Emily Oken, Rita Patel, Lauren B Guthrie, Konstantin Vilchuck, Natalia Bogdanovich, Natalia Sergeichick, Tom M Palmer, Michael S Kramer, and Richard M Martin ABSTRACT Background: Differences between mothers who do and do not succeed in breastfeeding are likely to confound associations of lactation with later maternal adiposity. Objective: We compared adiposity and blood pressure (BP) in women randomly assigned to an intervention to promote prolonged and exclusive breastfeeding or usual care. Design: We performed a cluster-randomized trial at 31 hospitals in Belarus in 1996–1997. Results: Of 17,046 women enrolled at delivery, we assessed 11,867 women (69.6%) at 11.5 y postpartum. The prevalence of exclusive breastfeeding $3 mo was 44.5% in 6321 women in the intervention group and 7.1% in 5546 women in the control group. At 11.5 y postpartum, mean (6SD) body mass index (BMI; in kg/m2) was 26.5 6 5.5, the percentage of body fat was 33.6% 6 8.3%, and systolic BP was 124.6 6 14.6 mm Hg. On intention-to-treat analysis (without imputation) with adjustment for clustering by hospital, mean outcomes were lower in intervention compared with control mothers for BMI (mean difference: 20.27; 95% CI: 20.91, 0.37), body fat (20.49%; 95% CI: 21.25%, 0.27%), and systolic BP (20.81 mm Hg; 95% CI: 23.33, 1.71 mm Hg), but effect sizes were small, CIs were wide, and results were attenuated further toward the null after adjustment for baseline characteristics. Results were similar in sensitivity analyses [ie, by using conventional observational analyses disregarding treatment assignment, instrumental variable analyses to estimate the causal effect of breastfeeding, and multiple imputation to account for missing outcome measures (n = 17,046)]. Conclusion: In women who initiated breastfeeding, an intervention to promote longer breastfeeding duration did not result in an important lowering of adiposity or BP. This trial was registered at clinicaltrials.gov as NCT01561612 and at Current Controlled Trials as ISRCTN37687716. Am J Clin Nutr 2013;98:1048–56.

INTRODUCTION

The reset hypothesis suggests that lactation plays a central role in mobilizing fat stores that a mother accumulates during pregnancy (1). Two small randomized trials showed that mothers who breastfed exclusively for 6 compared with 4 mo lost more weight by 6 mo pospartum (2, 3). However, these trials were

1048

conducted in a developing country with a very low prevalence of obesity and, thus, may have limited generalizability to populations at risk of obesity. In observational studies, most of which were conducted in Western populations, a longer breastfeeding duration has been associated with lower maternal weight, central adiposity, and blood pressure (BP)5 and reduced future risk of diabetes and cardiovascular disease (4–13). 1 From the Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA (EO and LBG); the School of Social and Community Medicine (RP and RMM) and the Medical Research Council/University of Bristol Integrated Epidemiology Unit (RMM), University of Bristol, Bristol, United Kingdom; the National Research and Applied Medicine Mother and Child Centre, Minsk, Republic of Belarus (KV, NB, and NS); the Division of Health Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom (TMP); the Departments of Pediatrics, Epidemiology, and Biostatistics, McGill University Faculty of Medicine, Montreal, Canada (MSK); and the University of Bristol/University Hospitals Bristol National Health Service Foundation Trust, National Institute for Health Research Bristol Nutrition Biomedical Research Unit, Bristol, United Kingdom (RMM). 2 Funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of data; or preparation, review, or approval of the manuscript. 3 The Promotion of Breastfeeding Intervention Trial III was supported by grants from the Canadian Institutes of Health Research (MOP-53155), the US National Institute of Child Health and Development (R01 HD 050758), and the European Union’s project on Early Nutrition Programming: Longterm Efficacy and Safety Trials (FOOD-DT-2005-007036). EO was supported by the US National Institutes of Health (K24 HD069408 and P30 DK092924). RMM works within the Integrated Epidemiology Unit, which is supported by the Medical Research Council and the University of Bristol. The National Institute for Health Research (NIHR) Bristol Nutrition Biomedical Research Unit (RMM) is funded by the National Institute for Health Research (NIHR) and is a partnership between University Hospitals Bristol National Health Service Foundation Trust and the University of Bristol. 4 Address correspondence to E Oken, Obesity Prevention Program, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, 133 Brookline Avenue, Boston, MA. E-mail: [email protected]. 5 Abbreviations used: BP, blood pressure; FFMI, fat-free mass index; FMI, fat mass index; ICC, intraclass correlation; PROBIT, Promotion of Breastfeeding Intervention Trial. Received April 30, 2013. Accepted for publication July 9, 2013. First published online August 14, 2013; doi: 10.3945/ajcn.113.065300.

Am J Clin Nutr 2013;98:1048–56. Printed in USA. Ó 2013 American Society for Nutrition Supplemental Material can be found at: http://ajcn.nutrition.org/content/suppl/2013/09/16/ajcn.113.0 65300.DCSupplemental.html

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

Effects of an intervention to promote breastfeeding on maternal adiposity and blood pressure at 11.5 y postpartum: results from the Promotion of Breastfeeding Intervention Trial, a cluster-randomized controlled trial1–4

BREASTFEEDING TRIAL AND MATERNAL ADIPOSITY

SUBJECTS AND METHODS

Study setting and population In 1996–1997, in the Republic of Belarus, 34 maternity hospitals and one each of their affiliated outpatient polyclinics were randomly assigned to receive a breastfeeding-promotion intervention (experimental group) or to continue the prevailing practices at the time of random assignment (control group) (17). The experimental intervention was based on the Baby Friendly Hospital Initiative developed by the WHO and UNICEF to promote and support breastfeeding, particularly in mothers who have chosen to initiate breastfeeding (17). Mothers were eligible if they initiated breastfeeding, had no contraindications to breastfeeding, and delivered a healthy singleton infant of $37 wk of gestation, $2500 g birth weight, and with a 5-min Apgar score $5. Less than 5% of women did not initiate breastfeeding, and only 1–2% of eligible women declined participation. The experimental intervention included 10 steps that maternity hospitals implemented to become certified as Baby Friendly (17), as summarized in detail previously (16). A total of 17,046 mothers and their surviving children were randomly assigned, and 97% of subjects completed the initial 12-mo follow-up. The random assignment produced groups with similar distributions of potentially confounding sociodemographic and clinical characteristics (16). Physicians (obstetricians and pediatricians) at participating study sites recorded sociodemographic and clinical information during the postpartum stay. At polyclinic visits at 1, 2, 3, 6, 9, and 12 mo, pediatricians completed a data form that contained detailed information about infant feeding. On the basis of WHO definitions (16, 17), we considered mothers to be exclusively breastfeeding at each time point if they provided their infant with only breast milk but no solids, nonhuman milk, water, or other liquids (other than vitamins or medications). A routine audit indicated excellent agreement between polyclinic charts and study documents regarding breastfeeding duration and exclusivity (16). At a study visit at 6.5 y postpartum, we asked mothers

to report their own current heights and weights (18), the number of children born since the birth of the PROBIT child, and the number of months they breastfed each subsequent child. Outcome assessment at 11.5 y postpartum From January 2008 through December 2010, we conducted follow-up visits (19). One study pediatrician collected all outcome measures at each of the 23 smaller polyclinics, and 2 pediatricians shared the workload at each of the 8 larger sites. Pediatricians obtained all measurements by using standard techniques. To ensure high-quality, standardized measurements (20), we held a 3-d training workshop for participating pediatricians, which included lectures and practical sessions. We provided the pediatricians with a manual of procedures and training DVD. All materials were translated into Russian and back-translated into English. We held refresher (retraining) workshops every 6 mo thereafter. Pediatricians measured maternal standing height by using a research stadiometer (Medtechnika) and weight by using a Tanita TBF 300GS body fat analyzer (Tanita Corporation Inc). If a mother did not attend the in-person visit, the pediatrician attempted to obtain self-reported weight and height via telephone. We calculated BMI (in kg/m2) and defined obesity as BMI $30.0 (21). The Tanita analyzer also measured bioelectrical impedance from leg to leg and provided estimates of the percentage of body fat, fat mass, and fat-free mass. We calculated the fat mass index (FMI; in kg/m2) and fat-free mass index (FFMI; in kg/m2). We applied a threshold of body fat $40% on the basis of correlations with obesity in US surveillance data (22). Pediatricians obtained a single measurement of maternal systolic BP and diastolic BP on the right arm by using the OMRON 705IT oscillometric device and an appropriate sized cuff (ie, the bladder of the cuff encircled $80% of the upper arm but not .100%). All mothers were sitting quietly, with their feet flat on the floor. Pediatricians also asked mothers whether they ever had a diagnosis of hypertension, type 2 diabetes, or gestational diabetes. Because of small numbers of reported outcomes, we did not further study type 2 (0.3%) or gestational diabetes (0.1%). We defined hypertension as measured systolic BP $140 mmHg, diastolic BP $90 mmHg (23), or a maternal report of hypertension diagnosis or treatment. Because confirmed hypertension generally requires elevated BP measurements on $2 occasions, we secondarily evaluated diagnosed hypertension as a maternal report of hypertension diagnosis or treatment and disregarded the single research measure of BP. Quality assurance Once data-collection began, each pediatrician was observed by 1 of 5 trained Belarussian monitors, who repeated key measurements made during 1 d of study visits per pediatrician. The monitors did not identify any problems with maternal outcome measurements. Once the data collection was complete, we conducted remeasurement audit visits to assess interobserver reproducibility, which was an important step because pediatricians were not blinded to the randomization status. So that all mothers seen in follow-up were eligible for selection, repeated measurements were carried out on randomly selected mothers at

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

Perhaps the most important limitation of previous observational studies is that both obesity and breastfeeding are strongly socially patterned (14, 15). Substantial differences between mothers who do and do not choose to breastfeed may confound results from observational studies. Thus, it is difficult to determine whether observational associations of breastfeeding duration with maternal adiposity and other health outcomes are causal or have alternative explanations. The Promotion of Breastfeeding Intervention Trial (PROBIT) trial, which was the largest breastfeeding trial ever conducted, successfully randomized units of medical care (maternity hospitals and their affiliated outpatient clinics) to a breastfeedingpromotion intervention (experimental group) or usual care (control group). The PROBIT intervention resulted in large contrasts in breastfeeding duration and exclusivity between the 2 groups (16). In the current study, we examined maternal adiposity outcomes in more than 11,800 PROBIT mothers evaluated at 11.5 y postpartum. We hypothesized that mothers randomly assigned to receive the breastfeeding intervention would have lower weight, body fat, and BP than would women randomly assigned to usual care.

1049

1050

OKEN ET AL

each of the polyclinics after completion of the primary data collection, which was an average of 1.3 y (range: 0.2–2.2 y) after the initial study visit. Auditors were blinded to initial measurements (but not to the experimental or control status). Correlations for 143 repeat measures were excellent for height (Spearman’s correlation coefficient = 0.99), weight (r = 0.93), and body fat (r = 0.83) and very good for systolic BP (r = 0.53) and diastolic (r = 0.57) BP.

The PROBIT III study protocol was approved by Human Subjects Committees at Harvard Pilgrim Health Care, the Montreal Children’s Hospital, and the Law and Ethics Committee at the University of Bristol. All procedures were conducted in accordance with ethical standards (24). All mothers provided written informed consent in Russian at enrollment and outcome visits. Statistics For each of the measured outcomes, we calculated the intraclass correlation (ICC) (20). The ICC is a statistic that can be used to quantify the degree to which observations within a cluster differ from those between clusters. An ICC close to 0 indicates a low degree of clustering (ie, most of the variation occurs within polyclinics), whereas an ICC close to 1 suggests a high degree of clustering (ie, most of the variation occurs between polyclinics). We calculated cluster-adjusted HRs for discontinuing any and exclusive breastfeeding with the SAS phreg procedure (SAS version 9.3; SAS Institute). In a modified intention-to-treat analysis (ie, without imputation for missing outcome measures), we evaluated the effect of the breastfeeding intervention on maternal BMI, body composition, and systolic and diastolic BPs by using the MIXED procedure in the SAS program (SAS version 9.3; SAS Institute) and on maternal obesity, excess body fat, and hypertension by using the GLIMMIX procedure in the SAS program. We accounted for clustering by clinic in all analyses by adding a random effect (for study site). We carried out additional analyses in which we also adjusted for maternal baseline characteristics and for polyclinic location (urban compared with rural; East compared with West Belarus). We excluded from each analysis women missing the outcome measurement. Because it was possible that the effect of the experimental intervention may have waned by 11 y postpartum, we also examined associations with self-reported maternal BMI at 6.5 y postpartum. In addition, we conducted sensitivity analyses as follows. We performed observational analyses (ie, on the basis of actual breastfeeding duration and exclusivity, disregarding randomized treatment allocation), in which we estimated the effects of the duration of any breastfeeding and exclusive breastfeeding (both continuous and in categories) on the same outcomes, accounting for clustering and baseline characteristics by using the same statistical procedures. We also conducted instrumental variable analyses to estimate the causal effect of the duration of any breastfeeding and exclusive breastfeeding by using the randomly assigned PROBIT intervention group as an instrument, akin to the use of instrumental variables to adjust for noncompliance in randomized controlled trials (25). We performed instrumental

RESULTS

Of the 17,046 mothers and their infants originally randomly assigned, 13,882 children attended PROBIT III visits, of whom 3 children were unable to provide verifiable assent, which left 13,879 eligible children (19, 31). We had information on one or more adiposity measures for 11,867 mothers (69.6%). The median follow-up time for mothers was 11.6 y (range: 10.3–13.9 y). We obtained BMI in 11,751 women (9491 women had measured weight, and the remainder was self-reported), body fat measures from 9392 mothers, and BP measures from 9383 mothers (Figure 1). Baseline characteristics of included women were similar across the randomly assigned groups (Table 1). Small differences in some categorical variables were attributable to the cluster-based random assignment and incomplete follow-up. The intervention substantially increased the breastfeeding duration and exclusivity; eg, compared with mothers in the control arm, intervention mothers were 6 times more likely to be exclusively breastfeeding at 3 mo postpartum (44.5% compared with 7.1%, respectively) and were breastfeeding to any degree at higher rates throughout the postpartum period. Risk of discontinuing breastfeeding was substantially lower in intervention compared with control mothers [cluster-adjusted HR for any breastfeeding 0.67 (95% CI: 0.55, 0.81); and for exclusive breastfeeding, HR was 0.25 (95% CI: 0.18, 0.36)]. At 11.5 y postpartum, mean (6SD) maternal BMI was 26.5 6 5.5, FMI was 9.4 6 4.2, FFMI was 17.4 6 2.0, percentage of body fat was 33.6 6 8.3%, systolic BP was 124.6 6 14.6 mm Hg, and diastolic BP was 77.7 6 10.4 mm Hg; 22.5% of mothers were obese (BMI $30). There was a low degree of within-polyclinic clustering for all outcomes (all ICCs #0.05; Table 2). Measures of adiposity and BP were lower in women who had been randomly assigned to the intervention group (Table 2). However, in an intention-to-treat analysis that accounted for clustering, we observed only weak evidence for any differences in intervention compared with control mothers for BMI (mean difference: 20.27; 95% CI: 20.91, 0.37), FMI (20.23; 95% CI: 20.64, 0.17), FFMI (20.05; 95% CI: 20.27, 0.16), percentage of body fat (20.49%; 95% CI: 21.25%, 0.27%), systolic BP (20.81 mm Hg; 95%

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

Ethics

variable estimation by using the generalized 2-stage least-squares estimator (26) implemented in the xtivreg command in Stata software (version 11.2; Stata Corp), while accounting for clustering by study site. We also examined associations of breastfeeding consistent with recommendations of the American Academy of Pediatrics (27) and the WHO (28) (ie, exclusive breastfeeding for 6 mo plus continued breastfeeding for $12 mo). Not all of the mothers who were randomly assigned at birth attended the 11.5-y postpartum visit. In an additional sensitivity analysis, we investigated whether an incomplete follow-up influenced the results by undertaking multiple imputation to generate plausible values of missing 11.5-y maternal outcomes (29, 30). In this analysis, we imputed data for all of the 17,046 mothers who were randomly assigned at birth. We used SAS imputations (Proc MI) to impute 5 values for each missing observation and combined multivariable modeling estimates by using Proc MI ANALYZE in the SAS program.

BREASTFEEDING TRIAL AND MATERNAL ADIPOSITY

1051

CI: 23.33, 1.71 mm Hg), or diastolic BP (21.09 mm Hg; 95% CI: 22.43, 0.25 mm Hg). Results were attenuated markedly toward the null when further adjusting for recruitment hospital/polyclinic location, maternal age and education, smoking, and the number of children in the household (Table 2). Although mean BMI and obesity prevalence were higher in mothers with measured weight than in mothers who self-reported, differences between intervention and control mothers were similar when they were based on measured weight than when they were based on all data (self-reported and measured), and results of regression analyses of BMI outcomes were almost identical when we limited analyses to women with measured weight (Table 2). Evidence of any interventioncompared with control-group effect was similarly weak for odds of obesity, high body fat, and any or diagnosed hypertension (Table 2). At 6.5 y postpartum, we also saw little evidence for lower self-reported weight (mean difference: 0.35 kg; 95% CI: 20.45, 1.17 kg), BMI (0.07; 95% CI: 20.27, 0.42), or obesity (OR: 0.93; 95% CI: 0.77, 1.12) in intervention compared with control mothers.

We performed observational analyses by examining the effects of each mother’s achieved breastfeeding duration and exclusivity, regardless of her treatment allocation (Table 3). There was no evidence for an association of exclusive breastfeeding with lower maternal BMI, body fat, or systolic BP or of any breastfeeding with BMI or body fat (Table 3). There was some indication of lower BP with any breastfeeding, although the association was strongest in the group with any breastfeeding $6 to ,9 mo and less strong with longer durations (Table 3). Results for weight and FMI were similar to those for the percentage of body fat, and results for diastolic BP were similar to those for systolic BP (data not shown). Exclusive breastfeeding for $6 mo followed by any breastfeeding for $12 mo, as recommended by the American Academy of Pediatrics and WHO, was paradoxically associated with somewhat higher maternal BMI (0.47; 95% CI: 20.31, 1.25) and percentage of body fat (1.30%; 95% CI: 0.01, 2.56%) (Table 3). Intervention women reported a substantially longer lactation duration than that of control women not only for the children enrolled in PROBIT (Table 3) but also for subsequent children

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

FIGURE 1. Flow of participants in the PROBIT study. *Numbers of deaths and mother-infant pairs with 12 mo of completed follow-up differed slightly from those previously reported because of ongoing data cleaning. BP, blood pressure; PROBIT, Promotion of Breastfeeding Intervention Trial.

1052

OKEN ET AL

TABLE 1 Baseline characteristics of 11,867 mothers enrolled in the PROBIT study1 Intervention (n = 6321)

Control (n = 5546)

Mean maternal age (y) Maternal education (%) Incomplete secondary Complete secondary Advanced secondary/some university University graduate Maternal smoking during pregnancy (%) Older children living in household (%) 0 1 $2 Male infant (%) Polyclinic location (%) East, urban East, rural West, urban West, rural Mean infant birth weight (g)

25.0 6 5.02

25.1 6 4.9

3.9 34.3 48.3 13.5 2.5

2.5 28.6 56.0 12.9 1.3

58.9 33.5 7.6 50.8

54.7 36.4 8.9 52.0

34.8 15.2 30.5 19.4 3439 6 417

28.4 14.4 19.6 37.6 3447 6 423

1 2

PROBIT, Promotion of Breastfeeding Intervention Trial. Mean 6 SD (all such values).

(see Supplemental Table 1 under “Supplemental data” in the online issue). When we used cumulative duration of lactation (including the PROBIT child and any subsequent children) as the exposure in observational analyses, we did not see any evidence for an effect of cumulative lifetime lactation for any outcome [eg 20.01 (95% CI: 20.05, 0.02) per month of exclusive breastfeeding and 0.001 (95% CI: 20.01, 0.01) per month of any breastfeeding; see Supplemental Table 1 under “Supplemental data” in the online issue]. Instrumental variable estimates by using the randomly assigned PROBIT intervention group as the instrument were in the same direction as observational estimates for both the duration of exclusive and any breastfeeding; however, all estimates had wide CIs including the null (Table 4). In an analysis that used multiply imputed data with modeling of outcomes for all 17,046 mothers randomly assigned at birth, estimated effects of the randomized intervention (see Supplemental Table 2 under “Supplemental data” in the online issue) and observational associations with breastfeeding (see Supplemental Table 3 under “Supplemental data” in the online issue) were virtually identical to analyses that were based on observed outcome data.

DISCUSSION

Contrary to the hypothesis, we did not show that women who initiated breastfeeding and were randomly assigned to a successful intervention to promote longer breastfeeding duration and exclusivity had significantly lower adiposity or BP at 11.5 y postpartum. A number of factors that predict postpartum weight and BP also strongly predict the initiation and duration of breastfeeding, including prepregnancy weight and gestational weight gain (32). In our study, confounding was minimized by the stringent random-allocation procedure, which produced intervention and control groups with similar baseline characteristics (16). We were also able to account for concurrently

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

Maternal, family, and offspring characteristics at baseline

measured baseline characteristics in our analyses. Breastfeeding was assessed repeatedly during infancy, not recalled remotely, and exposure and outcome measures were shown to be valid and reproducible. In agreement with our findings, a 2009 systematic review of observational studies concluded that the effect of lactation on return to prepregnancy weight was negligible, and many other factors had larger effects (33). Methodologic challenges in reviewed studies included accurately measuring weight change, adequately controlling for numerous covariables (including weight entering pregnancy and weight gained during pregnancy), and precisely quantifying the exclusivity and duration of breastfeeding. In large observational studies that have evaluated maternal outcomes over years or decades after the pregnancies or mothers, results have similarly been inconsistent. In 570 US women in the Project Viva longitudinal cohort study, a longer duration of lactation was not associated with lower weight retention, BMI, or blood cardiometabolic risk markers at 3 y postpartum (34). In w4000 US women in the Nurses’ Health Study II, lactation to any degree was associated with an w1-kg greater weight increase over the period 1989–1993 than in women who became pregnant but did not breastfeed (35). In both study populations, women who breastfed for a longer duration were less likely to be overweight or to smoke before pregnancy. In .700,000 women with a mean age of 57 y who participated in the UK Million Women Study, BMI was 0.13 (95% CI: 20.11, 20.13) lower for every 6 mo lifetime lactation (13), which was a small effect that was, nevertheless, excluded by our result in which the lower confidence limit of the estimate for 6 mo of cumulative breastfeeding was 20.06. Prepregnancy BMI was not assessed, although women who had breastfed for a longer duration were of a higher socioeconomic status. In the WHO Multicentre Growth Reference Study, which included 1743 women from Brazil, Ghana, India, Norway, Oman, and the United States, breastfeeding frequency did not vary by prepregnancy BMI, and the lactation duration and intensity explained little of the variation in maternal weight change to 24 mo postpartum (36). Lactation could theoretically influence body composition or metabolism without a direct effect on weight. Two studies showed that lactation $3 mo per child (compared with none) was associated with lower visceral adiposity, although no effect was observed on overall adiposity (4, 5). In postpartum women with recent gestational diabetes, a higher intensity of lactation was associated with improved fasting glucose and lower insulin concentrations at 6–9 wk postpartum (12). In PROBIT, we did not assess visceral adiposity or measure glucose tolerance, and the number of women who developed type 2 diabetes by 11.5 y postpartum was too small to permit analysis. Physiologic changes associated with breastfeeding also may benefit maternal BP (37, 38). Only a few long-term studies have examined these associations. In 4700 women enrolled in NHANES III (7), breastfeeding .1 compared with #1 mo was associated with a reduced odds of the metabolic syndrome (OR: 0.78; 95% CI: 0.61, 0.99), of which elevated BP is one component. In the Women’s Health Initiative, women with a lifetime history of $12 mo lactation were less likely to have hypertension (OR: 0.88; P , 0.001) than were women who had never breastfed. In the Nurses’ Health Study II, never or curtailed lactation was associated with increased risk of incident maternal

DBP, diastolic blood pressure; FFMI, fat-free mass index; FMI, fat mass index; ICC, intraclass correlation; PROBIT, Promotion of Breastfeeding Intervention Trial; SBP, systolic blood pressure. All values are means; 95% CIs in parentheses. Results are from multilevel linear regression analyses. 3 Results are from multilevel logistic regression analyses. 4 Location of polyclinic, maternal age, maternal education, number of children in the household, and smoking during pregnancy. 2

0.96 0.89 0.88 0.81 1.13) 1.01) 1.12) 1.37) 0.90 0.84 0.85 0.79 n (%) — — — — — — — 1323 (24.0) 1152 (25.3) 1657 (35.0) 1191 (22.6) n (%) — — — — — — — 1320 (21.1) 1082 (22.4) 1512 (30.3) 1086 (17.2) 20.06 (20.62, 0.48) 20.10 (20.59, 0.39) 20.08 (20.43, 0.27) 0.01 (20.17, 0.18) 20.17 (20.80, 0.45) 20.23 (22.71, 2.25) 20.74 (22.02, 0.53) — — — — 0.37) 0.27) 0.17) 0.16) 0.27) 1.71) 0.25) (20.91, (20.90, (20.64, (20.27, (21.25, (23.33, (22.43, — — — — 20.27 20.31 20.23 20.05 20.49 20.81 21.09 0.02 0.02 0.01 0.02 0.01 0.05 0.03 — — — — 5.6 5.7 4.3 1.9 8.3 15.3 10.6 26.7 6 27.0 6 9.6 6 17.4 6 33.9 6 125.2 6 78.2 6 — — — — 5501 4574 4555 4555 4560 4600 4600 — — — — 5.5 5.6 4.2 2.0 8.3 13.8 10.2 6 6 6 6 6 6 6 26.3 26.6 9.3 17.3 33.3 123.9 77.3 — — — — 6250 4851 4819 4819 4832 4783 4783 — — — — Any BMI (kg/m2) Measured weight BMI (kg/m2) FMI (kg/m2) FFMI (kg/m2) Percentage of body fat SBP (mm Hg) DBP (mm Hg) Obesity (BMI $30 kg/m2) Body fat $40% Any hypertension Diagnosed hypertension

1

— — — — — — — (0.78, (0.75, (0.65, (0.45, — — — — — — — (0.71, (0.70, (0.64, (0.46,

1.19) 1.06) 1.19) 1.44)

Further adjusted for baseline factors4 Cluster adjusted Control Intervention Further adjusted for baseline factors4 Cluster adjusted n Outcome

n

Mean 6 SD

Mean 6 SD

ICC

Difference2 Control Intervention

1053

hypertension compared with that for the recommended $6 mo exclusive or $12 mo total lactation per child (39). In the Coronary Artery Risk Development in Young Adults study, mothers who had breastfed had lower systolic BP (101.6 compared with 105.3 mm Hg) and diastolic BP (64.9 compared with 66.4 mm Hg) that did mothers who had not breastfed, although differences did not persist after adjusted analyses (6). A major limitation of the observational breastfeeding literature is that unmeasured or residual confounding may well explain the findings, because the exposures have been based on maternal choices about or success with breastfeeding. The Coronary Artery Risk Development in Young Adults study, which has followed a cohort of young women longitudinally, has the advantage of being able to account for differences in measured prepregnancy, as well as prenatal and postpartum, factors that may confound relations (6, 40). However, inadequate control for factors that predict both breastfeeding success and later outcomes remains a plausible alternative explanation for observed associations. Despite the large sample size, it is possible that our study was underpowered to detect a small but real difference in measured outcomes. Confidence limits were widened by accounting for the clustering of both the intervention (mothers were clustered within the polyclinic and, thus, would tend to have more-similar characteristics) and measurement (a tendency for measurements made by a polyclinic pediatrician to be more similar than measurements between clinics). (41). However, intraclass correlation coefficients for all outcomes of interest were low. Furthermore, any true difference between groups was likely to have been small (eg, the OR for obesity was 0.96, and there was a 0.23mm Hg lowering of systolic BP. It is possible that breastfeeding has a shorter-term effect on maternal adiposity and BP that is no longer present at 11.5 y postpartum, although we also saw no evidence for an effect on weight at 6.5 y postpartum. Our breastfeeding-promotion intervention did not result in complete separation between groups; some women in the control group exclusively breastfed .3 mo, and many women in the intervention group discontinued exclusive breastfeeding. Therefore, even if longer lactation duration is associated with modest improvements in maternal adiposity, a breastfeeding-promotion intervention is not likely to substantially improve maternal obesity or BP on a population level. All women in the cohort initiated breastfeeding, and thus, our study could not evaluate outcomes in women who never breastfed. We obtained a single measure of systolic and diastolic BP. Repeat measures may be more reflective of basal BP (23). Leg-leg bioimpedance methods may underestimate body fat compared with other methods such as a 4-compartment model, although comparisons between different methods suggested that this method has reasonable validity and is able to accurately rank individuals and groups (42, 43). Any measurement error in continuous outcome variables would have added to the residual variance and, therefore, decreased power. Although the polyclinic pediatricians could not be blinded to the intervention or control status, the primary measures (weight, bioimpedance, and BP) were assessed by using calibrated, automated equipment and, thus, should not have been operator dependent. Furthermore, maternal outcomes were not a primary study hypothesis, and the duplicate audit visits provided evidence of validity. Losses to follow-up and incomplete data were potential sources of bias, but

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

TABLE 2 Associations of experimental intervention with maternal adiposity and blood pressure at 11.5 y postpartum in 11,867 women in the PROBIT study1

OR (95% CI)3

BREASTFEEDING TRIAL AND MATERNAL ADIPOSITY

1054

OKEN ET AL

TABLE 3 Observational associations of the duration of breastfeeding with maternal adiposity and blood pressure at 11.5 y postpartum in 11,867 women in the PROBIT study1 Duration of breastfeeding

Intervention (n = 6321)

Control (n = 5546)

BMI2,3

Percentage of body fat3

SBP3

kg/m2

mm Hg

55.5 37.9 6.6 — — 3.0 6 1.8

92.9 6.3 0.8 — — 1.1 6 1.2

0 (referent) 20.25 (20.51, 0.01) 0.37 (20.17, 0.91) 0.68 0.04 20.02 (20.09, 0.04)

31.7 25.1 15.0 8.7 19.4 — — 6.5 6 4.2 3.2

40.5 25.7 13.6 8.3 11.9 — — 5.4 6 3.9 0.2

0 (referent) 20.05 (20.32, 0.21) 20.01 (20.33, 0.32) 20.05 (20.44, 0.35) 0.10 (20.22, 0.41) 0.61 0.93 20.001 (20.02, 0.02) 0.47 (20.31, 1.25)

0 (referent) 20.44 (20.87, 20.01) 0.41 (20.48, 1.30) 0.45 0.06 20.05 (20.16, 0.05)

0 (referent) 20.39 (21.16, 0.38) 0.40 (21.16, 1.97) 0.75 0.48 20.06 (20.26, 0.14)

0 20.17 20.05 20.20 0.06

0 (referent) 0.10 (20.67, 0.87) 21.19 (22.12, 20.25) 20.71 (21.85, 0.42) 20.57 (21.48, 0.33) 0.05 0.06 20.06 (20.13, 0.01) 0.07 (22.16, 2.31)

(referent) (20.62, 0.28) (20.59, 0.49) (20.86, 0.46) (20.46, 0.59) 0.90 0.89 20.01 (20.05, 0.03) 1.30 (0.01, 2.60)

1

PROBIT, Promotion of Breastfeeding Intervention Trial; SBP, systolic blood pressure. Calculated by using any (measured or reported) weight and height. 3 All values are cluster- and baseline-adjusted mean differences; 95% CIs in parentheses. Results are from random-effect linear regression analyses. 2

the analysis that was based on multiple imputation provided some reassurance against such biases. We did not collect information on maternal prepregnancy BMI, gestational weight gain, or BP before or during pregnancy or on menopausal status at the outcome assessment, but the randomization procedure should have balanced these factors across intervention and control sites. Because the study was in Belarus, all participants were white. However, PROBIT includes an economically diverse population, and fewer than 2% of eligible mothers refused initial participation in PROBIT (16). Belarus resembles developed Western countries in both basic health services and sanitary conditions, although obesity rates are lower than in many other settings.

In conclusion, by using data from a large, cluster-randomized controlled trial that minimized confounding, we did not see evidence that the longer breastfeeding duration and greater exclusivity achieved by the intervention led to an important lowering of maternal adiposity or BP. Breastfeeding has many advantages, but it is unlikely that interventions to promote a longer duration of exclusive breastfeeding will confer important long-term benefits for maternal adiposity or BP. We acknowledge Matthew Gillman and George Davey Smith for assistance with funding and data collection, Janet Rich-Edwards for input regarding data collection, and Jennifer Thompson and Sheryl Rifas-Shiman for assistance with data analysis. We are grateful to cohort members and their parents

TABLE 4 Instrumental variable estimates of the causal effect of breastfeeding duration on maternal adiposity and blood pressure at 11.5 y postpartum1 BMI

SBP2

Percentage of body fat

kg/m2 Duration of exclusive breastfeeding ,3 mo $3 but ,6 mo $6 mo Duration of any breastfeeding ,3 mo $3 but ,6 mo $6 but ,9 mo $9 but ,12 mo $12 mo Exclusive breastfeeding $6 mo with any breastfeeding $12 mo 1 2

mm Hg

0 (referent) 20.35 (21.97, 1.28) 20.03 (22.26, 2.20)

0 (referent) 20.66 (22.42, 1.10) 20.39 (23.05, 2.27)

0 (referent) 20.74 (28.62, 7.14) 20.91 (212.74, 10.93)

0 (referent) 21.90 (28.35, 4.56) 20.51 (23.19, 2.16) 20.72 (23.02, 1.57) 20.03 (20.85, 0.79) 20.54 (25.52, 4.44)

0 22.61 20.76 21.24 20.30 21.53

0 22.91 21.60 0.02 20.39 21.85

(referent) (212.21, 6.98) (24.11, 2.58) (23.93, 1.44) (21.32, 0.71) (27.19, 4.12)

(referent) (233.57, 27.74) (214.68, 11.47) (210.79, 10.84) (24.08, 3.31) (225.0, 21.3)

All values are cluster- and baseline-adjusted mean differences; 95% CIs in parentheses. Results are from random-effect linear regression analyses. SBP, systolic blood pressure.

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

Duration of exclusive breastfeeding (%) ,3 mo $3 but ,6 mo $6 mo P-trend P-heterogeneity Breastfeeding (effect/mo) Duration of any breastfeeding (%) ,3 mo $3 but ,6 mo $6 but ,9 mo $9 but ,12 mo $12 mo P-trend P-heterogeneity Breastfeeding (effect/mo) Exclusive breastfeeding $6 mo with any breastfeeding $12 mo

BREASTFEEDING TRIAL AND MATERNAL ADIPOSITY

17.

18.

19.

20.

21.

REFERENCES 1. Stuebe AM, Rich-Edwards JW. The reset hypothesis: lactation and maternal metabolism. Am J Perinatol 2009;26:81–8. 2. Dewey KG, Cohen RJ, Brown KH, Rivera LL. Age of introduction of complementary foods and growth of term, low-birth-weight, breast-fed infants: a randomized intervention study in Honduras. Am J Clin Nutr 1999;69:679–86. 3. Dewey KG, Cohen RJ, Brown KH, Rivera LL. Effects of exclusive breastfeeding for four versus six months on maternal nutritional status and infant motor development: results of two randomized trials in Honduras. J Nutr 2001;131:262–7. 4. McClure CK, Catov J, Ness R, Schwarz EB. Maternal visceral adiposity by consistency of lactation. Matern Child Health J 2012;16:316– 21. 5. McClure CK, Schwarz EB, Conroy MB, Tepper PG, Janssen I, SuttonTyrrell KC. Breastfeeding and subsequent maternal visceral adiposity. Obesity (Silver Spring) 2011;19:2205–13. 6. Gunderson EP, Lewis CE, Wei GS, Whitmer RA, Quesenberry CP, Sidney S. Lactation and changes in maternal metabolic risk factors. Obstet Gynecol 2007;109:729–38. 7. Cohen A, Pieper CF, Brown AJ, Bastian LA. Number of children and risk of metabolic syndrome in women. J Womens Health (Larchmt) 2006;15:763–73. 8. Baker JL, Gamborg M, Heitmann BL, Lissner L, Sorensen TI, Rasmussen KM. Breastfeeding reduces postpartum weight retention. Am J Clin Nutr 2008;88:1543–51. 9. Schwarz EB, Brown JS, Creasman JM, Stuebe A, McClure CK, Van Den Eeden SK, Thom D. Lactation and maternal risk of type 2 diabetes: a population-based study. Am J Med 2010;123:863.e1–6. 10. Schwarz EB, McClure CK, Tepper PG, Thurston R, Janssen I, Matthews KA, Sutton-Tyrrell K. Lactation and maternal measures of subclinical cardiovascular disease. Obstet Gynecol 2010;115:41–8. 11. Schwarz EB, Ray RM, Stuebe AM, Allison MA, Ness RB, Freiberg MS, Cauley JA. Duration of lactation and risk factors for maternal cardiovascular disease. Obstet Gynecol 2009;113:974–82. 12. Gunderson EP, Hedderson MM, Chiang V, Crites Y, Walton D, Azevedo RA, Fox G, Elmasian C, Young S, Salvador N, et al. Lactation intensity and postpartum maternal glucose tolerance and insulin resistance in women with recent GDM: the SWIFT cohort. Diabetes Care 2012;35: 50–6. 13. Bobrow KL, Quigley MA, Green J, Reeves GK, Beral V. Persistent effects of women’s parity and breastfeeding patterns on their body mass index: results from the Million Women Study. Int J Obes (Lond) 2013;37:712–7. 14. Liu J, Smith MG, Dobre MA, Ferguson JE. Maternal obesity and breast-feeding practices among white and black women. Obesity (Silver Spring) 2010;18:175–82. 15. Phares TM, Morrow B, Lansky A, Barfield WD, Prince CB, Marchi KS, Braveman PA, Williams LM, Kinniburgh B. Surveillance for disparities in maternal health-related behaviors–selected states, Pregnancy Risk Assessment Monitoring System (PRAMS), 2000-2001. MMWR Surveill Summ 2004;53:1–13. 16. Kramer MS, Chalmers B, Hodnett ED, Sevkovskaya Z, Dzikovich I, Shapiro S, Collet JP, Vanilovich I, Mezen I, Ducruet T, et al. Promotion

22.

23.

24.

25. 26.

27.

28.

29.

30. 31.

32.

33.

34.

35.

36.

of Breastfeeding Intervention Trial (PROBIT): a randomized trial in the Republic of Belarus. JAMA 2001;285:413–20. WHO/UNICEF. Protecting, promoting and supporting breastfeeding: the special role of maternity services. Geneva, Switzerland: World Health Organization, 1989. Patel R, Martin RM, Kramer MS, Oken E, Bogdanovich N, Matush L, Smith GD, Lawlor DA. Familial associations of adiposity: findings from a cross-sectional study of 12,181 parental-offspring trios from Belarus. PLoS ONE 2011;6:e14607. Patel R, Oken E, Bogdanovich N, Matush L, Sevkovskaya Z, Chalmers B, Hodnett ED, Vilchuck K, Kramer MS, Martin RM. Cohort profile: the Promotion of Breastfeeding Intervention Trial (PROBIT). Int J Epidemiol 2013 (Epub ahead of print 8 May 2013). Guthrie LB, Oken E, Sterne JA, Gillman MW, Patel R, Vilchuck K, Bogdanovich N, Kramer MS, Martin RM. Ongoing monitoring of data clustering in multicenter studies. BMC Med Res Methodol 2012;12:29. National Heart Lung and Blood Institute. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. Bethesda, MD: National Institutes of Health, US Department of Health and Human Services, 1998. Publication No. 98-4083. Available from: http://www.nhlbi.nih.gov/guidelines/obesity/ob_gdlns.htm (cited 1 August 2013). Flegal KM, Shepherd JA, Looker AC, Graubard BI, Borrud LG, Ogden CL, Harris TB, Everhart JE, Schenker N. Comparisons of percentage body fat, body mass index, waist circumference, and waist-stature ratio in adults. Am J Clin Nutr 2009;89:500–8. National High Blood Pressure Education Program. The seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-7). Washington, DC: National Heart, Lung and Blood Institute, 2004. Available at: http:// www.nhlbi.nih.gov/guidelines/hypertension/jnc7full.pdf (cited 19 December 2011). World Medical Association. World Medical Association Declaration of Helsinki. recommendations guiding physicians in biomedical research involving human subjects. JAMA 1997;277:925–6. Angrist JD, Imbens GW, Rubin DB. Identification of causal effects using instrumental variables. J Am Stat Assoc 1996;91:444–72. Balestra P, Varadharajan-Krishnakumar J. Full information estimations of a system of simultaneous equations with error component structure. Econometric Theory 1987;3. Gartner LM, Morton J, Lawrence RA, Naylor AJ, O’Hare D, Schanler RJ, Eidelman AI. Breastfeeding and the use of human milk. Pediatrics 2005;115:496–506. World Health Organization. Infant and young child nutrition: global strategy on infant and young child feeding. Geneva, Switzerland: World Health Organization, 2002. Available at: http://apps.who.int/gb/ archive/pdf_files/WHA55/ea5515.pdf (cited 28 December 2011). Horton NJ, Kleinman KP. Much ado about nothing: a comparison of missing data methods and software to fit incomplete data regression models. Am Stat 2007;61:79–90. Rubin DB. Multiple imputation for non response in surveys. New York, NY: J Wiley & Sons, 1987. Martin RM, Patel R, Kramer MS, Guthrie L, Vilchuck K, Bogdanovich N, Sergeichick N, Gusina N, Foo Y, Palmer T, et al. Effects of promoting longer-term and exclusive breastfeeding on adiposity and insulin-like growth factor-I at age 11.5 years: a randomized trial. JAMA 2013;309:1005–13. Wojcicki JM. Maternal prepregnancy body mass index and initiation and duration of breastfeeding: a review of the literature. J Womens Health (Larchmt) 2011;20:341–7. Ip S, Chung M, Raman G, Trikalinos TA, Lau J. A summary of the Agency for Healthcare Research and Quality’s evidence report on breastfeeding in developed countries. Breastfeed Med 2009;4(suppl 1): S17–30. Stuebe AM, Kleinman K, Gillman MW, Rifas-Shiman SL, Gunderson EP, Rich-Edwards J. Duration of lactation and maternal metabolism at 3 years postpartum. J Womens Health (Larchmt) 2010;19:941–50. Sichieri R, Field AE, Rich-Edwards J, Willett WC. Prospective assessment of exclusive breastfeeding in relation to weight change in women. Int J Obes Relat Metab Disord 2003;27:815–20. Onyango AW, Nommsen-Rivers L, Siyam A, Borghi E, de Onis M, Garza C, Lartey A, Baerug A, Bhandari N, Dewey KG, et al. Postpartum weight change patterns in the WHO Multicentre Growth Reference Study. Matern Child Nutr 2011;7:228–40.

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

for their generous participation in the study, polyclinic pediatricians who examined all children and ensured a very high follow-up rate, and the dedication of Belarussian team members who ensured that this trial could start and who continue to support and guide the PROBIT study through all stages. The authors’ responsibilities were as follows—EO, RMM, and MSK: study conception and design; EO, RP, LBG, KV, NB, NS, MSK, and RMM: acquisition of data; EO, RP, LBG, RMM, and MSK: analysis and interpretation of data; EO, LBG, RMM, TMP, and MSK: statistical analysis; RP, LBG, KV, NB, and NS: administrative, technical, or material support; EO and RMM: full access to all study data and responsibility for the integrity of data and the accuracy of the data analysis; and all authors: critical revision of the manuscript. EO gave an invited talk for Nestle Nutrition Institute on secular trends in birth weight in 2011. RMM gave an invited talk for the Nestle Nutrition Institute in 2010 on the role of the insulin-like growth factor system in growth and chronic disease risk. RP, LBG, KV, NB, NS, TMP, and MSK declared no conflicts of interest.

1055

1056

OKEN ET AL syndrome in women of reproductive age according to gestational diabetes mellitus status: a 20-Year prospective study in CARDIA (Coronary Artery Risk Development in Young Adults). Diabetes 2010;59: 495–504. 41. Kramer MS, Martin RM, Sterne JA, Shapiro S, Dahhou M, Platt RW. The double jeopardy of clustered measurement and cluster randomisation. BMJ 2009;339:b2900. 42. Ritchie JD, Miller CK, Smiciklas-Wright H. Tanita foot-to-foot bioelectrical impedance analysis system validated in older adults. J Am Diet Assoc 2005;105:1617–9. 43. Boneva-Asiova Z, Boyanov MA. Body composition analysis by legto-leg bioelectrical impedance and dual-energy X-ray absorptiometry in non-obese and obese individuals. Diabetes Obes Metab 2008; 10:1012–8.

Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on March 10, 2015

37. Mezzacappa ES, Kelsey RM, Katkin ES. Breast feeding, bottle feeding, and maternal autonomic responses to stress. J Psychosom Res 2005;58: 351–65. 38. Light KC, Smith TE, Johns JM, Brownley KA, Hofheimer JA, Amico JA. Oxytocin responsivity in mothers of infants: a preliminary study of relationships with blood pressure during laboratory stress and normal ambulatory activity. Health Psychol 2000;19: 560–7. 39. Stuebe AM, Schwarz EB, Grewen K, Rich-Edwards JW, Michels KB, Foster EM, Curhan G, Forman J. Duration of lactation and incidence of maternal hypertension: a longitudinal cohort study. Am J Epidemiol 2011;174:1147–58. 40. Gunderson EP, Jacobs DR Jr, Chiang V, Lewis CE, Feng J, Quesenberry CP Jr, Sidney S. Duration of lactation and incidence of the metabolic

Effects of an intervention to promote breastfeeding on maternal adiposity and blood pressure at 11.5 y postpartum: results from the Promotion of Breastfeeding Intervention Trial, a cluster-randomized controlled trial.

Differences between mothers who do and do not succeed in breastfeeding are likely to confound associations of lactation with later maternal adiposity...
436KB Sizes 0 Downloads 0 Views