点击显示 收起
1 From the Division of Nutrition and Physical Activity, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta.
2 Address reprint requests to R Li, Division of Nutrition and Physical Activity (MS K-25), Centers for Disease Control and Prevention, 4770 Buford Highway, NE, Atlanta, GA 30341-3717. E-mail: ril6{at}cdc.gov.
ABSTRACT
Background: Maternal obesity has been associated with poor lactation in animal models, but the results of related research in humans are inconclusive.
Objective: We tested the hypothesis that women who are obese before pregnancy or who gain excessive weight during pregnancy are less likely to initiate and maintain breast-feeding than are their normal-weight counterparts.
Design: We analyzed 124 151 mother-infant pairs from the Pediatric Nutrition Surveillance System and the Pregnancy Nutrition Surveillance System. Body mass index (BMI) before pregnancy and gestational weight gain were categorized according to guidelines from the Institute of Medicine. Multiple logistic regression was used to identify the association between maternal obesity and breast-feeding initiation (n = 51 329), and multiple linear regression was used to examine the effect of maternal obesity on breast-feeding duration among women who initiated breast-feeding (n = 13 234).
Results: Regardless of gestational weight gain, obese women were less likely to initiate breast-feeding than were women with a normal BMI before pregnancy who also gained the recommended weight during pregnancy. Maternal BMI before pregnancy and gestational weight gain were each independently associated with duration of breast-feeding. Women who were obese before pregnancy breast-fed 2 wk less than did their normal-weight counterparts, and women who either failed to reach or exceeded the recommended gestational weight gain breast-fed 1 wk less than did those who gained the recommended gestational weight.
Conclusions: Both obesity before pregnancy and inadequate weight gain during pregnancy have a negative effect on breast-feeding practice. Women who are obese before pregnancy or who gain inadequate weight during pregnancy need extra support for breast-feeding.
Key Words: Maternal obesity • BMI before pregnancy • gestational weight gain • breast-feeding initiation • breast-feeding duration • Pediatric Nutrition Surveillance System • Pregnancy Nutrition Surveillance System
INTRODUCTION
Despite the recognized benefits of breast-feeding (1–3), breast-feeding rates in the United States are far from national goals (4). Risk factors for poor breast-feeding practices include individual factors such as maternal age, level of education, smoking, and ethnicity; biomedical factors such as parity, method of delivery, pregnancy complications, and infant health; and social or environmental factors such as hospital practices, work environment, and social support (5, 6).
Recently, maternal obesity has also been associated with failure to breast-feed. Descriptive studies found that the initiation and duration of breast-feeding were poor among mothers who had a body mass index (BMI; in kg/m2) above the normal range (7, 8); several prospective studies also found that maternal fatness negatively affected breast-feeding. Investigators in Australia reported that among women who successfully breast-fed 2 wk, those who were heavier 1 mo after delivery (BMI > 26) had a risk of early breast-feeding cessation that was 1.5 times that of those who were lighter (BMI 26) (9). On the basis of clinical chart reviews, researchers in the United States found that both overweight (BMI of 26.1–29.0) and obese (BMI >29.0) women had a significantly higher risk of failing to initiate breast-feeding successfully than did normal-weight women [odds ratios of 2.54 (P < 0.05) and 3.65 (P < 0.0008), respectively] (10). In addition, overweight and obesity were negatively associated with the duration of exclusive or any breast-feeding (10). In a study of 192 primiparous women from Connecticut, those with a "heavy/obese build" were significantly more likely to experience delayed onset of lactogenesis II than were those with a "slim/average build" (11). In contrast with the results of most of these studies, some researchers did not find an inverse correlation between BMI before pregnancy and breast-feeding practices (12, 13).
Because the prevalence of obesity has increased dramatically among women of reproductive age in the United States (14) and because increasing breast-feeding rates is among our national health goals (4), systematic investigations of maternal obesity in relation to breast-feeding are needed. To test the hypothesis that women who are obese before pregnancy or who gain excessive weight during pregnancy are less likely to initiate and maintain breast-feeding than are their normal-weight counterparts, we used data from the Pediatric Nutrition Surveillance System (PedNSS) and the Pregnancy Nutrition Surveillance System (PNSS) to conduct the present study.
SUBJECTS AND METHODS
Both the PedNSS and the PNSS were developed in the 1970s by the Centers for Disease Control and Prevention in collaboration with state health departments. The PedNSS and the PNSS are designed to monitor the general health and nutritional status of low-income US women and children participating in federally funded public health programs, such as the Special Supplemental Nutrition Program for Women, Infants, and Children. Most of the PedNSS and PNSS data are obtained from clinics that participate in the Women, Infants, and Children Program (75% in the PedNSS and 99% in the PNSS). The details of the study design and data collection were published elsewhere (15). In brief, the PedNSS collects data from public health clinics primarily on children < 5 y old. Every child’s visit to the participating clinic has a record; on average, children visit the clinic twice per year. In addition to obtaining demographic information about the children and monitoring their weight, height, and hemoglobin, the PedNSS collects 3 breast-feeding–related data items for children < 2 y old, including whether the child is currently breast-fed, has ever been breast-fed, and if ever, for how many weeks. The PNSS, in contrast, collects data, including demographic information, hemoglobin values, smoking during pregnancy, prenatal care, BMI before pregnancy, and weight gain during pregnancy, on pregnant and postpartum women.
To correlate maternal factors from the PNSS with breast-feeding data from the PedNSS, we linked the PNSS records for births in 1996 to the PedNSS records for children’s visits in 1996, 1997, and 1998 by identification number, state, and date of birth. To determine the duration of breast-feeding, we selected the last record of a child’s clinic visits for this study. Among 22 states participating in the PNSS in 1996, only 7 of them (Idaho, Illinois, Massachusetts, Michigan, Utah, Wisconsin, and Wyoming) were able to link to their PedNSS records. As a result, 124 151 records were linked for children aged < 24 mo, accounting for 82% of the total PNSS records.
The outcome variables for this study were breast-feeding initiation (defined as ever breast-feeding) and breast-feeding duration (the number of weeks the children received any breast milk). The independent variables of primary interest were BMI before pregnancy and gestational weight gain. BMI before pregnancy, which was calculated from self-reported weight before pregnancy and height measured at the initial clinic visit, was categorized as underweight (< 19.8), normal (19.8–26.0), overweight (> 26.0–29.0), or obese (> 29.0) according to guidelines from the Institute of Medicine (IOM) (16). Gestational weight gain was self-reported by the mother at the postpartum visit. The IOM recommendations (16) guided the categorization of gestational weight gain: below IOM (< 12.5, < 11.5, < 7.0, and < 6.0 kg for underweight, normal-weight, overweight, and obese women, respectively), above IOM (> 18, > 16, > 11.5, and > 9.1 kg for underweight, normal-weight, overweight, and obese women, respectively), and within IOM (gain between the cutoffs for below and above IOM). The IOM gave no upper limit for weight gain by obese women; our choice of 9.1 kg has been used by others (17).
Because the initiation and duration of breast-feeding are influenced by multiple factors, we assessed the effects of maternal obesity on breast-feeding practice by using multiple regression models that adjusted for a series of sociodemographic and health factors. The controlled variables included in the analysis were birth weight of the infant, gestational age of the infant (based on the onset of the women’s last menstrual period), parity, maternal age, education, marital status, race or ethnicity, smoking during the last 3 mo of pregnancy, initiation of prenatal care (month the woman began prenatal visits to a doctor or certified nurse midwife), and poverty-income ratio (the ratio of family income to the 1996 poverty thresholds by size of family produced by the US Census Bureau).
We applied multiple logistic regression analysis to examine the odds of failure to initiate breast-feeding. In this analysis, we excluded records with multiple births (n = 6254) and records that were missing data for the following: BMI before pregnancy and gestational weight gain (n = 12 939), characteristics of the children (n = 14 082) or their mothers (n = 35 281), and breast-feeding initiation (n = 3680). In addition, we excluded 586 records with biologically implausible values for birth weight of the infant, gestational age of the infant, maternal BMI, maternal age, and parity. This process yielded a final sample of 51 329 mother-infant pairs. To examine the relation between maternal obesity and breast-feeding duration, we applied multiple linear regression analysis to those women who initiated breast-feeding. Because nearly all of the women (98%) in this population had discontinued breast-feeding 10 mo after delivery, we excluded children who were younger than 10 mo or who were still breast-feeding at 10 mo of age so that we used only records of completed breast-feeding duration. Our final sample size for multiple linear regression analysis had 13 234 mother-infant pairs, which included only those women who initiated breast-feeding and who had complete records of breast-feeding duration. In the PedNSS database, some infants who were coded as currently breast-fed at the time of their clinic visit had actually been weaned. Because we could not separate these poor-quality records from those of the infants who were actually breast-fed at that time, we analyzed only the data obtained after the children were weaned instead of analyzing all the records by using Cox’s proportional hazard model.
We tested the significance of an interactive effect between BMI before pregnancy and gestational weight gain on breast-feeding initiation by using the likelihood ratio test to compare the baseline and interaction logistic models. The baseline logistic model, which was formulated in simple terms as breast-feeding initiation = (BMI before pregnancy)(weight gain)(. . .), examined the main effects of BMI before pregnancy and gestational weight gain adjusted for controlled variables, whereas the interaction logistic model included an additional interaction term, which was formulated in simple terms as breast-feeding initiation = (BMI before pregnancy)(weight gain)(. . .)(BMI before pregnancy x weight gain). To test the significance of an interactive effect between BMI before pregnancy and gestational weight gain on breast-feeding duration, a simultaneous test based on the sum of squares from full and reduced linear regression models was used to compare the baseline and interaction multiple linear regression models. In addition, we estimated the adjusted average breast-feeding duration for each level of BMI before pregnancy and gestational weight gain by using predicted values and the coefficients obtained from the multiple linear regression model. All statistical analyses were conducted with SAS version 6.12 (18).
RESULTS
Comparing the sociodemographic characteristics of the women who were included in the present study with those of the women who were excluded, we found no differences in BMI before pregnancy, gestational weight gain, parity, birth weight of the infant, gestational age of the infant, maternal age, education, or poverty-income ratio. The proportions of African Americans were similar in both populations (18%), whereas the study population had a slightly higher proportion of whites and a slightly lower proportion of Hispanics than did the excluded population (70% compared with 63% and 10% compared with 16%, respectively; data not shown).
The characteristics of the study population are shown in Table 1. In the total population, less than one-half of the women (47.6%) had a normal BMI before pregnancy, and 73.5% did not meet the recommended weight gain during pregnancy (31.0% gained less and 42.5% gained more than the IOM recommended amount). Overall, 46% of the women initiated breast-feeding, with a median duration of 6 wk (data not shown). The analysis of breast-feeding duration included only the women who initiated breast-feeding and had complete data on breast-feeding duration (n = 13 234). The distribution of maternal obesity in this subgroup was similar to that of the total population.
View this table:
TABLE 1 . Characteristics of the study population in multiple logistic regression (n = 51329) and multiple linear regression (n = 13234) analyses1
The adjusted odds ratios in multiple logistic regression analysis of failure to initiate breast-feeding for different characteristics are shown in Table 2. Because the likelihood ratio test showed a significant interaction effect between BMI before pregnancy and gestational weight gain on the odds of failure to initiate breast-feeding (chi-square test: 23.3; P < 0.001), the individual effects of prepregnant BMI and gestational weight gain are not shown in this table. The modifying effect of gestational weight gain on the relation between BMI before pregnancy and the odds of failure to initiate breast-feeding is shown in Figure1. Among the women who were underweight (BMI < 19.8) or normal-weight (BMI of 19.8–26.0) before pregnancy, the odds of never initiating breast-feeding were lower the greater the gestational weight gain, whereas no such relations were observed in the overweight (BMI of > 26.0–29.0) or obese (BMI > 29.0) women. Compared with the women who had a normal BMI before pregnancy and who also gained the recommended weight during pregnancy, obese women, regardless of their gestational weight gain, had higher odds of failure to initiate breast-feeding. Finally, using the women who met the guidelines for gestational weight gain as the referent for each category of BMI before pregnancy, gaining less than the recommended weight during pregnancy increased the odds of failure to initiate breast-feeding in all but the obese women.
View this table:
TABLE 2 . Adjusted odds ratios (ORs) in multiple logistic regression analysis of failure to initiate breast-feeding among 51329 mother-infant pairs1
FIGURE 1. . Adjusted odds ratios (ORs) and 95% CIs of failure to initiate breast-feeding by BMI values before pregnancy in women having different gestational weight gains [categorized as below (), within (), or above () the guidelines from the Institute of Medicine]. The women who had a normal BMI before pregnancy and who gained the recommended weight during pregnancy were considered as the reference group. *Significantly different from the reference group, P < 0.01.
Because the interaction test for the effect of BMI before pregnancy and gestational weight gain on breast-feeding duration was not significant (F = 0.6, P > 0.05), the baseline model was used for the final multiple linear regression analysis. The coefficients in Table 3 represent how many weeks less a mother with an abnormal BMI before pregnancy or an abnormal gestational weight gain breast-fed than did the reference group after all the other factors were controlled for. The results show that BMI before pregnancy and gestational weight gain were each significantly related to breast-feeding duration. Breast-feeding lasted 2 wk less in the women who were obese before pregnancy than in those who were normal-weight before pregnancy, and breast-feeding lasted 1 wk less in the women who either failed to reach or exceeded the recommended weight gain during pregnancy than in those who gained the recommended weight during pregnancy. On the basis of the predicted values and coefficients obtained from the multiple linear regression model, we estimated that the adjusted breast-feeding duration was 14 wk in the women who had a normal BMI before pregnancy or who gained the recommended weight during pregnancy, whereas the adjusted breast-feeding duration was 12 wk in the women who were obese before pregnancy and 13 wk in the women who gained less than or more than the recommended weight during pregnancy (Figures 2 and 3).
View this table:
TABLE 3 . Coefficients in multiple linear regression analysis for breast-feeding duration among 13234 mother-infant pairs1
FIGURE 2. . Adjusted mean breast-feeding durations (and 95% CIs) by BMI values before pregnancy among the women who initiated breast-feeding. The women who had a normal BMI before pregnancy were considered as the reference group. *Significantly different from the reference group, P < 0.01.
FIGURE 3. . Adjusted mean breast-feeding durations (and 95% CIs) by gestational weight gain categorized according to guidelines from the Institute of Medicine (IOM) among the women who initiated breast-feeding. The women who gained the recommended weight during pregnancy were considered as the reference group. *Significantly different from the reference group, P < 0.01.
The multiple logistic regression analysis identified several other significant risk factors for failure to initiate breast-feeding: low infant birth weight, young maternal age, low maternal education, being unmarried or African American, smoking during pregnancy, late introduction of prenatal care, and poverty (Table 2). In contrast, the multiple linear regression analysis did not show that African American race or the late introduction of prenatal care was related to a shorter breast-feeding duration (Table 3). Although breast-feeding initiation rates did not differ significantly between the primiparous and the multiparous mothers, the duration of breast-feeding among the primiparous mothers was 5 wk shorter than that among the multiparous mothers.
DISCUSSION
The results of this study indicate that obese women, regardless of their gestational weight gain, are less likely to initiate breast-feeding than are women who have a normal BMI before pregnancy and who also gain the recommended weight during pregnancy. In addition, among the different groups of BMI before pregnancy (ie, underweight, normal-weight, overweight, and obese), the women who gained less than the recommended weight during pregnancy had a significantly higher chance of never breast-feeding than did those who gained the recommended weight for all the groups except the obese women. We also found that both maternal BMI before pregnancy and gestational weight gain independently affected the continuation of breast-feeding. The women who were obese before pregnancy breast-fed 2 wk less than did the women who were of normal weight, and the women who gained either less than or more than the recommended weight during pregnancy breast-fed 1 wk less than did those who gained the recommended weight.
On the basis of the results of the present study, the effect of maternal obesity on breast-feeding initiation and duration appears to be relatively small, but these results were obtained after adjustment for 10 potentially confounding variables and are compatible with results obtained for many other risk factors identified previously and in the present study, such as race or ethnicity and poverty. Among all the factors examined, low maternal education and smoking during the last trimester had the greatest effect on breast-feeding initiation (adjusted odds ratios of 2.47 and 1.73, respectively), and primiparity and smoking during the last trimester led to the shortest durations of breast-feeding (ß coefficients of -5.45 and -4.07, respectively). This provides further evidence that breast-feeding initiation and duration are influenced by multiple factors. A better understanding of how these factors are associated with infant-feeding decisions and practices may facilitate the development of more effective breast-feeding promotion strategies.
Although animal studies indicated that both milk volume and milk composition changed in obese nursing rats (19–22), very few human studies have examined the effect of maternal "overnutrition" on breast-feeding practices, and the results of those studies are inconsistent. Prospective studies conducted by Rutishauser and Carlin (9), Hilson et al (10), and Chapman and Pérez-Escamilla (11) suggest that obese women have a higher risk of not breast-feeding or of stopping breast-feeding prematurely than do normal-weight women, but a Danish cohort study (12) and a national survey in Italy (13) did not find that BMI before pregnancy was related to breast-feeding initiation or duration. Possible reasons for such inconclusive results include different study designs, measures of maternal obesity, and definitions of breast-feeding, as well as variations in how confounding factors were handled. Our results support the hypothesis that both abnormal BMI before pregnancy and abnormal gestational weight gain have a negative influence on breast-feeding practices. The present study is unique because it contained a large sample size and the data were obtained from a low-income US population with at least one-third of the women having a BMI before pregnancy that was above the normal range.
The mechanisms for the association of maternal obesity with poor breast-feeding practice are unclear. Obesity may produce various alterations in the hypothalamic-pituitary-gonadal axis and in fat metabolism, consequently affecting milk production and composition (23, 24). Alternatively, women with large breasts may have physical difficulties in breast-feeding (11). In addition to possible physiologic and biological explanations, psychosocial factors related to obesity may play a role in the failure to breast-feed (11, 25–27). The present study has several limitations. First, self-reported weight before pregnancy and gestational weight gain may be underestimated in overweight or obese women (28, 29). Even if this is true, we believe that this misclassification is not related to women’s breast-feeding practices. For nondifferential misclassification, the study results would be biased toward the null hypothesis (30). Second, because all the study participants were from low-income families, the results may not be extrapolated to other populations. Third, we have no data to elucidate the causal mechanism behind the association observed in the present study.
In conclusion, the present study shows that both obesity before pregnancy and inappropriate weight gain during pregnancy have a negative effect on breast-feeding practices. Their effects on breast-feeding initiation were interactive, whereas their effects on breast-feeding duration were independent. Health professionals should be aware that women who are obese before pregnancy or who gain too little or too much weight during pregnancy need extra support for breast-feeding.
ACKNOWLEDGMENTS
We acknowledge the state PNSS and PedNSS coordinators, whose cooperation made this work possible.
RL designed the study, analyzed the data, and wrote the manuscript. SJ helped with extraction and analysis of data. LG-S assisted with the study design, supervised data analyses, and contributed to the writing of the manuscript. None of the authors had any financial or personal interest in any company or organization connected in any way with the research represented in the article.
REFERENCES