Pregnancy is associated with excessive weight gain and central fat deposition(Reference Butte and Hopkinson1), which are suggested to serve as an energy reservoir to ensure adequate nutrient supply for the infant(Reference Zafon2). Although the increase in maternal fat mass may be essential for the infant, it is evident that pregnancy poses a considerable risk to the mother's health, as manifested by increased blood pressure(Reference Magee, Abalos and von Dadelszen3), atherogenic lipid profile and insulin resistance(Reference Lain and Catalano4). Fortunately, postpartum lactation enables women to reverse this detrimental trend; during lactation, fat is mobilized primarily from the trunk and thighs(Reference Butte, Garza and Stuff5), reducing central fat accumulation, which is considered a major risk factor for metabolic disorders(Reference Despres and Lemieux6).
In spite of the recommendations to breast-feed(7) and the known beneficial effects of breast-feeding on both infant and maternal health, only 18% of mothers in Finland breast-feed until 12 months(Reference Erkkola, Salmenhaara and Kronberg-Kippila8). In the USA(Reference Gartner, Morton and Lawrence9) and Europe(Reference Yngve and Sjostrom10) the trends are similar, indicating that more information about the effects of breast-feeding on maternal health is called for. Animal studies have already suggested that repeated pregnancies without lactation may predispose mothers to obesity(Reference Jen, Juuhl and Lin11, Reference Zhong, Almario and Dubrinsky12). In humans, however, the weight-reducing effects of postpartum lactation have remained controversial. Some studies have shown that breast-feeding significantly reduces weight(Reference Hatsu, McDougald and Anderson13–Reference Dewey, Heinig and Nommsen15), whereas others have reported no effect on maternal anthropometrics and body composition(Reference Dugdale and Eaton-Evans16, Reference Boardley, Sargent and Coker17). These discrepancies in research results may be due to inconsistency in descriptions of breast-feeding patterns, variation in follow-up durations and parity. However, given the increasing trend towards overweight and obesity in all developed countries(Reference Flegal, Carroll and Ogden18, Reference Kautiainen, Koivisto and Koivusilta19), and the consequent detrimental health effects, the impact of lactation on maternal health and body composition during the reproductive years and beyond deserve attention.
If the anticipatory role for mothers is to deposit fat during pregnancy and lose it during lactation, we may ask: are mothers who do not lactate at risk for obesity? We hypothesized that repeated pregnancies in the absence of or followed by only a short period of postpartum lactation would have detrimental effects on body composition in women and increase risk for cardio-metabolic disorders in later life. To test our hypothesis, we investigated the long-term cumulative effects of repeated pregnancies and duration of postpartum breast-feeding on maternal anthropometrics, body composition and metabolic risk 16–20 years after the last parturition.
Materials and methods
The present retrospective study was part of the Calex-family study, which has been described elsewhere(Reference Cheng, Volgyi and Tylavsky20, Reference Xu, Nicholson and Wang21). Briefly, the study was conducted in the city of Jyväskylä and its surroundings in Central Finland in 2007 to 2008. Medical history and lifestyle factors, such as level of education, previous (stages of life: 20–29, 30–39, 40–49, > 50 years of age) and current participation in leisure-time physical activity, and information on breast-feeding, number of pregnancies and biological children, and pre-pregnancy weight and height, were collected via a self-administered questionnaire. Participants were also asked to provide detailed information on weight change patterns during each pregnancy from their personal maternity tracking records, which have been issued to mothers since the socialization of maternal health care in Finland. Intakes of total energy and energy-yielding nutrients were assessed from food records(Reference Lyytikainen, Lamberg-Allardt and Kannas22). Breast-feeding was expressed in total exclusive (giving the infant no food or liquid other than breast milk) and partial (giving the infant some breast milk in addition to other liquid or solid foods) breast-feeding months.
The recruitment of the study population is presented schematically in Fig. 1. Briefly, 212 mothers (mean age 48, range 36–60 years) participated in the body composition and other laboratory assessments. Of this number, 206 women provided valid information on the number of biological children they had. Thirteen (6·3 %) women reported having one child, seventy-one (34·5 %) had two, seventy-two (35·0 %) had three, thirty-five (17·0 %) had four and fifteen (7·3 %) had more than four biological children. Participants were excluded if they had gestational diabetes or hypertension, were currently pregnant or reported being pregnant within 5 years before the present measurements. In addition, we excluded those who reported twin pregnancies (n 5) or did not have body composition data (n 3). Thus, the final sample consisted of 198 mothers.
The 198 participants were divided into three groups according to the average duration of breast-feeding they reported (total months of breast-feeding divided by the number of biological children). WHO recommends 6 months for a minimum duration to breast-feed(7). The lowest cut-off point was selected on the basis of this recommendation. Sixty-seven (34 %) mothers reported a short duration of breast-feeding (SDB, defined as breast-feeding their infants for less than 6 months), sixty-eight (34 %) reported a medium duration of breast-feeding (MDB, defined as breast-feeding their infants for more than 6 months but less than 10 months) and sixty-three (32 %) reported a long duration of breast-feeding (LDB, defined as breast-feeding for longer than 10 months). Five (7 %) of the SDB and two (3 %) of the LDB mothers reported that they had never exclusively breast-fed their infants. One hundred and fifty-one mothers provided pre-pregnancy anthropometric data and 126 mothers provided maternity tracking records. The study protocol was approved by the ethical committee of the Central Health Care District, Central Finland. Written informed consent was given by all participants prior to the assessments.
Anthropometric and body composition assessments
Body height (cm) and weight (kg) were measured using standardized protocols, and BMI (kg/m2) was calculated. Bone mass (BM in kg), lean tissue mass (LM in kg) and fat mass (FM in kg) of the whole body and in different areas of the body (arm, leg, trunk, android and gynoid, femoral and gluteal regions) were assessed using dual-energy X-ray absorptiometry (Prodigy; GE Lunar Corp., Madison, WI, USA). The CV of two repeated measurements on the same day was on average 0·7 % for BM, 1·0 % for LM and 2·2 % for FM in the present study. Blood pressure was measured twice in the morning, by the oscillometric method, after the participants had arrived at the laboratory and sat at rest for 10 min.
Blood samples and analysis
Blood samples were drawn from the antecubical vein in the morning (07.00–09.00 hours) after overnight fasting (12 h). If the mothers were in the premenopausal state, the blood sample was drawn on the fifth day from the start of menstruation. Serum was separated within 30 min, and stored at −80°C until analysis. Serum glucose, total cholesterol (TC), HDL cholesterol (HDL-C) and TAG concentrations were measured by enzymatic photometry on a Kone Pro Clinical Chemistry Analyser (Thermo Clinical Labsystems Oy, Espoo, Finland) with commercial kits. LDL cholesterol (LDL-C) was calculated using the Friedewald equation(Reference Friedewald, Levy and Fredrickson23). An IMMULITE Analyser (Diagnostic Products Corporation, Los Angeles, CA, USA) and an IMMULITE Insulin Kit were used for quantitative determination of insulin. The CV of two repeated measurements on the same day was on average 3·4 % for insulin.
All data were checked for normality using the Shapiro–Wilk W test in the SPSS for Windows statistical software package version 15·0 (SPSS Inc., Chicago, IL, USA). If data were not normally distributed, natural logarithms were used. ANOVA with the Least Significant Difference post hoc test was used to compare differences among the breast-feeding groups in pre-pregnancy anthropometrics, age at baseline and at each pregnancy, and weight change during each pregnancy. To validate the self-estimated weight before the first pregnancy, we compared the self-estimated weight with the measured weight at the beginning of the first pregnancy using Bland–Altman analysis(Reference Bland and Altman24).
A general linear model was employed to compare the differences in anthropometric, body composition, serum glucose and insulin and lipid profiles among the groups 16–20 years after the last parturition. All the analyses were adjusted for relevant factors. Statistical significance was set at P < 0·05.
To determine which factors were associated with weight gain, total body fat mass and android fat mass, a generalized estimating equations (GEE) model was used. The following predictors were included in the model: pre-pregnancy weight and BMI, age at first pregnancy, smoking, menopause status, level of education, previous and current participation in leisure-time physical activity, current dietary energy intake, number of biological children, and duration of exclusive and total breast-feeding months. The R 2 of the GEE model was computed according to Hardin(Reference Hardin and Hilbe25), and can be interpreted as a similar measure of the proportion of the outcome variance explained by the model as that in common linear regression.
The characteristics of the participants in the year before the first pregnancy are given in Table 1. No significant differences were found in age, BMI or systolic and diastolic blood pressure between the groups before the first pregnancy. Gestational durations and inter-pregnancy intervals were similar among all groups (P > 0·05 for all, data not shown). Consequently, there were no significant differences in average age at the subsequent pregnancies (second pregnancy, SDB 28·6 (sd 3·8) years, LDB 28·2 (sd 3·1) years and MDB 29·0 (sd 3·8) years; third pregnancy, SDB 31·3 (sd 4·2) years, LDB 29·9 (sd 3·3) years and MDB 31·1 (sd 3·4) years; fourth pregnancy, SDB 34·0 (sd 4·3) years, MDB 33·5 (sd 4·3) years and LDB 33·0 (sd 2·1) years) among the groups (all P > 0·05).
SDB, short duration of breast-feeding; MDB, medium duration of breast-feeding; LDB, long duration of breast-feeding; BP, blood pressure.
*Significant at P < 0·05.
†Blood pressure obtained at the beginning of the first pregnancy.
Between the years 1980 and 2008, the SDB mothers gained significantly more body weight (14·0 (sd 9·1) kg) than either the MDB mothers (8·3 (sd 6·5) kg; P = 0·001) or the LDB mothers (7·6 (sd 6·6) kg; P < 0·001; Fig. 2). Gestational weight gains were similar among groups in all pregnancies. However, between the first parturition and the beginning of the second pregnancy, weight reduction was more significant in the MDB (−10·5 (sd 4·5) kg) and LDB (−11·4 (sd 3·4) kg) mothers than in the SDB mothers (−8·0 (sd 4·4) kg; P < 0·01 for all). In fact, only the LDB mothers had returned to their pre-pregnancy weight (59·5 (sd 6·7) kg; P = 0·084), whereas the MDB mothers were 1·8 (sd 2·6) kg and SDB mothers 5·0 (sd 4·2) kg heavier by the beginning of the second pregnancy (P < 0·001 for both). Although gradual weight gain was observed in all groups, this phenomenon followed a similar pattern in the mothers’ subsequent pregnancies (Fig. 2). The self-estimated weight before the first pregnancy correlated significantly with the weight measured at the first time point (8 weeks) of the first pregnancy (R 2 = 0·898, P < 0·001). Furthermore, a Bland–Altman analysis showed a good agreement (mean difference 2·6 %) and the disagreement was 7·7 % between the self-estimated and first measured weight (data not shown), regardless of the fact some weight gain is experienced during early pregnancy.
At 16–20 years after the last parturition, no significant differences between the groups in age, height, total body lean mass, total bone mass, number of biological children, smoking status, menopause, current total energy and macronutrient intake, or participation in current leisure-time physical activity were found (all P > 0·05; Table 2). In addition, no significant differences between the groups in previous participation in leisure-time physical activity were observed (data not shown). However, the SDB mothers weighed more and had higher BMI and total fat mass than either the MDB or LDB mothers (P < 0·001 for both; Table 2). In addition, a notably higher proportion of the SDB mothers had only a basic level of education compared with the MDB and LDB mothers (P < 0·05 for both). Furthermore, the SDB mothers had significantly more fat in the android region (46·5 (sd 8·2) %) than either the MDB (39·0 (sd 10·8) %) or the LDB mothers (38·4 (sd 10·9) %; P < 0·01 for both). After adjusting for level of education, the level of significance remained the same (all P < 0·05; Fig. 3).
SDB, short duration of breast-feeding; MDB, medium duration of breast-feeding; LDB, long duration of breast-feeding; BP, blood pressure; HRT, hormone replacement therapy; B, basic education (high-school or upper secondary level); H, higher education (polytechnic/academic level); PA, physical activity; TBF, total breast-feeding months; %E, percentage of daily energy intake.
Results are adjusted for level of education.
*Significant at P < 0·05.
†Significance from natural log values comparison.
In accordance with their higher fat mass, the SDB mothers had significantly higher fasting serum glucose concentration than the LDB mothers (P = 0·033), and higher insulin concentrations and insulin resistance index (HOMA-IR) than either the LDB or MDB mothers (all P < 0·05; Table 3). Similarly, the SDB mothers had higher serum TC, LDL-C and TAG levels than either the LDB or MDB mothers (all P < 0·05), and lower HDL-C than the MDB mothers (P = 0·016; Table 3). Furthermore, the SDB mothers had higher systolic and diastolic blood pressure than either the MDB or LDB mothers. After adjusting for level of education, the level of significance remained unchanged (P < 0·05 for both).
TC, total serum cholesterol; HDL-C, HCL cholesterol; LDL-C, LDL cholesterol; GLUC, serum fasting glucose; INS, serum fasting insulin; HOMA-IR, homeostasis model assessment – insulin resistance index.
Results were adjusted for level of education.
*Significant at P < 0·05.
†Significance from natural log values comparison.
A GEE model was used to evaluate the risk factors associated with body composition. We found that a common predictor for low gain in body weight, total fat mass and android fat mass was long duration of breast-feeding (explaining 20 %, 24 % and 14 % of the variance, respectively; Table 4). Specific predictors for high total weight gain were high age at first pregnancy (11 %), menopause (7 %) and short duration of exclusive breast-feeding (6 %). Significant predictors for high total fat mass were high pre-pregnancy weight, menopause, short duration of exclusive breast-feeding and low current leisure-time physical activity (explaining 16 %, 5 %, 11 % and 7 % of the variance, respectively). Specific predictors for high android fat mass were high pre-pregnancy BMI (11 %), menopause (9 %), high age at first pregnancy (7 %) and short duration of exclusive breast-feeding (5 %).
TBF, total duration of breast-feeding; EBF, exclusive breast-feeding; LTPA, leisure-time physical activity.
Only statistically significant (P < 0·05) predictors are presented in the table.
In the current retrospective study, we found that body weight and body composition during the reproductive years were associated with the duration of postpartum lactation. Mothers who had breast-fed their infants for less than 6 months had higher total and relative body fat mass, particularly in the android region, compared with mothers who had breast-fed for longer than 6 months. This difference was observed 16–20 years after the last parturition and was independent of pre-pregnancy weight and BMI, menopausal status, smoking status, level of education, participation in previous and present leisure-time physical activity, and current dietary energy intake. Short duration of breast-feeding was also associated with increased risk of metabolic disorders manifested by higher fasting serum glucose concentration and insulin resistance, serum TAG, LDL-C and TC concentrations, as well as higher systolic and diastolic blood pressures.
In theory, weight loss is supported by negative energy balance due to either increased energy expenditure or reduced energy intake, or both. However, we did not find differences in current dietary energy or energy-yielding nutrient intake, or in previous or current participation in leisure-time physical activity, among the different breast-feeding duration groups. Although postpartum lactation significantly increases energy expenditure due to the production of milk in the mammary glands(Reference Butte and Hopkinson1), it is also accompanied by increased energy intake(Reference Dewey, Heinig and Nommsen15). Unfortunately, we were unable to obtain postpartum dietary data from our participants. However, we speculate that given their long duration of breast-feeding and short-spaced pregnancies, it is unlikely that the participants had tried to lose weight postpartum by restricting their dietary energy intake. This indicates that postpartum weight loss cannot be explained merely by changes in energy expenditure or energy intake alone. Thus, we speculate that postpartum weight loss may relate mainly to hormonal/metabolic changes induced by lactation.
Indeed, after parturition, withdrawal of progesterone and the suckling of the breast by the infant facilitate the release of prolactin, thereby decreasing the level of oestrogen(Reference Butte and Hopkinson1), which in turn enhances the mobilization of adipose tissue stores(Reference Pansini, Bonaccorsi and Genovesi26). Furthermore, since prolactin also inhibits lipogenesis(Reference Ben-Jonathan, Hugo and Brandebourg27) and suppresses glucose uptake in adipose tissue(Reference Nilsson, Roepstorff and Kiens28), it is conceivable that the pregnancy-induced pattern of fat deposition may be reversed during lactation by the fluctuating web of hormones. Unfortunately, due to the retrospective design of our study, we were unable to obtain data on changes in the hormonal milieu during and after pregnancy that would have allowed us to explore the underlying metabolism and confirm that the hormonal changes were induced by lactation. Nevertheless, we found that short duration of breast-feeding was the strongest independent predictor for high weight gain as well as total and central fat mass accumulation 16–20 years after the last parturition, explaining 14–24 % of the variance. Therefore, the long-lasting effect of hormonal changes induced by lactation on fat accumulation cannot be ruled out.
Accumulation of fat mass, especially in the android region, is associated with high risk of chronic diseases such as type 2 diabetes and CVD(Reference Despres and Lemieux6). The present study shows the long-term effects of breast-feeding duration not only on body weight and body composition but also on blood glucose and the lipid profile. Short duration of breast-feeding was associated with long-term accumulation of cardio-metabolic risks, suggesting that proper duration of lactation may not only reverse gestational hyperlipidaemia and impaired glucose metabolism, but also stabilize it thereafter, possibly even providing protection against type 2 diabetes in later life(Reference Stuebe, Rich-Edwards and Willett29). Furthermore, most previous studies have only investigated the effect of breast-feeding in single pregnancies(Reference Hatsu, McDougald and Anderson13–Reference Dugdale and Eaton-Evans16, Reference Brewer, Bates and Vannoy30–Reference Gigante, Victora and Barros32), and therefore the plausible cumulative effect of weight gain and duration of breast-feeding in repeated pregnancies over time has been overlooked or neglected. In our study we found that body weight accumulated with each consecutive full-term pregnancy and that after each pregnancy women with short duration of breast-feeding gained more weight than women in the other groups. This indicates that prolonged breast-feeding may be essential for prevention of weight gain during reproductive age in women, thereby reducing the high risk for chronic diseases in later life.
Apart from duration of breast-feeding, several other factors have also been found to predict fat mass accumulation and weight retention postpartum. In our study cohort we found no differences in BMI, blood pressure or age at pre-pregnancy. However, we found that weight gain was associated with age at first pregnancy and menopausal status. Accumulation of fat mass was also moderately associated with current participation in leisure-time physical activity in addition to age at first pregnancy and menopausal status. Furthermore, mothers with the shortest duration of breast-feeding had the lowest level of education, suggesting that to meet the current breast-feeding recommendations enhanced counselling is needed in particular for women with a low level of education.
The strength of our study is that we were able to obtain detailed and accurate data on weight change during each pregnancy from the maternal tracking records. In addition, we were able to use data with multiple factors to assess whole body composition, including bone, muscle and fat distribution in different body compartments. However, the study has some limitations. First, breast-feeding data collected retrospectively, particularly after a considerable period of time, may be subject to recall bias. However, the limited evidence available suggests that maternal recall does provide accurate estimates of initiation and duration of breast-feeding with high validity(Reference Li, Scanlon and Serdula33) even after 20 years(Reference Kark, Troya and Friedlander34). Second, it is difficult to establish the exact physiological mechanisms that drive changes of maternal body composition during the reproductive period on account of the array of potentially confounding factors. Furthermore, we did not obtain maternal tracking records and pre-pregnancy data from all of our participants. However, there were no differences in body weight between those who provided the pre-pregnancy anthropometrics and those who did not at 16–20 years after the first pregnancy. Given that weight gain pattern is heavily dependent on accurate estimation of the pre-pregnancy weight, we validated the self-estimated pre-pregnancy weight with the documented maternity tracking records. Our results showed a good agreement between the self-estimated weight and measured weight. Considering the consistency and the power of our results, the missing data did not change the paradigm of the study, which partially justifies the authenticity of the results.
In conclusion, our results provide an important public health message that short duration of breast-feeding may induce weight retention and fat mass accumulation, resulting in increased risk of cardio-metabolic disorders in later life. Encouraging women to prolong breast-feeding beyond 6 months may provide an important strategy for reducing the growing obesity epidemic and obesity-related cardio-metabolic disorders in women.
Sources of funding: This study was financially supported by the Academy of Finland, Ministry of Education of Finland, University of Jyväskylä, Juho Vainio Foundation, and ASBMR Bridge Funding Research Grant 2006. Conflict of interest: All authors declared no conflicts of interest. The study sponsors played no role in the study design, data collection, analysis and interpretation, writing of the report, or in the decision to submit the paper for publication. The authors were solely responsible for writing and submitting the manuscript for publication. Author contributions: S.-L.C. has full access to all of the data in the study and takes full responsibility for the integrity of the data and for the accuracy of the data analysis. Study concept and design: S.-L.C., P.W., L.-T.X., M.A., S.K.-K. Acquisition of data: S.-M.C., P.W., L.-T.X., A.L., J.S., Q.W., E.V., E.M., S.-L.C. Analysis and interpretation of data: P.W., L.-T.X., A.L., J.S., Q.W., E.V., E.M., S.-M.C., M.A., S.K.-K., S.-L.C. Drafting of the manuscript: P.W., L.-T.X., Q.W., M.A., S.K.-K., S.-L.C. Critical revision of the manuscript for important intellectual content: P.W., L-T.X., A.L., J.S., Q.W., E.V., E.M., S.-M.C., M.A., S.K.-K., S.-L.C. Statistical expertise: P.W., L.-T.X., Q.W., S.-M.C. Obtained funding: M.A., S.K.-K., S.-L.C. Administrative, technical or material support: P.W., Q.X., A.L., J.S., Q.W., E.V., E.M., S.-M.C., M.A., S.K.-K., S.-L.C. Acknowledgments: The authors would like to thank the entire research staff, and especially Heli Vertamo and Erkki Helkala, for their valuable work and technical assistance on this project and all participants for their support.