Association of breastfeeding , maternal anthropometry and body composition in women at 30 years of age Associação entre aleitamento materno

This study aimed at assessing the association of breastfeeding with maternal body mass index (BMI), waist circumference, fat mass index, fat free mass index, android/gynoid fat ratio and bone mineral density. In 1982, the maternity hospitals in Pelotas, Rio Grande do Sul State, Brazil, were daily visited and all live births were identified and examined. These subjects underwent followup for several times. At 30 years of age, the participants were interviewed and examined. Parous women provided information on parity and duration of breastfeeding. Multiple linear regression was used in the multivariate analysis, controlling for genomic ancestry, family income, schooling and smoking at 2004-2005. After controlling for confounding factors, breastfeeding was inversely associated with BMI and fat mass index, whereas breastfeeding per live birth was negatively associated with BMI, waist circumference and fat mass index. Women who had had a child in the last 5 years and had breastfed, showed lower BMI (β = -2.12, 95%CI: -4.2; -0.1), waist circumference (β = -4.46, 95%CI: -8.3; -0.6) and fat mass index (β = -1.79, 95%CI: -3.3; -0.3), whereas no association was observed among those whose last childbirth was > 5 years, but the p-value for the tests of interaction were > 0.05. Our findings suggest that breastfeeding is associated with lower BMI and other adiposity measures, mostly in the first years after delivery. Besides that, it has no negative impact on bone mineral density. Breast Feeding; Lactation; Anthropometry; Body Composition; Women ARTIGO ARTICLE This article is published in Open Access under the Creative Commons Attribution license, which allows use, distribution, and reproduction in any medium, without restrictions, as long as the original work is correctly cited.


Introduction
Breastfeeding has clear short-term benefits, decreasing mortality and morbidity from infectious diseases in childhood 1,2 .Additionally, breastfeeding protects against type 2 diabetes and obesity 3 , and is positively associated with human capital 4,5 .Beyond the benefits of breastfeeding to those who have been breastfed, studies report that women who breastfed would have lower risk of breast and ovarian cancer, and larger birth spacing 6 .
A recently published meta-analysis reported that breastfeeding mothers have lower postpartum weight retention of 380g (95%CI: -640; -110) than those that bottle-fed their child 7 .Furthermore, it has been suggested that breastfeeding would also be associated with the maternal body composition, such as skinfold thickness, fat free mass and fat mass, but most studies failed to observe an association or observed weak associations 8 .During lactation, women have a fetal demand for calcium 9 , but the evidence regarding the association between breastfeeding and bone mineral density shows conflicting results and is inconclusive 6,10 .Most studies were conducted in high-income countries, where there is a positive association between socioeconomic status and breastfeeding 11 .Because socioeconomic status is negatively associated with obesity in these settings 12,13,14 , the observed associations could be to residual confounding.
This study aimed at evaluating the association between breastfeeding and body mass index (BMI), waist circumference, fat mass index, fat free mass index, android/gynoid fat ratio and bone mineral density in parous women enrolled in the 1982 Pelotas (Brazil) Birth Cohort Study, a setting where duration of breastfeeding is not associated with socioeconomic status 15 .Therefore, this study should not be susceptible to residual confounding by socioeconomic status.

Methods
In 1982, all maternity hospitals in Pelotas, a Southern Brazilian city, were daily visited.The live births whose family lived in the urban area of the city were examined and their mothers interviewed (N = 5,914).These subjects underwent several follow-up visits; further details on the study methodology have been previously published 16,17 .
From June 2012 to February 2013, the cohort was invited to a new follow-up study using different strategies to locate cohort members.The subjects (mean age 30.2 years) were invited to the research clinic, where they were interviewed and examined by a trained team 18 .In this study, we included women who had had at least one live birth and were not pregnant at the time of interview.
In the 2012-2013 visit, we collected information on socioeconomic conditions, such as skin color (white; black; brown/indigenous/Asian), years of schooling (0-4; 5-8; 9-11; ≥ 12), and asset index according to the Brazilian Association of Research Companies criterion (A/B; C; D/E).Besides, women were asked about their parity and the duration of breastfeeding of each child.Total duration of breastfeeding was obtained by summing the duration of lactation of all children (in months).Women who breastfed < 1 month was classified as never breastfeeding.Weight was measured using a scale coupled to a BodPod (COSMED, Chicago, USA) equipment with a capacity up to 150kg, and height was measured using a portable stadiometer (aluminum and wood; Pelotas, Brazil).BMI was calculated dividing the weight by the squared height (kg/m 2 ).Individuals were classified as overweight if BMI was ≥ 25.0kg/m 2 .Waist circumference was measured with an inextensible tape with an accuracy of 0.1cm.This measure was collected twice during the same visit; if the difference between the two measures was above 1cm, a third measurement was performed.Total fat mass and free fat mass were evaluated by air displacement plethysmography (BodPod).Fat mass index and fat free mass index were assessed dividing the total fat and fat free mass by the squared height (kg/m 2 ).Android/gynoid fat ratio was calculated dividing the fat mass in the android region by the gynoid region, which were measured using dual-energy x-ray absorptiometry (DXA).The mineral density of the femoral neck bone was also measured by DXA.Individuals with metal body parts in the femoral neck (plates, pins) and those whose surgical intervention altered the anatomic structure of that segment of the skeleton were excluded from bone mineral density analysis.
Cad. Saúde Pública 2019; 35(2):e00122018 Analyses were carried out using Stata, version 14.0 (https://www.stata.com).We used chi-square test and analysis of variance (ANOVA) to compare proportions and means, respectively.Data distribution was assessed, and all outcomes were normally distributed.Multiple linear regression analysis included confounding factors with a p-value < 0.20 for the association with both body composition outcomes and breastfeeding variables.The following potential confounding variables were collected during the 2004-2005 visit: European genomic ancestry (based on approximately 370,000 single nucleotide polymorphisms mutually available for the Pelotas cohort and selected samples of the HapMap and Human Genome Diversity -ADMIXTURE was used to estimate the genomic ancestry of each subject) 19 , family income (in Brazilian reais), years of schooling (0-4; 5-8; 9-11; ≥ 12), leisuretime physical activity assessed through the International Physical Activity Questionnaire (minutes/week) 20 , and self-reported tobacco smoking (reported as "yes" or "no").Statistical comparisons were based on tests of heterogeneity and linear trend, and the one with the lower p-value was presented.Analysis was also performed stratifying results by time since last birth (< 5; ≥ 5 years) and interaction was tested fitting a linear regression model with an interaction term.
This study was conducted according to the guidelines established in the Declaration of Helsinki and all procedures involving human subjects were approved by the Research Ethics Committee of the Faculty of Medicine, Federal University of Pelotas (protocol number: Of. 16/12).Written informed consent was obtained from all subjects.

Results
In 2012-2013, we interviewed 3,701 individuals from the cohort that, added to 325 known deaths, represented a follow-up rate of 68.1%.With respect to women, 1,914 were interviewed and 130 had died, corresponding to 71.1% of the original cohort.Of those interviewed, 1,147 met the eligibility criteria.Information on breastfeeding was available for 1,146 women, while complete data on breastfeeding and at least one of the maternal body composition measures were available for 1,126 women.
Among the subjects included in this analysis, 73.6% were white, 34.4% had completed between 9 and 11 years of schooling, and the mean proportion of European ancestry was 75%.Most women were primiparous (52.6%) and 27.4% had breastfed for at least 24 months.The prevalence of overweight was 59% (Table 1).
With respect to the confounding variables, breastfeeding was higher in women with lower socioeconomic status in 2004-2005 (Table S1 on Supplementary Material: http://cadernos.ensp.fiocruz.br/site/public_site/arquivo/suppl-e00122018_6233.pdf), but no association was observed after taking parity into account (Table S2 on Supplementary Material: http://cadernos.ensp.fiocruz.br/site/public_site/arquivo/suppl-e00122018_6233.pdf).Fat free mass index and bone mineral density were slightly higher in women with lower family income and schooling in 2004-2005, whereas BMI and fat mass index were only associated with income, but these associations did not show a clear pattern (Tables S3 and S4 on Supplementary Material: http://cadernos.ensp.fiocruz.br/site/public_site/arquivo/suppl-e00122018_6233.pdf).
Tables 2 and 3 show the association of breastfeeding with maternal anthropometry and body composition.Even after controlling for possible confounding variables, women who had ever breastfed showed lower BMI (β = -1.57,95%CI: -2.8; -0.4), waist circumference (β = -3.41,95%CI: -5.8; -1.0) and fat mass index (β = -1.32,95%CI: -2.2; -0.4).In addition, the total duration of breastfeeding was inversely associated with BMI and fat mass index.For waist circumference, we observed reduction in all categories of total breastfeeding.For android to gynoid ratio, adjustment for confounding variables slightly decreased the magnitude of the associations and most of the confidence intervals included the reference.Duration of breastfeeding per live birth was negatively associated with BMI, waist circumference and fat mass index.For android to gynoid ratio, the association was not linear, but those who had breastfed for longer periods showed lower values.On the other hand, bone mineral density was not associated with breastfeeding.
Tables 4 and 5 show the analyses stratified by time since last birth.After controlling for confounding variables, women who had had a child in the last 5 years and had ever breastfed, showed lower  BMI (β = -2.12,95%CI: -4.2; -0.1), waist circumference (β = -4.46,95%CI: -8.3; -0.6) and fat mass index (β = -1.79,95%CI: -3.3; -0.3), compared to the never breastfeeding group.Among those whose last childbirth was > 5 years, the associations were weaker and the confidence intervals included the reference, but the tests for interaction were not statistically significant (p-value for interaction > 0.05).

Discussion
In this cohort, which has been prospectively submitted to follow-up since birth, breastfeeding was associated with lower fat mass, whereas there was no association with bone mineral density.We observed a trend toward weaker associations among those who had the last birth > 5 years before the interview, but the formal tests for interaction were not statistically significant.Therefore, we cannot exclude that these differences were due to random variation.With respect to adiposity, the association of breastfeeding with BMI 21,22,23,24,25,26,27 and abdominal adiposity measures 22,24,28,29,30,31,32 have been reported by some studies, whereas others have failed to describe such associations 33,34,35,36 .A systematic review observed that most of the studies reported little or no association between breastfeeding and body composition 8 , but the authors did not estimate the pooled effect.In the same token, a lower total fat mass 28,37 was reported among women who breastfed, but others failed to observe such association 32,38,39 .Concerning bone mineral density, two systematic reviews described that the evidence of an association with breastfeeding was not clear because of the high heterogeneity among studies 6,10 .As most studies were conducted in high-income countries, where socioeconomic status is positively associated with breastfeeding 11 and inversely associated with obesity 12,13,14 , these results might be due to residual confounding.In our study, when evaluating the duration of breastfeeding per live birth, with the intention of taking parity into account, we observed no association between breastfeeding and socioeconomic variables.Therefore, lower adiposity among women with longer duration of breastfeeding should not be attributed to residual confusion by socioeconomic status.
The Promotion of Breastfeeding Intervention Trial (PROBIT) evaluated the effect of breastfeeding on BMI, fat mass index and fat free mass index 40 .In an intention-to-treat analysis, mothers who had been allocated to the breastfeeding promotion group showed, at 11.5 years postpartum, lower BMI (-0.27kg/m 2 , 95%CI: -0.91; 0.37), fat mass index (-0.23kg/m 2 , 95%CI: -0.64; 0.17) and fat free mass index (-0.05kg/m 2 , 95%CI: -0.27; 0.16) than those in the control group, but the confidence intervals included nullity.In that trial, the intervention showed a higher duration of breastfeeding 41 , but at six months only 49.8% of mothers in the intervention group were still breastfeeding, as well as 36.1% in the control group.Due to low compliance to study protocol, there is a decrease in the statistical power 42 ; therefore, the non-significant association observed in this study should not be considered as an indication that there are no associations.
Concerning the possible mechanisms for the observed association between breastfeeding and maternal adiposity, it has been suggested that lactation may mobilize fat accumulations,   "resetting" maternal metabolism after pregnancy 43 .Lactation would improve beta-cell function, reducing insulin secretion, and suppress hypothalamic-pituitary-adrenal axis activity through the action of oxytocin and other lactogenic hormones, lowering cortisol levels 44 .Beside this, it has been estimated that exclusive breastfeeding in the first six months of life requires additional energy of approximately 500kcal/day 45 .Therefore, lactation may have an effect on maternal adiposity.Our findings suggest that the benefits of breastfeeding on maternal fatness decrease with time since last birth.This moderation on the effect might be due to the concept of energy balance, i.e. the effect decreases because the energy demand stops after ending the breastfeeding.Similarly with our results, previous studies also reported that the effect of breastfeeding on maternal metabolic outcomes reduces as women age 7,23,24,46,47 and with time since last birth 48,49,50,51 .
As strengths of the study, we assessed the association between breastfeeding, maternal anthropometry and body composition using information from a large birth cohort in which all the data was prospectively collected by a trained research team.In 2012-2013, we followed-up 71.1% of women from the original cohort, representing a high follow-up rate.The attrition rate was slightly higher Cad.Saúde Pública 2019; 35(2):e00122018 and daily frequency of breastfeeding.The latter might have led to a misclassification in the breastfeeding status, but this is a limitation in most of the studies evaluating breastfeeding and maternal health.Further, the breastfeeding measure was collected retrospectively; however, we believe that a possible error in the information on the duration of breastfeeding is independent of the outcomes evaluated, so this classification error would be non-differential.It is also important to mention that, because we tested multiple outcomes, some of the associations could have been statistically significant due to inflation of the type 1 error.However, we observed more significant associations than it would be expected by error.In addition, the outcomes are correlated measures, being unlikely that the results are due to type 1 error.
Because postpartum weight retention is associated with the development of overweight and obesity 52,53 , which increases the risk of chronic non-communicable diseases 54,55 , the global leading cause of morbidity and mortality 56 , ourt study brings further evidence on the beneficial effect of breastfeeding to the mother.These evidence should be taken into consideration when estimating the consequences of breastfeeding on maternal health.
Concluding, our results suggest that recent breastfeeding is associated with lower BMI and other adiposity measures.And, at the same time, it has no negative impact on bone mineral density.These findings suggest that breastfeeding may also have a positive consequence on maternal health, reinforcing, the relevance of interventions aimed at increasing breastfeeding duration.
Contributors N. P. Lima collaborated in the data collection, conducted the data analysis, interpreted the results and drafted the article.D. G. Bassani, B. G. C. Silva, J. V. S. Motta and E. I. S. Magalhães collaborated in the data analysis, interpretation and drafted the article.F. C. Barros designed the cohort study, collaborated in the data analysis and drafted the article.B. L. Horta participated in the conception, design, analysis, interpretation and discussion of the data.
BMD: bone mineral density; SD: standard deviation.* Brazilian Association of Research Companies criterion.

Table 3
Maternal body composition according to total sum of breastfeeding (N = 1,126).Pelotas, Rio Grande do Sul State, Brazil, 2012.