Birth Weight, Adult Body Composition, and Subcutaneous Fat Distribution

Authors

  • Saskia J. te Velde,

    1. Institute for Research in Extramural Medicine (EMGO), Amsterdam Growth and Health Research Group; Amsterdam, The Netherlands
    Search for more papers by this author
  • Jos W.R. Twisk,

    1. Institute for Research in Extramural Medicine (EMGO), Amsterdam Growth and Health Research Group; Amsterdam, The Netherlands
    Search for more papers by this author
  • Willem van Mechelen,

    1. Institute for Research in Extramural Medicine (EMGO), Amsterdam Growth and Health Research Group; Amsterdam, The Netherlands
    2. Department of Social Medicine and Body@work Research, Centre for Physical Activity, Work and Health, TNO VU, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • Prof. Dr. Han C.G. Kemper

    Corresponding author
    1. Institute for Research in Extramural Medicine (EMGO), Amsterdam Growth and Health Research Group; Amsterdam, The Netherlands
      AGAHLS Research Group, EMGO Institute, VU University Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands. E-mail: HCG.Kemper.emgo@med.vu.nl
    Search for more papers by this author

AGAHLS Research Group, EMGO Institute, VU University Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands. E-mail: HCG.Kemper.emgo@med.vu.nl

Abstract

Objectives: To investigate if birth weight is related to both body mass index (BMI) and distribution of subcutaneous fat at adult age.

Research Methods and Procedures: A 9-year longitudinal study was performed in 229 subjects (192 women) with ages ranging from 27 to 36 years. Birth weight was retrieved by a questionnaire, and adult weight, height, skinfold thicknesses, and waist-to-hip ratio (WHR) were repeatedly measured at mean ages 27, 29, 31, and 36 years. BMI, sum of four skinfolds (S4S), the ratio between two truncal skinfolds and S4S (SS/S4S), and the ratio between WHR and the cross-sectional area of the left thigh were calculated with the available data.

Results: The adjusted model showed that in women, birth weight was significantly negatively related to adult S4S [β = −5.211; (−9.768 to −0.654)], waist circumference [β = −1.449; (−2.829 to −0.069)], and SS/S4S ratio [β = −3.579; (−5.296 to −1.862)]. In men, a significant negative association was observed between birth weight and adult WHR [β = −1.096; (−2.092 to −0.100)] only. Other relationships showed, although not significantly, the same negative trend, namely that lower birth weight is related to higher adult body fat mass (S4S) and a more truncal subcutaneous fat distribution (SS/S4S). No associations were found between birth weight and either adult BMI or the cross-sectional area of the thigh.

Discussion: Lower birth weight is, in both adult men and women, related to a higher adult subcutaneous fat mass and a more truncal distribution of subcutaneous fat, indicating a higher risk for obesity.

Introduction

In the last decade, much research has been published about the fetal origins of the development of chronic diseases later in life (1, 2, 3, 4). These studies suggest that intrauterine life may be a critical period for the development of several risk factors for chronic diseases. To date, many epidemiological studies have assessed the relationship between birth weight, as a marker of intrauterine growth, and cardiovascular diseases. However, less is known about the relationship between birth weight and overweight, obesity, or body fat distribution in adult life. This relationship is relevant for public health because overweight and central obesity are important risk factors for metabolic diseases (5, 6, 7, 8, 9). Previous studies assessing the relationship between birth weight and adult overweight, fat mass, or body fat distribution showed inconsistent results (2, 10, 11, 12, 13, 14, 15, 16, 17, 18). Furthermore, most of these studies used single measures of body fat or body fat distribution as outcomes in relation to birth weight. Therefore, the aim of this study was to investigate the relationship between birth weight and both adult body fat distribution and adult body composition in men and women.

Research Methods and Procedures

Subjects and Design

The subjects described here are all participants of the Amsterdam Growth and Health Longitudinal Study (AGAHLS),1 which started in 1976 as an observational study. In this longitudinal study, measurements concerning growth and lifestyle were carried out in pupils of two secondary schools in and around Amsterdam. For the study presented here, only the anthropometric data of the last four measurements were used. These four repeated measurements were performed in 1991 (mean age of 27 years), in 1993 (mean age of 29 years), in 1996/1997 (mean age of 32/33 years), and in 2000 (mean age of 36 years). The design of this longitudinal study is graphically represented in Figure 1. More detailed information about the AGAHLS has been described in previous publications (19, 20, 21). All participants were born between 1961 and 1965, and are all residents of The Netherlands. All subjects gave informed consent, and the Medical Ethical Committee of the Vrije Universiteit approved the protocol.

Figure 1.

Design of the longitudinal study. Recalled birth weight is used as a measure of intrauterine growth, and indicators for body composition and body fat distribution are measured up to a maximum of four times during adulthood.

Data Collection

Anthropometric measurements of body height, body mass, and four skinfolds (biceps, triceps, subscapular, and suprailiacal) were performed according to standard procedures (22, 23) and with the same instruments throughout the study. Body mass index (BMI) was calculated by dividing body mass (kilograms) by body height squared (meters squared). The sum of the four skinfolds (S4S) was used as a measure of subcutaneous fat mass. The ratio between the two truncal skinfolds (SS) and S4S [SS/S4S, which is defined as (subscapular + suprailiacal)/S4S] was used as an indicator of a central pattern of subcutaneous body fat. Waist-to-hip ratio (WHR) was used as another indicator of a central pattern of body fat distribution. Therefore, circumferences of the waist (at umbilicus height) and hip were measured with a flexible steel tape to the nearest 0.1 cm. Waist circumference alone was used as a measure of abdominal fat (24). Finally, as a measure of muscle mass, the lean mass of the thigh was calculated. For that purpose, thigh circumference and anterior thigh skinfold thickness were measured, and the lean mass of the thigh was estimated by computing a cross-sectional area of the lean thigh (14).

Data on birth weight and gestation were retrieved at age 36 by a retrospective questionnaire. Of all the participants, 91% (396 subjects) filled in the questionnaire. Of those, 238 subjects (60%) were able to retrieve written information, such as a birth certificate, and another 147 subjects (38%) retrieved the required information from their parents. The remaining 2% of subjects did not mention the source and were, therefore, excluded from this study. Previous research has shown that recalling birth information by parents is a valid method to obtain birth weight (25, 26, 27, 28, 29, 30, 31). There was no difference in the outcome variables between those who were included or excluded for further analyses (p > 0.217; 385 vs. 49 subjects, respectively). After the exclusion of twins (11 subjects) and those born preterm (i.e., before 37 weeks of gestation; 18 subjects), 137 men and 192 women had complete data sets about birth weight, as well as at least one measurement of adult anthropometry.

Statistics

Linear generalized estimating equation (GEE) analyses (32) were used to study the relationship between birth weight and the development of adult body fat distribution and adult body composition. GEE takes into account that the observations within each subject are correlated. To investigate if the relationships between birth weight and body composition and body fat distribution are time dependent, the interaction between time of measurement and birth weight was also studied.

GEE analyses were performed with the Statistical Package for Interactive Data Analyses (33). As a result, regression coefficients were estimated that reflect the linear relationships between the predictor variable (i.e., birth weight) and the outcome variables throughout the longitudinal period. For analyses, a crude model, as well as an adjusted model, was used, in which adult body mass was included as a time-dependent covariate in the adjusted model.

To study the linear relationships with adult BMI, S4S, waist circumference, WHR, SS/S4S, and the lean mass of the thigh, analyses were performed with birth weight as a continuous variable. Additionally, the same relationships were studied using tertiles of birth weight to investigate whether the above-mentioned relationships were linear or whether there was a threshold effect, which is a possibility when no significant results are observed in the first analyses. All analyses were performed separately for men and women.

Results

Of the 229 subjects included in this study, the mean birth weight for women was 3.41 kg (SD = 0.52 kg) and for men it was 3.58 kg (SD = 0.47 kg). Only six subjects had a birth weight below 2.5 kg, which is often used as a cut-off point for low birth weight (15, 34, 35, 36, 37). Characteristics of the subjects are presented in Table 1.

Table 1.  Characteristics of the male (M) and female (F) subjects at mean ages of 27, 29, 33, and 36 years
 Age 27 years (N = 53; 53)Age 29 years (N = 51; 70)Age 33 years (N = 126; 182)Age 36 years (N = 100; 143)
 MeanSDMeanSDMeanSDMeanSD
Height (m)        
 M1.830.061.830.071.840.061.840.06
 F1.700.061.700.061.700.061.710.06
Body mass (kg)        
 M76.69.279.09.981.59.083.610.4
 F64.37.865.98.366.29.068.19.8
BMI        
 M22.72.223.42.224.12.424.62.7
 F22.22.522.72.622.83.023.43.3
S4S (mm)        
 M37.414.942.418.643.116.646.715.5
 F47.917.252.719.752.520.155.320.1
Waist circumference (cm)        
 M78.96.179.96.483.06.585.08.0
 F68.35.770.26.971.06.773.48.9
WHR (×100)        
 M90.55.990.64.492.64.395.35.4
 F78.94.080.97.879.24.886.846.2
SS/S4S (×100)        
 M62.54.963.85.863.56.066.06.0
 F51.36.352.46.751.17.150.66.9
Lean thigh (cm2)        
 M    219.925.6222.930.1
 F    184.229.5180.928.9

Results from the linear GEE analyses for the crude models, as well as the adjusted models, are presented in Table 2.

Table 2.  Results from the GEE analyses for 137 men (M) and 192 women (F)
 Crude modelAdjusted for adult body mass
 β95% CIβ95% CI
  • *

    p ≤ 0.05.

  • p ≤ 0.001.

BMI    
 M0.401[−0.369; 1.171]  
 F0.558[−0.304; 1.438]  
S4S (mm)    
 M1.855[−2.590; 6.300]−2.413[−6.449; 1.623]
 F1.682[−4.243; 7.607]−5.211*[−9.768; −0.654]
Waist (cm)    
 M1.293[−0.683; 3.269]−0.814[−1.919; 0.291]
 F1.100[−1.191; 3.391]−1.449*[−2.829; −0.069]
WHR    
 M−0.640[−1.649; 0.369]−1.096*[−2.092; −0.100]
 F1.644[−0.445; 3.733]1.162[−0.696; 3.020]
SS/S4S    
 M0.355[−0.993; 1.703]−0.277[−1.633; 1.079]
 F−3.065[−4.807; −1.323]−3.579[−5.296; −1.862]
Lean thigh (cm2)    
 M5.629[−2.209; 13.467]0.319[−3.848; 4.486]
 F10.118[−2.773; 23.009]2.266[−8.077; 12.609]

The crude model showed that in women only the relationship between birth weight and adult SS/S4S was statistically significant. When adding adult body mass to the model, the associations between birth weight and the outcome variables were either becoming negative or strongly negative. In women, results of the adjusted model showed statistically significant associations between birth weight and adult S4S, waist circumference, and SS/S4S. This indicates that, in women, lower birth weight was related to higher adult subcutaneous fat mass (S4S), higher waist circumference, and a more truncal fat distribution. In men, the same trend was observed. However, only the relationship between birth weight and adult WHR reached statistical significance.

Also, in men, current body mass was a significant effect modifier in the relationship between birth weight and the S4S; this means that the negative relationship between birth weight and the S4S was becoming less negative when adult body mass was larger.

Moreover, in women, time of measurement was a significantly associated effect modifier in the relationship between birth weight and the adult SS/S4S ratio. This means that the negatively associated effect of birth weight on the SS/S4S ratio was becoming less negative, or weaker, with time.

Additional analyses were performed to assess the linearity of the relationships considered above. Therefore, tertiles of birth weight were used as predictor variables. In the GEE analyses, the lowest tertile was used as the reference group. Results of the adjusted model of these additional GEE analyses are presented in Table 3.

Table 3.  Results of GEE analyses in men and women with tertiles of birth weight; lowest birth weight tertile is used as reference
  MenWomen
 Tertilesβ95% CIβ95% CI
  • *

    p ≤ 0.05.

  • p ≤ 0.01.

  • p ≤ 0.001.

BMI1Ref Ref 
 20.468[−0.526; 1.462]−0.024[−1.049; 1.001]
 30.448[−0.479; 1.375]0.758[−0.318; 1.834]
S4S (mm)1Ref Ref 
 22.268[−2.883; 7.419]−3.224[−8.563; 2.115]
 3−2.660[−8.128; 2.808]−6.422*[−12.004; −0.840]
Waist (cm)1Ref Ref 
 2−0.615[−2.152; 0.992]−0.829[−2.358; 0.670]
 3−1.078[−2.605; 0.449]−1.270[−2.863; 0.323]
WHR (×100)1Ref Ref 
 2−1.843*[−3.466; −0.220]2.818[−3.489; 9.125]
 3−1.123[−2.699; 0.453]0.357[−1.417; 2.131]
SS/S4S1Ref Ref 
 2−0.279[−2.480; 1.922]−3.204[−5.305; −1.103]
 3−0.586[−2.634; 1.462]−4.768[−6.893; −2.643]
Lean thigh (cm2)1Ref Ref 
 2−0.964[−7.152; 5.224]3.623[−2.631; 9.877]
 3−3.252[−9.093; 2.589]3.790[−3.437; 11.017]

It can be concluded from the regression coefficients presented in Table 3 that the relationships in men between birth weight and adult BMI and between birth weight and adult WHR are not linear. The regression coefficients suggest a threshold effect and not a linear effect, meaning that a birth weight lower than a cut-off point is related to adverse outcomes, but the effect levels off as birth weight increases. The same holds for the relationships between birth weight and adult S4S and lean thigh mass in women.

The relationship in men between birth weight and adult S4S and in women between birth weight and adult WHR seems more hyperbolic, with the middle tertile having the highest regression coefficients.

Furthermore, in men, the relationship with waist circumference and with adult SS/S4S ratio and, in women, the relationship with adult S4S and with adult waist circumference can be considered linear, because the regression coefficients of the middle and highest tertile increase equally.

Discussion

The aim of this study was to investigate the relationship between birth weight and the development of the distribution of subcutaneous fat and body composition at adult age. The results support what a few other studies have already reported (2, 5), namely that under correction of adult body mass, birth weight is negatively associated with most indicators of adult body fatness and adult body fat distribution. The overall trend in these results was that lower birth weight, corrected for adult body mass, was associated with a central pattern of subcutaneous body fat in adulthood, indicated by an increased SS/S4S ratio and more fat stored abdominally (waist circumference). Despite the latter finding, no relationship was observed between birth weight and adult WHR, which is also considered a measure of body fat distribution, although some authors cast doubt on whether WHR is a correct anthropometric measure for assessing central adiposity. It has been suggested that hip circumference may be strongly affected by pelvic structure, including muscle mass (24, 38), whereas waist circumference measures predominantly abdominal fat. When waist and hip circumferences are combined in a ratio, any of the WHR values can be heterogeneous with regard to waist and hip circumferences; therefore, WHR values do not necessarily correlate with the amount of abdominal fat (38). This can explain why, in this study, no associations were found between birth weight and adult WHR, but they were found between birth weight and adult waist circumference.

No relationship was found between birth weight and adult BMI. BMI, however, is only an indirect measure of body fatness, because it also includes lean and bone mass. This study supports the idea that BMI is not an appropriate measure of body fatness, and that when using BMI as an outcome, it is possible that important associations will not be detected (39). Birth weight is related to total adult fat mass as measured by S4S, but it is not related to BMI. In women, but not in men, this can partly be explained by the weak positive association between birth weight and adult muscle mass.

Results from studies in newborns suggest that preterm small-for-gestational age infants store excess calories as fat, whereas protein reserves in the form of muscle mass remain low (35, 40, 41). There is no reason to believe that this same pattern does not apply to newborns born at term with low birth weight (41). If this pattern continues during life, this can explain why subjects with lower birth weight have higher fat mass and reduced muscle mass. In contrast with the findings of Kahn et al. (14), a positive association between birth weight and adult lean thigh mass in men was not observed in this study. Therefore, it can also be suggested that in men, birth weight is not associated with muscle mass, and with the current data, no clear explanation can be given why men with lower birth weight have a higher S4S but do not have higher BMI.

The fact that adult fat mass is more centrally distributed in those with lower birth weights, especially in women, can be attributed to altered hormonal status (17). It is possible that factors influencing low birth weight at term may also be associated with fat patterning in later life, e.g., sustained adrenal overactivity initiated by early growth restraint (42). Recent studies have shown that the hypothalamic-pituitary-adrenal axis (HPA) might play an important role; exposure of pregnant rats to a variety of stressors, including low-protein diets, resulted in persistent changes in HPA activity in their offspring. It is also known that, in humans, an increase in HPA activity and the consequent rise in circulating cortisol concentrations are associated with the pathogenesis of the metabolic syndrome (43), of which abdominal obesity is a symptom.

As referred to before, in many studies, a cut-off point for birth weight of 2.5 kg is used to define a group of “low birth weight” subjects (15, 34, 35, 36, 37). This suggests that a threshold effect exists, with an adverse effect for those below the cut-off value and with a leveling-off effect with increasing birth weight above the cut-off value. Considering regression coefficients for the middle and the highest tertile in current analyses, arguments for a threshold effect could be observed for adult BMI and adult WHR in men and for SS/S4S ratio in women, because regression coefficients for the middle and the highest tertile have the same strength. This is true despite the fact that in this study population the mean birth weight of the lowest birth weight tertile ranged from 2.20 to 3.31 kg (mean, 3.05 kg) in men and from 2.10 to 3.18 kg (mean, 2.89 kg) in women. These values are not really low compared with the low birth weight groups in other studies. Taking this into account, current findings can be considered an indication for a threshold effect. However, to find out what the cut-off point should be, more analyses are needed in which different groups of birth weight are compared. For the other outcome variables, no arguments were found to support this phenomenon.

Although some longitudinal studies have been performed to study effects of low birth weight on adult health (18, 44), no study used special statistical techniques to analyze the data. A special feature of the current study is that a statistical technique was used that had the advantage of studying all available data at the same time. Results are based on measurements at two to four different time points during adult life, which give more valid outcomes than results based on just one measurement. From the results of this study, however, it can be concluded that the obtained results are stable. This is supported by the fact that in only one relationship, which was the relationship between birth weight and SS/S4S in women, a modifying effect of time of measurement was observed. Finally, from the results of this study, it can be concluded that low birth weight is associated with unfavorable body composition and subcutaneous body fat distribution at adult age, especially for women, which, in turn, is related to metabolic diseases (6, 7, 8, 10, 11).

Acknowledgment

The AGAHLS is supported by multiple grants from the Dutch Prevention Fund (ZON), Dutch Heart Foundation (NHS), Dutch Ministry of Education and Science, Dutch Ministry of Well Being, Public Health and Sports (VWS), the Dairy Foundation on Nutrition and Health (ZVG), Dutch Olympic Committee/Dutch Sports Foundation (NOC/NSF), Scientific Board Smoking and Health, and Heineken Inc. We would like to thank all the men and women who participated in the AGAHLS.

Footnotes

  • 1

    Nonstandard abbreviations: AGAHLS, Amsterdam Growth And Health Longitudinal Study; BMI, body mass index; S4S, sum of the four skinfolds; SS, two truncal skinfolds; WHR, waist-to-hip ratio; GEE, generalized estimating equation; HPA, hypothalamic-pituitary-adrenal axis.

Ancillary