点击显示 收起
1 From the Departments of Pediatrics (MSK and RWP) and Epidemiology and Biostatistics (MSK, RWP, J-PC, and SS), McGill University Faculty of Medicine, Montreal, PQ, Canada; the National Research and Applied Medicine Mother and Child Centre, Minsk, Belarus (LM, IV, NB, ZS, ID, and GS); the Department of Social Medicine, University of Bristol, Bristol, United Kingdom (RMM and GDS); the Department of Ambulatory Care and Prevention, Harvard Medical School/Harvard Pilgrim Health Care, Boston, MA (MWG); the Department of Epidemiology and Community Health, Queen's University, Kingston, ON, Canada (BC); and the Faculty of Nursing, University of Toronto, Toronto, ON, Canada (EH)
2 Supported by a grant from the Canadian Institutes of Health Research and by grant no. FOOD-DT-2005-007036 from the European Union's project on Early Nutrition Programming: Long-term Efficacy and Safety Trials (to RMM and GDS). 3 Reprints not available. Address correspondence to MS Kramer, The Montreal Children's Hospital, 2300 Tupper Street (Les Tourelles), Montreal, PQ H3H 1P3, Canada. E-mail: michael.kramer{at}mcgill.ca.
ABSTRACT
Background: The evidence that breastfeeding protects against obesity and a variety of chronic diseases comes almost entirely from observational studies, which have a potential for bias due to confounding, selection bias, and selective publication.
Objective: We assessed whether an intervention designed to promote exclusive and prolonged breastfeeding affects children's height, weight, adiposity, and blood pressure at age 6.5 y.
Design: The Promotion of Breastfeeding Intervention Trial (PROBIT) is a cluster-randomized trial of a breastfeeding promotion intervention based on the WHO/UNICEF Baby-Friendly Hospital Initiative. A total of 17 046 healthy breastfed infants were enrolled from 31 Belarussian maternity hospitals and their affiliated clinics; of those infants, 13 889 (81.5%) were followed up at 6.5 y with duplicate measurements of anthropometric variables and blood pressure. Analysis was based on intention to treat, with statistical adjustment for clustering within hospitals or clinics to permit inferences at the individual level.
Results: The experimental intervention led to a much greater prevalence of exclusive breastfeeding at 3 mo in the experimental than in the control group (43.3% and 6.4%, respectively; P < 0.001) and a higher prevalence of any breastfeeding throughout infancy. No significant intervention effects were observed on height, body mass index, waist or hip circumference, triceps or subscapular skinfold thickness, or systolic or diastolic blood pressure.
Conclusions: The breastfeeding promotion intervention resulted in substantial increases in the duration and exclusivity of breastfeeding, yet it did not reduce the measures of adiposity, increase stature, or reduce blood pressure at age 6.5 y in the experimental group. Previously reported beneficial effects on these outcomes may be the result of uncontrolled confounding and selection bias.
Key Words: Breastfeeding • adiposity • obesity • blood pressure • programming
INTRODUCTION
Between 3 and 12 mo of age, infants who receive prolonged and exclusive breastfeeding become thinner and shorter than those who are predominantly bottle-fed, although the former group partially catches up in the second year of life (1). This evidence led the World Health Organization (WHO) to develop new growth standards for infants who follow WHO recommendations for exclusive and prolonged breastfeeding (2). Many studies over the past several decades have suggested that breastfeeding offers long-term protection against obesity and that the degree of protection increases with greater duration and exclusivity of breastfeeding (3). More recent systematic reviews and meta-analyses, however, suggest that inadequate control for confounding differences between breastfed and formula-fed infants may explain the reported differences; these reviews also provide evidence of publication bias—ie, of selective publication of studies that reported a protective effect of breastfeeding (4, 5).
Fewer studies have examined long-term effects of breastfeeding on child and adult stature (height), although studies of the Boyd-Orr cohort paradoxically (in view of the negative effects cited above on length-for-age in the first 2 y of life) report an association between breastfeeding and greater stature (6). Recent studies of the Avon Longitudinal Study of Parents and Children cohort in the United Kingdom suggested a long-term, dose-response effect on serum insulin-like growth factor-I concentrations according to the degree of breastfeeding, which may underlie a positive effect on height (7).
Systematic reviews and meta-analyses suggested that breastfed infants have a mean systolic blood pressure that is slightly lower (pooled estimate: 1.4 mm Hg lower) than that of formula-fed infants (8, 9). Studies also suggested that breastfeeding reduces atherosclerosis as compared with formula feeding (10) but does not reduce the risk of coronary heart disease mortality (11).
The evidence for breastfeeding's beneficial effects on adiposity, stature, blood pressure, and other chronic disease outcomes derives almost exclusively from observational studies. As shown in the systematic review by Owen et al (5), residual confounding and publication bias may at least partially explain the published evidence. In addition, studies that compare the growth of infants receiving prolonged and exclusive breastfeeding with the growth of infants with shorter durations and lesser degrees of breastfeeding are susceptible to selection bias. Infants who grow more slowly for genetic or other constitutional reasons may be satisfied with their mothers' breast-milk output, whereas those who grow more rapidly may require a larger energy intake (12, 13), and mothers may be reluctant (or, in some cases, unable) to increase their milk supply by augmenting the frequency of nursing. To overcome the methodologic limitations inherent in observational studies, we designed a 6.5-y follow-up of children participating in the Promotion of Breastfeeding Intervention Trial (PROBIT), a cluster-randomized trial in the Republic of Belarus (14).
SUBJECTS AND METHODS
Subjects
The detailed methods of PROBIT and the results during the first year of follow-up were previously reported (14). The units (clusters) of randomization were maternity hospitals and one affiliated polyclinic (outpatient clinic where the children are followed for well-child and illness care) per hospital, and double randomization was based on both a random numbers table and a coin toss (14). The experimental intervention was based on the Baby-Friendly Hospital Initiative, which was developed by WHO and UNICEF to promote and support breastfeeding, particularly among mothers who have chosen to initiate breastfeeding (15); the control maternity hospitals and polyclinics continued the practices and policies in effect at the time of randomization. The trial results are based on a total of 17 046 healthy breastfed infants from 31 maternity hospitals or polyclinics; all were born at term, weighed 2500 g, and were enrolled during their postpartum stay (14). To our knowledge, PROBIT is the largest randomized trial ever undertaken in the area of human lactation. It has been registered as ISRCTN-37687716, and it conforms to the Consolidated Standards of Reporting Trials guidelines for the design, analysis, and reporting of cluster-randomized trials (16).
As previously reported (14), the 2 randomized groups were similar in baseline sociodemographic and clinical variables, including maternal age, education, number of other children at home, the proportion of mothers who had breastfed a previous child for 3 mo, cesarean delivery, maternal smoking during pregnancy, birth weight, gestational age, and 5-min Apgar score. The experimental intervention led to a substantial difference in the duration of breastfeeding, which persisted throughout the first year of follow-up: in the experimental and control groups, respectively, 72.7% and 60.0% of mothers were still breastfeeding at 3 mo, 49.8% and 36.1% were doing so at 6 mo, 36.1% and 24.4% were doing so at 9 mo, and 19.7% and 11.4% were doing so at 12 mo (14). In addition, the prevalence of exclusive breastfeeding in the experimental group at 3 mo was 7 times that in the control group (43.3% and 6.4%, respectively), although it was low in both groups at 6 mo (7.9% and 0.6%, respectively) (14). When breastfeeding was supplemented or discontinued, the primary substitute was locally manufactured infant formula. Details concerning the intakes of other liquids, cereals, and other solid foods during infancy were reported previously (17).
Follow-up interviews and examinations at age 6.5 y were performed by 1 pediatrician each at 24 of the 31 polyclinics; in the remaining 7 high-volume clinics, the follow-up visits were shared by 2 pediatricians. All anthropometric measurements were obtained in duplicate and averaged. These included standing and sitting height, measured with a wall-mounted stadiometer for standing height and a uniform, standard wooden bench for sitting height (both: Medtechnika, Pinsk, Belarus); weight, measured on an electronic digital scale (Bella 840; Seca Corporation, Hamburg, Germany); and head, waist, hip, midupper arm, and midthigh circumferences, measured with a nonstretchable cloth tape. Triceps and subscapular skinfold thicknesses were measured (also in duplicate) with the use of Lange skinfold calipers (Beta Technology, Santa Cruz, CA). Systolic and diastolic blood pressures were also measured in duplicate with the use of a digital oscillometric device (M1; Omron Healthcare, Milton Keynes, United Kingdom). When blood pressure was unobtainable with the Omron M1, a standard clinic sphygmomanometer was used. Training and standardization of pediatricians in obtaining the anthropometric and blood pressure measurements were the focus of a week-long training session involving a convenience sample of school-aged children living in a residential facility near Minsk, Belarus. The training included printed materials in Russian and a training video recording used in the National Health and Nutrition Examination Survey (NHANES) and provided by the National Center for Health Statistics. Each participating pediatrician was given a copy of the video, along with detailed written instructions (in Russian) for performing each measurement.
Because interpediatrician and intrapediatrician measurement agreement (ie, remeasurement of the same child within a brief interval) was not formally examined, and because blinding of pediatricians to the experimental or control assignment was infeasible, a re-measurement audit was designed to assess the validity and reproducibility of the polyclinic data. For each of the 38 pediatricians, 5 children were randomly selected for the audit, for a total of 190 audited children. So that all children seen in follow-up were eligible for selection, the audit was carried out after primary data collection had been completed, at an average of 17.7 mo (range: 5.3–32.6 mo) after the initial clinic visit. The audit of the anthropometric and blood pressure measurements was carried out by one of the Minsk-based authors (LM, ZS, ID, or GS), all of whom were pediatricians who were blinded to the measures obtained at the initial clinic visit but not to the experimental or control treatment allocation. Because of the time elapsed between the audit and initial polyclinic visits, results were compared with the use of Pearson correlation coefficients.
Statistical analysis
All statistical analyses were based on intention to treat. Differences in outcome between the experimental and control groups were analyzed by using the MIXED procedure in SAS software (version 8.2; SAS Inc, Cary, NC). This statistical procedure accounts for the clustered randomization and thus permits inference at the level of the individual child, rather than at the level of the cluster (maternal hospital and polyclinic). The modeled differences presented below are based on this cluster-adjusted model but are very similar to those obtained from a model that also adjusts for stratum-level variables [ie, geographic region (west or east) and urban or rural location] and for the following individual-level covariates: infant age at follow-up, sex, and birth weight and maternal education, maternal and paternal height (for standing and sitting height), and maternal and paternal BMI (for adiposity measures and blood pressure) (results available on request). We also analyzed mixed models that included terms for the sex of each child and a multiplicative sex x treatment interaction term. In addition to these analyses based on the continuous anthropometric measurements, we compared the proportions of children in the 85th and 95th percentiles of BMI based on the Centers for Disease Control and Prevention (CDC) 2000 standards (18). The analogous penalized quasi-likelihood for generalized linear mixed models [GLIMMIX (19)] procedure in SAS was used to estimate the adjusted odds ratios and 95% CIs for these dichotomous outcomes. Missing data were not imputed.
RESULTS
A total of 13 889 children were seen in follow-up for the present study; they represented 81.5% of the 17 046 who had originally been randomly assigned. Of the 3157 (17 046–13 889) children who were randomly assigned but not followed up, 88 had died, 2938 were lost to follow-up, and 131 were unable or unwilling to attend their visit. Follow-up rates were similar in the experimental (80.2%) and control (82.9%) polyclinic groups, but they varied considerably by polyclinic: from 56.1% at one of the Minsk polyclinics to 94.6% at the small, rural polyclinic at Klimovichi. The mean ± SD age at follow-up was 6.6 ± 0.3 y.
As shown in Table 1, the children followed up in the experimental and control groups had similar baseline characteristics; the small differences paralleled those seen at randomization and previously reported (14). The children also were virtually identical in mean parental BMI, according to the heights and weights reported by the accompanying parent (usually the mother) at the follow-up visit: BMIs of 24.5 and 25.7 in both groups for the mother and father, respectively.
View this table:
TABLE 1. Baseline comparison of experimental and control groups of children followed up at age 6.5 y
The audit results are summarized in Table 2. The results shown are the Pearson correlation coefficients (and 95% CIs) for comparing the initial clinic visit results and the results at the audit visit. The test-retest correlations were high (>0.80) for height, BMI, waist circumference, and midupper arm circumference; intermediate (0.60–0.80) for leg length, head circumference, and subscapular skinfold thickness; and low (<0.60) for triceps skinfold thickness, midthigh circumference, and systolic (and, particularly, diastolic) blood pressure measurements.
View this table:
TABLE 2. Pearson correlation coefficients (r) (and 95% CIs) for anthropometric and blood pressure measurements at the audit and original polyclinic visits1
The main study results, including the number of children in whom each measure was obtained, the crude subject-based means ± SDs in the experimental and control groups, the intraclass correlation coefficients (ICCs) reflecting the degree of within-polyclinic clustering, and the cluster-adjusted differences in means (and 95% CIs), are shown in Table 3. The results showed a high degree of clustering for triceps skinfold thickness and systolic blood pressure. No significant between-group differences were observed in any of the anthropometric or blood pressure results except a slightly higher midthigh circumference in the experimental than in the control group. The CIs were reasonably narrow for those measures with good correlations between measures at the initial clinic visit and at the audit (as shown in Table 2) and with lower ICCs—ie, height, BMI, head circumference, and subscapular skinfold thickness—but were fairly wide for triceps skinfold thickness and both systolic and diastolic blood pressures.
View this table:
TABLE 3. Anthropometric measurements and blood pressure results1
Significant (P < 0.05) treatment x sex interactions were observed for head circumference, subscapular skinfold thickness, and midthigh circumference. A significant effect of treatment on head circumference was observed in girls only [cluster-adjusted difference = 0.3 (95% CI: 0.02, 0.5) cm], whereas the effect on midthigh circumference was significant in both sexes, but higher in girls than in boys (1.0 and 0.7 cm, respectively). Effects on subscapular skinfold thicknesses were nonsignificant in both sexes but were opposite in direction (ie, positive in boys, negative in girls).
The proportions of children with BMIs 85th and 95th percentiles were 13.4% and 5.9% in the experimental group and 12.2% and 5.0% in the control group. The corresponding adjusted odds ratios were 1.1 (95% CI: 0.8, 1.4) and 1.2 (0.8, 1.6).
To assess whether we could reproduce the protective associations of prolonged and exclusive breastfeeding reported in previous observational studies, we compared 2 extreme observational infant-feeding groups previously reported (20) to differ widely in their growth trajectories during the first 12 mo of life: 1) those completely weaned within the first month (n = 1136) and 2) those breastfed exclusively for 6 mo with continued breastfeeding to any degree until 12 mo of age (n = 215). The observational analyses were based on multiple linear regression models containing the same stratum- and individual-level covariates as the expanded mixed models described above. None of the observed differences suggested lower adiposity or blood pressure in the group with prolonged and exclusive breastfeeding than in the group with less breastfeeding. In fact, the former group had significantly higher mean BMIs [cluster-adjusted difference: 0.3 (95% CI: 0.4, 0.5)], triceps skinfold thicknesses [1.3 (0.7, 1.9) mm], and systolic blood pressure [1.5 (0.1, 2.9) mm Hg] than did the latter.
DISCUSSION
Our results, the first based on a randomized study of healthy, full-term infants, indicate that prolonged, exclusive breastfeeding provides no apparent beneficial effects on stature, BMI or other measures of adiposity, or blood pressure in 6.5-y-old Belarussian children. Our results are consistent with those of the recently published meta-analysis conducted by Owen et al (5), and they suggest that previous studies reporting differences in these outcomes may have been subject to biases due to residual confounding, subject selection, and selective publication. Residual confounding is of particular concern (21). In the individual-patient meta-analysis by Owen et al, an association between breastfeeding (rather than formula feeding) and mean BMI was virtually abolished after control for socioeconomic status, parental BMI, and maternal smoking (5). A recent analysis of the large Nurses' Health Study II also found no association between the duration or exclusivity of breastfeeding and obesity in adulthood (22).
Despite the large sample size and the high rate of follow-up in the present study, the precision of the observed differences in those anthropometric blood pressure measures that were more observer (ie, pediatrician) dependent was only modest: in particular, for triceps skinfold thickness and blood pressure. Wide CIs resulted not only from random measurement error among children examined by individual pediatricians, but also from systematic differences among the 38 participating pediatricians, which led to systematic differences in the means of these anthropometric and blood pressure measurements among the 31 polyclinics. These systematic measurement differences constituted a source of clustering within polyclinics and resulted in high intraclass correlations, ie, a tendency for children within a single polyclinic than for those from different polyclinics to have similar measurements. Because randomization was also clustered within polyclinics, this "double clustering" substantially reduced the precision (ie, yielded wider CIs) of the observed differences between the experimental and control groups. Unfortunately, for both geographic and economic reasons, it was infeasible for pediatricians who worked at one polyclinic to examine children enrolled in other polyclinics. Thus, the present study cannot exclude small differences that may be due to the intervention, especially with respect to triceps skinfold thickness and blood pressure.
It is important to emphasize that the experimental intervention was designed to increase the degree and duration of breastfeeding, not to increase its initiation. Thus, our findings may not apply to comparisons of breastfeeding versus formula feeding (the comparisons most often emphasized in previous studies). Nonetheless, many previous studies have reported graded, dose-response associations between the degree or duration (or both) of breastfeeding and adiposity measures. Our findings clearly do not support those associations.
Although Belarus is a developed country that has experienced lower rates of infant mortality and that has more-favorable health indicators than have other former Soviet countries, we are aware of no published data on the extent to which the obesity epidemic currently gripping North America and Western Europe is affecting Belarus. The fact that the observed proportions of children with BMIs 85th (13%) and 95th (5%) percentiles are far lower than those recently reported from the United States (23) suggests a far less severe problem among Belarussian children. Caution is therefore advised in generalizing our results to settings with a much higher prevalence of child obesity. Nonetheless, in an examination of reported protective effects of breastfeeding on obesity over the several decades that this association has been studied, no apparent differences are evident, nor are we aware of any biological mechanisms that would modify the potential protective effect of breastfeeding in different settings. Continued follow-up of the PROBIT cohort, along with attempts to provide more direct measurements of body fat and to reduce variations in the measurement of blood pressure, should help in detecting effects of smaller magnitude (which could nonetheless have an important public health effect), as well as those that may develop later in childhood or adulthood with increased deposition of body fat. Until that time, however, it seems unwise to depend on current efforts to promote exclusive and prolonged breastfeeding as an effective population health strategy for stemming the current obesity epidemic or reducing the risk of future hypertension.
ACKNOWLEDGMENTS
The additional contributing members of the PROBIT Study Group are named. National Research and Applied Medicine Mother and Child Centre (Minsk, Belarus): Natalia Bazulko, Olga Gritsenko, Lidia Ovchinikova, and Julia Rizkovskaya. Polyclinic pediatricians: Natalia Andreeva (Rogachev), Tatiana Avdeichuk (Brest), Elena Avsiuk (Vitebsk), Irina Baikevich (Slonim), Zinaida Bisucova (Zlobin), Irina Bujko (Oshmiany), Tamara Galushkina (Volkovysk), Marina Gotovchi (Brest), Danuta Iodkovskaya (Berestovitsa), Galina Ivanova (Mogilev), Larisa Kebikova (Minsk), Galina Kluchnikova (Ostrovets), Maria Kotliarovich (Soligorsk), Galina Kovalevskaya (Lepel), Natalia Krokas (Mosty), Nadezda Kushkova (Rechitsa), Afanasia Lazarenko (Klimovichi), Ludmila Lazuta (Minsk), Zinaida Liamkina (Borisov), Raisa Lisiura (Stolin), Tamara Nabedo (Novolukoml), Svetlana Pleskach (Baranovichi), Oksana Potapenko (Soligorsk), Svetlana Pridhodoskaya (Bereuza), Valentina Rahotskaya (Oshmiany), Irina Rogach (Mstislavl), Ludmila Rutkovskaya (Kobrin), Natalia Senchuk (Rechitsa), Elena Seraia (Baranovichi), Ludmila Sheveleva (Kobrin), Vera Shota (Svisloch), Anna Silvanovich (Shuchin), Lilia Smolskaya (Glubokoe), Valentina Solovey (Volkovysk), Zoya Solovyova (Dokshitsy), Natalia Tsarik (Svetlogorsk), Nadezda Turkovskaya (Zlobin), and Oxana Zarodova (Minsk Region).
The authors' responsibilities were as follows—MSK, RWP, J-PC, SS, EH, and BC: contributed to obtaining funding for this project and to the design, analysis, interpretation, and writing or revision (or both) of the manuscript; LM, IV, NB, ZS, ID, and GS: contributed to the design of the study and to the planning, implementation, and monitoring of the field work in Belarus; RMM, GDS, and MWG: contributed to the selection of, and measurement methods for, the anthropometric and blood pressure outcomes and to the content of the manuscript. MSK is Senior Investigator of the Canadian Institutes of Health Research; RWP is a Monat-McPherson Career Investigator of McGill University and a career investigator (chercheur-boursier) of the Fonds de la recherche en santé du Québec.
REFERENCES