点击显示 收起
1 From the Medical Research Council Environmental Epidemiology Unit, University of Southampton, Southamption, United Kingdom (AAS, HES, EMD, HJG, SLD, CC, DJB, and DIP), and University Geriatric Medicine, University of Southampton, Southampton, United Kingdom (AAS)
2 Supported by the Medical Research Council, United Kingdom. 3 Address reprint requests to A Aihie Sayer, MRC Environmental Epidemiology Unit, University of Southampton, Southampton SO16 6YD, United Kingdom. E-mail: aas{at}mrc.soton.ac.uk.
ABSTRACT
Background: Size in early life is related to adult body mass index, and early environmental influences have been proposed to have lifelong consequences for obesity. However, body mass index also reflects fat-free mass, and few studies have examined the relation between size in early life and direct measures of body composition in older people.
Objective: We investigated the associations of birth weight and weight at 1 y of age with body composition in older men.
Design: We carried out a retrospective cohort study in Hertfordshire, United Kingdom. Men who were born between 1931 and 1939 and for whom there were records of birth weight and weight at 1 y of age (n = 737) participated in the study. The main outcome measures were adult body mass index, fat-free mass, and fat mass.
Results: Birth weight was significantly and consistently positively associated with adult body mass index and fat-free mass but not with measures of adult fat mass. In contrast, weight at 1 y of age was associated with adult body mass index, fat-free mass, and fat mass.
Conclusions: The consistently reported positive relation between birth weight and adult body mass index may reflect prenatal and maternal influences on fat-free mass rather than on fat mass in older people. The postnatal environment may be more influential than prenatal factors in the development of obesity in later life.
Key Words: Body composition body mass index fat-free mass birth weight programming
INTRODUCTION
The antecedents of adult obesity are of considerable interest, and the relation between weight at birth and subsequent body mass index (BMI) has been investigated in many studies. Findings from a comprehensive systematic review confirmed the positive association between birth weight and BMI in both children and adults (1), and it has been proposed that both the prenatal environment (2) and maternal factors (3) may have lifelong consequences for obesity. BMI is frequently used as a marker of adult obesity because it is widely available and is correlated with direct measures of adiposity in children and younger adults (4). However, BMI also reflects fat-free mass and is more difficult to interpret when the relative proportions of fat, muscle, bone, and organ mass change, eg, during childhood (5) and in later life (6, 7).
Findings from the few studies that examined the relation between birth weight and more direct measures of adiposity are less consistent. Two studies showed positive relations between birth weight and both BMI and skinfold thickness in young children (8, 9), but a third study in adolescent women showed discordance: the relation between birth weight and skinfold thickness was negative or absent, whereas the relation between birth weight and BMI was positive (10). A further study in young adults showed a negative association between birth weight and skinfold thickness but failed to find any relation between birth weight and BMI; however, only recalled birth weight was available in this cohort (11). A recent study in children and adolescents in which whole-body dual-energy X-ray absorptiometry, skinfold-thickness measurements, and bioelectrical impedance analysis were used showed that increases in birth weight were associated with increases in fat-free mass but not in fat mass in both groups independently of age, sex, height, pubertal stage, socioeconomic status, and physical activity (12). A study in young Pima Indian adults had similar findings (13).
There is just one report of a study that investigated the relation between birth size and body composition in older people, and this study also showed significant associations of birth weight with muscle mass and bone mass but no association between birth weight and fat mass (14). In the present study, we explored these associations further by measuring anthropometric variables and body composition in a group of men aged 5970 y who had information on both prenatal and infant growth and participated in the Hertfordshire Cohort Study.
SUBJECTS AND METHODS
Study population
From 1911 to 1948, midwives collected detailed records, including information on birth weight and weight at 1 y of age, on infants born in the county of Hertfordshire, United Kingdom. The records for the persons born from 1911 to 1930 were used in a series of studies linking early growth to health in later life. In 1998 a younger cohort was recruited to participate in studies examining the interactions between early life, adult diet and lifestyle, and genetics as determinants of adult disease. Men (n = 1760) who were born in Hertfordshire between 1931 and 1939 and were still living in East Hertfordshire were traced with the aid of the National Health Service central registry in Southport and were confirmed as being currently registered with a general practitioner in Hertfordshire.
Permission to contact 1397 (79%) subjects was obtained from the general practitioners. Of those subjects, 768 (55%) agreed to take part in a home interview in which trained nurses collected information on their medical and social history. Of the subjects who were interviewed at home, 737 (96%) subsequently attended a clinic for several investigations. Anthropometric measurements included measurements of height, sitting height, weight, and waist, hip, midupper arm, and midthigh circumferences (15). Skinfold thicknesses were measured in triplicate with the use of Harpenden skinfold calipers (British Indicators, Luton, United Kingdom) at the triceps, biceps, subscapular, and suprailiac sites on the nondominant side (16). Grip strength was measured 3 times on each side by using a Jamar handgrip dynamometer (Promedics Ltd, Blackburn, United Kingdom) (17). Intraobserver and interobserver studies were carried out at regular intervals during the fieldwork to ensure comparability of measurements within and between observers. The study received ethical approval from the North and East Hertfordshire Local Research Ethics Committee, and all subjects gave written informed consent.
Statistical methods
BMI was calculated as weight (in kg)/height2 (in m). Leg length was calculated by subtracting sitting height from standing height. The averages of the triplicate skinfold-thickness measures at each site were taken, and the percentage of body fat was derived from the 4 average skinfold-thickness measures according to the method proposed by Durnin and Womersley (18). Fat mass was derived by multiplying body weight by percentage of body fat. Fat-free mass was estimated by subtracting fat mass from body weight. Waist-to-hip ratio was expressed as a proportion. The best of the 6 grip measurements were used in the analyses. Normality of variables was assessed, and BMI, weight, hip circumference, triceps skinfold thickness, biceps skinfold thickness, and fat mass were loge transformed. All statistical analyses were carried out by using STATA release 7 (Stata Corp, College Station, TX).
For presentational purposes, means and SDs of anthropometric variables were derived according to quintiles of birth weight and weight at 1 y of age. Geometric means and SDs were calculated for the loge-normally distributed variables. However, statistical tests of association were based on continuously distributed variables for size in early life and adult anthropometric measures. Multiple regression analyses (Pearson's pairwise and partial correlation coefficients) were used to investigate the relations between early size and adult measures without or with adjustment for age at clinic visit, social class at birth and current social class, smoking status, alcohol consumption, and physical activity. Because of the number of tests carried out and the large sample size, a stringent 1% significance level was used to screen for important correlations between size in early life and adult anthropometric variables.
RESULTS
Subject characteristics
The mean (± SD) age of the 737 subjects who attended the clinic was 64.3 ± 2.6 y. The descriptive data for their adult anthropometric variables and body composition are summarized in Table 1. The mean birth weight of the subjects was 3.5 ± 0.6 kg, and the mean weight at 1 y of age was 10.2 ± 1.1 kg. One hundred thirteen (15.3%) subjects were of the nonmanual social class at birth, 579 (78.6%) were of the manual social class at birth, and social class at birth was unclassified for 45 (6.1%) subjects. The proportion of subjects who were currently of the nonmanual social class was 36.5% (269), and 57.8% (426) were currently of the manual social class. Current social class was unclassified for 5.7% (42) of subjects.
View this table:
TABLE 1. Adult anthropometric and body-composition measures1
Two hundred thirty-eight (32.3%) subjects reported that they had never smoked, 375 (50.9%) were ex-smokers, and 124 (16.8%) were current smokers. Forty (5.4%) subjects were nondrinkers. Seventy-eight (10.6%) subjects reported very low weekly alcohol consumption [<1 unit (1 unit = 284 mL beer, 125 mL wine, 50 mL fortified wine, or 25 mL spirits)]; 294 (39.9%), low consumption (110 units); 149 (20.2%), moderate consumption (1121 units); 90 (12.2%), fairly high consumption (2235 units); and 86 (11.7%), high consumption (>35 units). The mean activity score (ranging from 0 to 100, with higher values indicating greater habitual activity levels) was 62.7 ± 15.4.
Relations of birth weight and weight at 1 y of age to adult anthropometric and body-composition measures
The relations of birth weight to adult anthropometric and body-composition measures are shown in Table 2. Birth weight was correlated strongly and positively with fat-free mass; height; sitting height; leg length; weight; hip, midthigh, and midupper arm circumferences; and grip strength. Birth weight was moderately positively correlated with BMI and waist circumference. Birth weight was not correlated with any of the 4 skinfold-thickness measures, percentage of body fat, or waist-to-hip ratio, and a weak significant association with fat mass was present only after adjustment for age, social class, and adult lifestyle factors.
View this table:
TABLE 2. Relations of birth weight to adult anthropometric and body-composition measures1
Similar patterns were observed for associations of weight at 1 y of age with anthropometric and body-composition measures in later adulthood (Table 3). Weight at 1 y of age was correlated strongly with height, sitting height, leg length, weight, hip and midthigh circumferences, fat-free mass, and grip strength. Weight at 1 y of age was moderately associated with BMI in the adjusted model. However, although weight at 1 y of age was not associated with adult skinfold thicknesses, percentage of body fat, or waist-to-hip ratio, it was correlated significantly and positively with fat mass in both the unadjusted and adjusted models.
View this table:
TABLE 3. Relations of weight at 1 y of age to adult anthropometric and body-composition measures1
DISCUSSION
In the present study, we showed that birth weight was significantly positively associated with BMI and fat-free mass but not with any of the measures of adiposity in this group of older men. In contrast, weight at 1 y of age was related to fat-free mass and fat mass. These findings show that relations with BMI may not adequately reflect the effects of influences that act differentially on different body compartments. This is particularly relevant to the stages of life when the relative proportions of fat, muscle, bone, and organ mass change, eg, in later life when the aging process affects body composition (6, 7).
Our results suggest that the relation between birth size and adult BMI in this cohort of older men reflects an association between higher birth weight and higher fat-free mass rather than higher fat mass. It is not possible to determine from these data whether the association of birth weight with fat-free mass reflects an association with muscle, bone, organ mass, or a combination of all 3. However, BMI is a strong predictor of skeletal muscle (19), and the strong relations of birth weight to grip strength, arm circumference, and thigh circumference, all of which are highly correlated with muscle mass (20, 21), that were shown in the present study suggest an association with muscle. This suggested association is consistent with the results of previous studies in older people that showed significant associations of size at birth with adult muscle mass and bone mass (22, 2325) and grip strength (26, 27). Evidence from animal models has shown that both prenatal and postnatal undernutrition may reduce the number of secondary muscle fibers and that this effect may not be reversed by subsequent improved nutrition (28). The relation between small size at birth and adult muscle strength may reflect early nutritional programming of the number of muscle fibers.
The divergence in findings for birth weight and weight at 1 y of age needs further consideration. Weight at 1 y of age was not related to the 4 skinfold-thickness measures, derived percentage of body fat, or waist-to-hip ratio but was significantly correlated with fat mass. The lack of consistency across the measures of adiposity suggests a chance finding, but an alternative explanation is that the correlation between weight at 1 y of age and fat mass provides preliminary evidence for the influence of the postnatal environment on the development of adult adiposity. In contrast, previous research on young men conceived during a wartime famine in the Netherlands showed that exposure to undernutrition in early pregnancy resulted in increased rates of obesity at 19 y of age, whereas later exposure in the third trimester or early postnatal life resulted in a reduced likelihood of obesity (29). However, subsequent follow-up of the men at 50 y of age showed that this effect on adiposity did not persist (30). Detailed information on infant feeding was not available in our cohort, but this is an area that warrants further investigation.
Several potential caveats should be considered in the interpretation of our findings. Losses to follow-up occurred during tracing and in gaining subjects' consent to participate, and response bias may have been introduced. However, we were able to characterize those who did not take part in the study in several ways. No substantial differences in birth weight or weight at 1 y of age were found between the subjects who were traced and eligible to participate in the study but did not and those who had a home interview. Furthermore, no major differences in social class (at birth or current), alcohol consumption, activity level, or age were found between interviewed subjects who did or did not visit the clinic. The proportion of current smokers was lower among the interviewed subjects who visited the clinic (16.8%) than among those who declined to visit (32.3%). This suggests that a "healthy subject" effect may have occurred in our study. However, our comparisons were internal; therefore, unless the relation between early size and adult anthropometric measures differed between those who did or did not visit the clinic, no bias should have been introduced.
We carried out several statistical tests in the present study, and some positive findings are likely to have arisen by chance alone. To address this possibility, we used a stringent 1% significance level to screen for important associations. Consistency with findings in studies of younger people suggests that this approach was effective. Alternative explanations of the associations observed in our study include confounding by social class, smoking status, alcohol consumption, or physical activity level. However, the relations remained after adjustment for all these factors. We were unable to determine whether the associations were independent of maternal factors other than social class because characteristics such as maternal size were not available in this cohort. We suggest, therefore, that our results are internally valid. However, generalization of these findings to the wider population would require replication in other study populations.
Our findings provide evidence for a relation between birth weight and adult fat-free mass but not for an association between birth weight and fat mass in older people. The significant relation between birth weight and adult grip strength suggests that muscle is an important component of fat-free mass that is influenced by prenatal environmental and maternal influences. In contrast, weight at 1 y of age was associated with adult fat mass, and postnatal environmental factors, such as infant feeding, may be more important than are prenatal factors for the development of adult adiposity. This is an area for further investigation. We conclude that the well-described positive relation between size at birth and adult BMI in older people reflects an association of size at birth with adult fat-free mass rather than fat mass.
ACKNOWLEDGMENTS
All authors participated in the conception, design, and conduct of the study. HES carried out the statistical analyses, and AAS drafted the first version of the manuscript. All authors contributed to and approved the final version of the manuscript. None of the authors had any conflicts of interest in connection with this article.
REFERENCES