点击显示 收起
1 From the Division of Nutritional Sciences, Cornell University, Ithaca, NY (JLB and KMR); the Department of Human Nutrition and Centre for Food Research, Faculty of Life Science, University of Copenhagen, Frederiksberg, Denmark (KFM and KMR); and the Institute of Preventive Medicine, Copenhagen University Hospital, Centre for Health and Society, Copenhagen, Denmark (JLB, TIAS, and KMR)
2 Data collection supported by the Danish National Research Foundation, Danish Pharmaceutical Association, Ministry of Health, National Board of Health, Statens Serum Institut, BIOMED, March of Dimes Birth Defects Foundation, Danish Heart Association, Danish Medical Research Council, and Sygekassernes Helsefond. Data analysis supported by Hatch grant NYC399405 (to KMR) and a grant from the Einaudi Center at Cornell University (to JLB). JLB was supported by National Institutes of Health training grant HD07331 (to KMR). 3 Reprints not available. Address correspondence to JL Baker, Institute of Preventive Medicine, Copenhagen University Hospital, Centre for Health and Society, Øster Søgade 18, 1st floor, DK1357 Copenhagen K, Denmark. E-mail: jba{at}ipm.regionh.dk.
ABSTRACT
Background: An association between high prepregnant body mass index (BMI) and early termination of breastfeeding has been observed, but this finding may have depended on the sociocultural context.
Objective: The objective was to determine whether this association was stronger with increasing maternal obesity, was modified by gestational weight gain, and still existed when there was greater social support for breastfeeding.
Design: Study participants (37 459 women) were drawn from the Danish National Birth Cohort. The association of prepregnant BMI and gestational weight gain with the termination of full or any breastfeeding by 1, 16, or 20 wk postpartum was assessed with logistic regression analyses, and the risk of early termination of full and any breastfeeding during the first 18 mo postpartum was assessed with Poisson regression analyses.
Results: The risk of early termination of any (with similar results for full) breastfeeding rose progressively with increasing prepregnant BMI values (in kg/m2), from 1.12 (95% CI: 1.09, 1.16) for overweight (BMI = 25.0–29.9) women to 1.39 (95% CI: 1.19, 1.63) for obese class III women (BMI 40) compared with normal-BMI women. Gestational weight gain did not add to or modify the association between prepregnant BMI and breastfeeding.
Conclusions: These findings extend the observation to a broader range of BMIs that the greater the prepregnant BMI, the earlier the termination of breastfeeding. Together with the fact that this association was evident in a more supportive social context for breastfeeding, these findings suggest a biological basis for the association.
Key Words: Breastfeeding • prepregnancy • body mass index • Danish women • obesity
INTRODUCTION
An association between higher prepregnant maternal body mass index (BMI) and shorter duration of breastfeeding has been observed in populations of the United States (1-3) and elsewhere (4-6). There is evidence for a biological basis for this association, inasmuch as being heavier is associated with a reduction in the prolactin response to suckling in the first week postpartum (7) and with a delay in the onset of copious milk secretion (lactogenesis II; 8). Data from both rats (9, 10) and mice (11) also suggest that fatter animals experience difficulties producing milk in the early postpartum period.
This association is of particular concern in the United States, where the proportions of overweight and obese women of reproductive age remain exceptionally high (51.7% and 28.9%, respectively), according to the latest available nationally representative data (1999–2004; 12). Even more worrisome is the historically high proportion (8.0%) of extremely obese women [BMI (in kg/m2) 40] aged 20–39 y (12). Moreover, although the proportion of American women who attempt to breastfeed is approaching the national target of 75% in some subgroups, the proportion who continue for as long as is recommended (50% at 6 mo postpartum and 25% at 12 mo postpartum) remains far below (13, 14) current national targets (15). A better understanding of how obesity affects the initiation and continuation of breastfeeding is thus essential for meeting these public health goals.
Many factors in addition to overweight or obesity make it difficult for women in the United States to breastfeed if they wish to do so (16, 17), but these factors are much reduced in other places. Notably, breastfeeding is nearly universal among women in Scandinavian countries, where maternity leave is generous and social support for breastfeeding is high (18). For the research presented herein, we used data from the Danish National Birth Cohort (DNBC), a large contemporary sample of women recruited from throughout Denmark, to ascertain whether the negative association between maternal prepregnant BMI and the duration of breastfeeding also occurred under these different social circumstances.
The exceptional statistical power available in the DNBC permitted us to explore whether the limited association between greater BMI and early termination of breastfeeding that we observed in a much smaller sample of women (1) would extend to higher degrees of obesity. In addition, we were able to investigate how gestational weight gain (GWG) affected the association between prepregnant BMI and the duration of breastfeeding. In smaller studies, we (19) and others (3) observed an additive relation.
SUBJECTS AND METHODS
Subjects
The study used data from the DNBC (20) and included subjects who gave birth from 1999 to 2002. Women were invited to participate in the cohort by their general practitioner at the first prenatal visit at 6–12 wk of pregnancy. To be eligible for the cohort, the women had to plan to continue with the pregnancy, live in Denmark, and speak Danish well enough to participate in telephone interviews (20). After enrollment, women were interviewed by telephone 4 times: at 12 and 26 wk of gestation and at 6 and 18 mo postpartum. Information about medical conditions that the woman experienced during pregnancy and delivery and about the characteristics of the infant at birth was obtained via a linkage between the subjects’ Civil Person's Registry number and the Danish National Patient Registry. This study was approved by both the University Committee on Human Subjects at Cornell University and the Danish Data Protection Agency.
To be considered eligible for this investigation, participants in the DNBC had to have completed at least interviews 1–3, have full data available from the linked registers, and have delivered a liveborn, singleton infant. From these women, we selected those who had responded to the version of the questionnaires that permitted us to determine the duration of full breastfeeding. A total of 5209 women participated more than once in the DNBC and completed interviews 1–3 or 1–4. When women participated more than once, we systematically chose the pregnancy in which a woman completed 4 interviews. If participants only completed 3 interviews, we systematically chose the second pregnancy within the DNBC. On the basis of these criteria, 42 369 women were available for this investigation (Figure 1). Crucial information was missing for 1263 subjects (maternal BMI: n = 659; GWG: n = 376; birth weight: n = 225); an additional 3 women did not wish to answer the infant feeding questions, which left 41 106 women who were eligible for the investigation. From this group, we excluded subjects who were aged <18 y or >45 y (n = 40), who had preexisting or gestational diabetes (n = 554), who chose to not breastfeed (n = 745) or used an alternative breastfeeding method (eg, breast pump only or feeding the infant banked human milk; n = 14), whose infant was born preterm (gestation <259 d; n = 1316) or at a very low birth weight (<2000 g; n = 29), or whose infant had a birth defect, severe illness, or other condition that might preclude successful breastfeeding (as determined by KFM; n = 420). In addition, we excluded women (n = 529) for whom the duration of breastfeeding could not be determined; most of these women had extreme inconsistencies in their answers to the infant feeding questions. A total of 37 459 women were included in the investigation.
FIGURE 1.. Flow diagram for selection of subjects for the study from the Danish National Birth Cohort. BF, breastfeeding.
Maternal variables
Maternal characteristics, such as age, parity, and self-reported prepregnant weight and height, were obtained during the interview at 12 wk of gestation. The woman's BMI was calculated by using her self-reported prepregnant weight and height. BMIs were classified into 6 categories according to the criteria of the World Health Organization (21): underweight (<18.5), normal weight (18.5–24.9), overweight (25.0–29.9), obese class I (30.0–34.9), obese class II (35.0–39.9), and obese class III (40.0).
Information about the development of maternal health conditions and smoking habits during and after pregnancy was obtained during the interviews at 12 and 26 wk of pregnancy and at 6 mo postpartum. Women were categorized as having never smoked or ever smoked in each of 2 periods: at any time during pregnancy and during the breastfeeding period. Total GWG was obtained by self-report during the interview at 6 mo postpartum and was characterized by using the groupings proposed by Cedergren (22) for Swedish women: <8 (lower weight gain), 8–15.9 (reference group), and 16 kg (higher weight gain).
For these analyses, women were categorized at the first pregnancy interview as living with a spouse or partner or living alone. Maternal occupation was coded by DNBC staff members into 6 categories: 1) worker at a job that required the highest level of education, 2) worker at a job that required a medium level of education, 3) skilled worker, 4) unskilled worker or unemployed, 5) student, and 6) information missing. Physical activity during late pregnancy was assessed during the interview at 6 mo postpartum and was used as a marker of a healthier lifestyle. We coded as yes or no the women's responses to the question "did you engage in any sort of exercise during the last part of pregnancy? "
Infant feeding variables
In the interview at 6 mo postpartum, women were asked about how they had been feeding and were currently feeding their infants, including the duration of full and any breastfeeding, the use of infant formula or cow milk, and the introduction of complementary food. The duration of full breastfeeding was reported by mothers in months and weeks or in weeks and days and was then converted into weeks, as was the duration of any breastfeeding and the infant's age at which mothers began feeding infant formula. If, at the time of this interview, the mother reported that her infant was currently being fed infant formula daily, she was asked when this began, and the response was recorded in 0.5-mo intervals. The infant's age at the time the mothers began feeding complementary foods (solid or mushy) was reported in weeks or in months and weeks and was then converted into weeks.
We termed our outcome "full" breastfeeding because we could not distinguish "full" breastfeeding from the more restrictive definition of "exclusive" breastfeeding (no liquids other than human milk and no solid foods given to the infant) (23) with certainty. The questionnaire used in the DNBC included a number of different ways to describe infant feeding practices, so it was possible for the various indicators of the duration of full breastfeeding to be inconsistent with one another. To resolve the inconsistencies in the responses to these questions, which occurred in 12% of the subjects, we used an approach that respected the definition of full breastfeeding and came as close as possible to a definition for exclusive breastfeeding (24). To establish the time at which infant formula was introduced, the earlier of the 2 possible dates (first use of infant formula or regular use of infant formula) was chosen. To establish the duration of full breastfeeding, the reported value for full breastfeeding was compared with the reported ages at which infant formula and complementary foods were introduced. If these values were within 2 wk of one another, the reported value for the duration of full breastfeeding was used. If these values differed by >2 wk, the duration of full breastfeeding was set to the infant's age at the introduction of infant formula or complementary food, whichever was earlier (n = 3120 subjects). Finally, for women with a reported duration of any breastfeeding that was shorter than that of full breastfeeding or who had other inconsistencies in responses that could not be resolved, the duration of full breastfeeding was declared to be missing (n = 529).
Infant variables
The length of gestation was calculated by DNBC staff members from the women's self-reports of the date of her last menstrual period or from data from an ultrasound assessment of fetal size early in the pregnancy. The sex of the infant was obtained from the Danish National Birth Register. Infant birth weight and length were also obtained from the Danish National Birth Register and were used as continuous variables.
Statistical methods
To test whether the characteristics of included and excluded women were similar, Student's t test or chi-square analysis was used as appropriate. To test whether the characteristics of the women differed by prepregnant BMI, analysis of variance was used. Dunnett's t test was used to compare women in the other prepregnant BMI categories with normal-weight women.
From the data obtained at the interview at 6 mo postpartum, we also determined whether maternal prepregnant BMI was negatively associated with the termination of full and any breastfeeding early in lactation (1 wk) and at 2 later times, 16 and 20 wk postpartum. These 2 later times correspond to the latest time (16 wk) before the Danish health authorities recommended the introduction of solid food (which would cause the end of full breastfeeding) and to a time after this transition was well under way (20 wk). We examined this association at each of these times (ie, at 1 wk compared with >1 wk, <16 wk compared with 16 wk, and <20 wk compared with 20 wk) by using logistic regression adjusted for potentially confounding variables. Women of normal weight were the reference category.
We assessed the associations between prepregnant BMI and the duration of full or any breastfeeding, adjusted for potentially confounding factors. Survival analyses were made with log-linear Poisson regression (25). Follow-up ended on the date when the subject terminated breastfeeding (event) or, in women who continued to breastfeed, on the date when information about breastfeeding status was obtained (censored).
After examining the unadjusted associations of the variables of interest, we examined potentially confounding factors. We adjusted the logistic and Poisson regression models for maternal age as a continuous variable and for the following categorical variables: GWG (3 categories), presence of a spouse or partner during early pregnancy (yes or no), maternal occupation in early pregnancy, parity (primiparous: yes or no), maternal smoking during pregnancy (yes or no), mode of delivery (cesarean: yes or no), physical activity during late pregnancy (yes or no), and infant sex. Variables were added to the basic model containing maternal prepregnant BMI in a stepwise fashion and were eliminated if they failed to cause a change to the estimated coefficients for prepregnant BMI and GWG that exceeded 10%. Interactions between categories of prepregnant BMI and GWG were investigated. Significance was set at = 0.05. All data were analyzed with the Statistical Analysis System (version 9.2; SAS Institute, Cary, NC) on a personal computer.
RESULTS
As a result of the large number of eligible women in this investigation, even differences that were presumably biologically unimportant between those who were included and those who were excluded were statistically significant (Table 1). Some of the possibly noteworthy differences are as follows: the women who were included in these analyses were 2.6 kg lighter and thus had BMIs 1 unit lower, gained 1.3 kg more during pregnancy, were less likely to be primiparous and to have had a cesarean delivery, and were more likely to be continuing to breastfeed at 6 mo postpartum than were the women who were excluded from the study.
View this table:
TABLE 1. Characteristics of mother-infant dyads included in or excluded from the investigation1
There were several systematic differences between the 6 prepregnant BMI groups in this investigation (Table 2). Women in all of the other prepregnant BMI groups were significantly younger than the normal-weight women. This difference was 1 y for those with the lowest and highest BMIs. Women in the higher-BMI groups were less likely to be employed in jobs that required the highest level of education and were more likely to be employed as a skilled or unskilled worker or to be unemployed than were the normal-weight women. The women in the higher BMI groups gained less weight during pregnancy; they were also less likely to exercise in late pregnancy and were more likely to have a cesarean delivery than were normal-weight women. Birth weight increased progressively across the prepregnant BMI groups and was 380 g higher in the women in the class III obesity group than in the underweight women. Although all of the women in this sample breastfed their infants, the proportion of those who had ever fully breastfed was significantly lower in the women with higher prepregnant BMIs (Table 3); similar differences were observed in the proportions of women who had ever fed formula and in the age at which solid foods had been introduced to the infants. In addition, the duration of full and any breastfeeding and the proportion of women who had continued to breastfeed to any extent at the time of the 6-mo interview differed significantly between the prepregnant BMI groups: the duration of breastfeeding was shorter and the proportion of women who continued to breastfeed was lower among the heavier women (Table 3).
View this table:
TABLE 2. Characteristics of women and their infants by prepregnant BMI category1
View this table:
TABLE 3. Characteristics of women's infant feeding practices by prepregnant BMI category1
At the end of the first week after delivery, 4.2% of the 36 392 women who had ever fully breastfed their infants had ended full breastfeeding, and 1.6% of the 37 459 women who had ever offered their infants human milk had completely ceased breastfeeding. These proportions varied by maternal prepregnant BMI category (Figure 2). At the end of the first week after delivery, 14.4% of the women in obesity class III had ceased full breastfeeding, compared with 3.5% of normal-weight women. This difference was smaller for any breastfeeding—1.2% of normal-weight women compared with 6.0% of obese class II women. These differences between overweight or obese women and normal-weight women were all statistically significant in the adjusted logistic regression analyses for both full and any breastfeeding (Figure 3).
FIGURE 2.. Proportion of women who continued full or any breastfeeding varied by prepregnant BMI category. Unadjusted Poisson regression model data, truncated at 26 wk, are shown.
FIGURE 3.. Adjusted odds of terminating full or any breastfeeding at 1, 16, or 20 wk postpartum varied by prepregnant BMI category in women who had ever attempted to breastfeed their infants. Odds ratios (±95% CI) from the logistic regression models were adjusted for gestational weight gain, cesarean delivery, maternal age, smoking during lactation, parity, maternal occupation, spouse or partner present in early pregnancy, physical activity during late pregnancy, and infant sex as appropriate. Normal-weight women were the reference group. P values for the overall association of prepregnant BMI with the breastfeeding outcome are included.
Before the time (at 16 wk postpartum) when the health authorities recommended the introduction of solid foods, 29–56% of the women in each prepregnant BMI group had terminated full breastfeeding and 15–44% had terminated any breastfeeding. The proportion of women who continued full or any breastfeeding decreased with increasing BMI (Figure 2). Each of the heavier groups differed significantly from the normal-weight group in the adjusted logistic regression analyses (Figure 3).
By 20 wk postpartum, 72–87% of the women in each prepregnant BMI group had terminated full breastfeeding, and 21–51% had terminated any breastfeeding (Figure 2). Again, the proportion of women who continued full or any breastfeeding decreased with increasing BMI (Figure 2). The overall differences between the BMI groups remained statistically significant (full breastfeeding: P < 0.01; any breastfeeding: P < 0.0001) in the adjusted logistic regression analyses (Figure 3). However, the differences between the overweight and obese women and the normal-weight women were no longer statistically significant in the adjusted logistic regression analyses for full breastfeeding (Figure 3, top) but were statistically significant for any breastfeeding (except for women in the obese III category of prepregnant BMI; Figure 3, bottom).
Overall, there was a progressively higher risk of terminating full or any breastfeeding sooner as prepregnant BMI increased (as shown in the Poisson regression analyses in Table 4). As expected, primiparity, cesarean delivery, and smoking during the breastfeeding period were also associated with an increased risk of early termination of full and any breastfeeding (Table 4). In addition, compared with working in the occupations that required the highest level of education or with being a student, working in other occupations was associated with an increased risk of early termination of breastfeeding. Being inactive in late pregnancy was also associated with an increased risk of early termination of breastfeeding. Women who were breastfeeding female infants continued full breastfeeding longer than did those breastfeeding male infants.
View this table:
TABLE 4. Final adjusted Poisson regression model for the association of maternal prepregnant BMI with the duration of full and any breastfeeding in 37 459 women in the Danish National Birth Cohort
Inasmuch as GWG also contributes to maternal fatness at the time lactation begins, we also examined whether GWG modified this association or was independently associated with the initiation and continuation of breastfeeding. GWG increased the odds of termination of breastfeeding inconsistently. It was significant for the termination of full breastfeeding at 1 (P < 0.0001), 16 (P < 0.05), and 20 (P < 0.05) wk. It was significant for the termination of any breastfeeding only at 16 and 20 wk (P < 0.0001 for both). In the unadjusted Poisson regression analysis, lower and higher GWG values were significantly associated with early termination of both full [relative risk (RR) of 1.13 (95% CI: 1.08, 1.18) and RR of 1.05 (95% CI: 1.03, 1.08), respectively] and any breastfeeding [RR of 1.16 (95% CI: 1.11, 1.22) and RR of 1.05 (95% CI: 1.03, 1.08), respectively]. However, GWG was not a significant predictor of early termination of full or any breastfeeding when prepregnant BMI was in the model, and, thus, GWG did not enter the final model (Table 4). We found that GWG did not interact with prepregnant BMI to affect the odds of terminating full or any breastfeeding in the logistic regression analyses nor the risk of early termination of full or any breastfeeding in the Poisson regression analyses. To be sure that this was not the result of inadequate statistical power among the most obese women, we also ran the Poisson analyses with fewer categories of maternal prepregnant BMI and tertiles [instead of Cedergren's (22) categories] of GWG and obtained similar results.
DISCUSSION
Using an exceptionally large contemporary sample of Danish women, who reported their infant feeding behavior in detail as it was progressing, we showed that higher prepregnant BMI was consistently associated with the early termination of breastfeeding. The risk of termination rose steadily with increasing prepregnant BMI. This association raises concerns about the ability of similar white, American women, among whom 8% of those of reproductive age now have a BMI 40 (12), to meet national goals for the duration of breastfeeding (26).
The social environment for breastfeeding in Denmark, which includes a 24-wk maternity leave and support for breastfeeding from health visitors who work with all new mothers, is quite different from that for American women. In previous studies (1-3), it was not possible to tell whether the association between high prepregnant BMI and early termination of breastfeeding was the result of some unmeasured aspect of poor social support for breastfeeding in the United States or whether it reflected an underlying causal, biological relation. There is currently more evidence of such a biological relation (7, 11; see reference 27 for a more comprehensive review). Moreover, the findings of the present investigation confirm that this relation exists when social support of breastfeeding is high and provide evidence of a "dose-response" relation, which suggests an underlying biological phenomenon, provided that there is no analogous unknown confounding factor.
The design of the DNBC provided an advantage over prior studies in American women that were based on a review of medical records (1, 2) or that used surveillance data based on clinic records collected as part of the Special Supplemental Nutrition Program for Women, Infants, and Children (3). In the DNBC, the women reported directly and systematically on multiple aspects of their ongoing infant feeding practices. The richness of these data was reflected in their complexity. When there were inconsistencies in these data, we used a conservative approach to determine the duration of full breastfeeding accurately. As a result, it may have caused us to underestimate the true duration of full breastfeeding in those women for whom it was necessary to reset the duration of full breastfeeding to the infant's age at the introduction of infant formula or complementary food. However, there was no significant difference (data not shown; P = 0.16 by analysis of variance) in prepregnant BMI between women with and without consistent responses for the duration of full breastfeeding. Therefore, this way of resolving the inconsistencies in the data was unlikely to have introduced bias into our comparisons.
The richness of these data also gave us access to information on an exceptional range of possible confounding factors, and we adjusted our analyses for them. Nonetheless, the DNBC is an observational study, so it is possible that unmeasured confounding remains. The consistency of these findings with the available data from American (1-3) and Australian (4-6) women—despite differences in research design and in the sociodemographic characteristics of the women—led us to conclude that the confounding of our findings by such unmeasured factors is likely to be minimal.
From the results of 2 prior studies (3, 19) in which the role of GWG was examined, we expected that GWG would have an additional, independent association with the duration of breastfeeding beyond that of prepregnant BMI. We showed that GWG, although shown to be associated with the early termination of breastfeeding at specific times in the first 22 wk postpartum according to the adjusted logistic regression analyses, was not a risk factor for the early termination of breastfeeding when prepregnant BMIs were in the adjusted Poisson regression models. In addition, GWG did not interact with prepregnant BMI to affect the duration of breastfeeding. In this investigation, the odds of early termination of breastfeeding were highest among women in the lower (<8 kg) GWG category. The women who gained this amount were likely to have come from the obesity class II (mean GWG: 8.9 kg) and class III (mean GWG: 6.1 kg) groups, so it is not surprising that when prepregnant BMI data were incorporated into the Poisson regression model, GWG became nonsignificant.
This investigation differed from the 2 prior studies (3, 19) conducted in American women in other important ways. In those studies, both maternal prepregnant BMI and GWG were categorized per the criteria of the Institute of Medicine (IOM) (28). In contrast, in the present investigation, maternal prepregnant BMI was categorized per the World Health Organization criteria (21) and GWG per cutoffs suggested by Cedergren (22) for Swedish women. Despite the fact that the cutoff for obesity was higher and the population was leaner overall in these Danish women than in the American women studied previously, this investigation still included nearly as many obese subjects as the investigation by Li et al (3) because it included nearly 3 times as many subjects in total.
The differences in criteria for GWG were more substantial. The IOM recommends that women gain less weight as their prepregnant BMI increases, so women in the 2 American studies were categorized as being above, within, or below the IOM's recommended range of GWG. Categorizing women according to the IOM's recommendations produces results that have clear implications for public health, but the biological interpretation of these findings is less clear. To assure ourselves that we were not failing to detect an association of GWG with the duration of breastfeeding that was present but hidden by these differences in analytic approach, we also characterized our data by the IOM categories for prepregnant BMI and GWG (28) and ran a similar series of Poisson regression models. As we observed previously (1), women who were overweight or obese before pregnancy had a significantly (P < 0.0001) higher risk of early termination of full breastfeeding [RR of 1.08 (95% CI: 1.04, 1.11) for overweight women and RR of 1.17 (95% CI: 1.13, 1.21) for obese women] or any breastfeeding [RR of 1.13 (95% CI: 1.08, 1.17) for overweight women and RR of 1.24 (95% CI: 1.19, 1.28) for obese women] in the fully adjusted Poisson regression models. In the unadjusted Poisson regression models, only GWG that was above the IOM recommendations was significantly associated with early termination of both full and any breastfeeding. When prepregnant BMI was added to the model, the regression coefficients for GWG were substantially diminished but were still statistically significant (data not shown). However, in the fully adjusted Poisson regression model, GWG above the recommendations was no longer significant (data not shown). From the results of the fully adjusted Poisson regression analyses—regardless of how GWG is characterized—we concluded that GWG did not add to or modify the risk of early termination of breastfeeding that is already captured by maternal prepregnant BMI.
Most of the maternal characteristics that were significant in the adjusted Poisson regression models behaved as expected from the literature on breastfeeding. Younger maternal age, primiparity, lower educational attainment or working in an occupation with a lower skill level, cesarean delivery, and maternal smoking have all been shown to be associated with less successful breastfeeding in prior studies (3, 29, 30). To our knowledge, a lower level of physical activity during pregnancy has not previously been studied in relation to the duration of breastfeeding. It is noteworthy that the association of obesity with early termination of breastfeeding was stronger than that of primiparity and was similar to that of smoking.
From the data available in the DNBC, we were able to examine full and any breastfeeding as separate outcomes. Prepregnant BMI was associated with early termination of breastfeeding in both cases but became attenuated over time for full breastfeeding, which was as expected because of the introduction of solid foods into the infants’ diets. However, this attenuation was less for any breastfeeding. We speculated that heavier women may have moved rapidly to the complete cessation of breastfeeding, not just to ending full breastfeeding by adding infant formula or by introducing solid foods early to their infants’ diets, as a way to deal with the challenges that they were experiencing with breastfeeding or for other reasons.
In conclusion, the results of this study show clearly that the greater the prepregnant BMI, the earlier the termination of breastfeeding in a sociocultural context in which breastfeeding is strongly supported. These results are in accord with other data from other species (27) that indicate that this association has a biological basis. The fact that heavier Danish women breastfeed for a much longer time than do heavier American women (1-3) suggests that social support plays an important role in overcoming the challenges to breastfeeding that are associated with being heavier. This is an argument for continuing to encourage women to breastfeed so that more will know how to do so successfully and be able to share this knowledge with other women. This is also an argument for providing heavier women with the additional assistance that they need to become successful breastfeeders. The results of this study also show that GWG (at least in the amounts gained by these Danish women) did not significantly add to or modify the association between prepregnant BMI and early termination of breastfeeding. The clear implication of these findings is that women should begin pregnancy at a healthy weight, although it is unknown whether it matters for the success of breastfeeding how women achieve this goal. The implications of this study are in accord with new American guidelines concerning preconceptional health (31), which call for treating every woman of reproductive age as prepregnant.
ACKNOWLEDGMENTS
The authors’ responsibilities were as follows—JLB: prepared the data set, conducted the data analysis, and contributed to the manuscript; KFM: developed the infant feeding questions that were included in the DNBC, identified the infants who were ill, and contributed to the manuscript; TIAS: participated in the initiation and planning of the DNBC, planned and coordinated the postpartum data collection, and contributed to the manuscript; and KMR: conceived the study, began the data preparation, and prepared the manuscript. None of the authors were affiliated with any advisory board or had any financial interests in the organizations that sponsored this research.
REFERENCES