the county from which they were sampled allowing for evaluation of this relationship in a sample that represents the communities with similar demographic characteristics. These results are likely generalizable to populations with similar characteristics, however, are not generalizable to women with underweight pre-pregnancy BMIs and should be replicated in other populations. Categorizing pre-pregnancy BMI status and pregnancy weight gain using birth certificate data (originally abstracted directly from the medical record by birth certificate coders) eliminates concerns about social desirability bias that is often present in self-report data regarding weight. Additionally, this association should be evaluated in regions with greater diversity in terms of cultural, ethnic and racial backgrounds, and with differing health care infrastructure around pregnancy and breastfeeding support in order to understand the importance of these factors in the association of interest. Finally, having data regarding breastfeeding outcomes extending to four months allowed us to evaluate the risk of cessation over a long follow up period, and provides detailed time to event breastfeeding data that captures the individual variation in breastfeeding outcomes in this population.