Body mass index in young Dutch adults : its development and the etiology of its development

Follow-up studies of long duration have shown a U-shaped relationship between mortality/morbidity and the body mass index (BMI, weight/height ²). The risk to health posed by obesity appears to be larger in younger subjects than in older subjects. Though this might suggest that a moderate weight gain after the termination of growth may not increase the risk to health, this contention is not supported by other observations (Chapter 1). These are: (1) the range of relative weight associated with minim= mortality does not seem to shift to higher values with increasing age, at least not in men. (2) Diseases associated with overweight at younger ages need not be the same as those associated with overweight at older ages. (3) Cohort effects may bias the age trend in the risk associated with overweight. (4) Longitudinal studies on adults show that changes in body weight are associated with changes in risk factors of diseases. (5) Anthropometric studies show that even at the same weight, the fat mass increases with increasing age. In addition, several studies suggest that weight gained during adulthood mainly accumulates in the abdomen, which means a shift to a fat distribution more strongly associated with risk to health. The risk to health therefore does not seem to remain constant if body weight increases during adulthood.

As part of an extensive mixed-longitudinal study (Chapter 2), this thesis deals with the development of the BMI and the etiology of this development in young adults. In addition, one methodological study is described. These studies build m cross-sectional observations that were made in the first examination of the study.

In spring 1980, all residents of Dutch nationality in the county of Ede, born in 1948-1950, 1953-1955, 1958-1960 were invited by mail to participate in the study. Participants in spring 1980 (N=3936) were followed for four years (1980-1984) in two groups, the follow-up group (N=1670) and the control group (N=2266). From spring 1981 onward the follow-up group was approached twice a year in seven consecutive examinations. The control group was re-measured in spring 1984 only. At each examination a questionnaire was completed and body weight in light indoor clothing and without shoes was measured to the nearest 0.5 kg. Body height was measured to the nearest 0.1 cm in spring 1980. Subgroups of the follow-up and control groups were additionally examined for the studies described in Chapter 7 and 8.

The development of the BMI is dealt with in Chapter 3, in which the mixed-longitudinal design of the study was utilized, and Chapter 4, in which the data were analyzed longitudinally. Over the four years, the change in BMI of the follow-up group was the same as that of the control group. Prom 19 to 35 years of age the median BMI increased from 22.1 kg/m ²to 24.4 kg/m ²in men and from 21.1 kg/m ²to 23.0 kg/m ²in women (Chapter 3). This increment in BMI gave rise to an incidence of moderate overweight (BMI>25 kg/m ²) which increased from 4.8% per 4 years to 15.5% per 4 years in men, and was more stable (5.3-5.7% per 4 years) in women (Chapter 4). Thus, especially in men, young adulthood appears to be a critical period for the development of overweight.

The age-reference curves (Chapter 3) suggested that the variation of the BMI over subjects was independent of age. In accordance, the longitudinal analyses showed that the initial BMI and the rate of gain in BMI were unrelated in men and only slightly negatively related in women. Thus, overweight subjects did not appear to gain more or less body mass than non-overweight subjects.

The within-subjects standard deviation of the yearly measured BMI was 0.69 kg/m ²in men and 0.74 kg/m ²in women (Chaper 4). This fluctuation was larger in overweight subjects than in normal-weight subjects and was larger in subjects with a larger long-term change (gain or loss) in BMI than in subjects who had a more stable BMI. The considerable fluctuation in BMI may mask the gradual long-term increment in BMI and may thus hinder young adults from being aware that they are becoming overweight.

The effect of dieting is dealt with in Chapter 5. After two years of follow-up, the average decrease in body mass as a result of dieting was -0.5 kg/m ²(95% confidence interval (CI):-1.0,+0.0) in men and -0.4 kg/m ²(95%CI:-0.8,+0.0) in women. This approximately amounts to an average weight reduction of 1.5 kg in men and 1.0 kg in women. The effect of dieting was most pronounced in men whose initial BMI was high, who dieted during the summer, or who dieted more frequently, and in women who were older than 30 or who dieted on medical advice. Though subjects who have already developed severe overweight may achieve a larger reduction in the risk to health by losing the same amount of body mass as subjects who are only moderately overweight, it is unlikely that the benefit is large unless they persist. However, the effect of dieting may be enough to prevent the age-related gain in body mass and thus dieting may especially play a part in the prevention of overweight.

Chapter 6 deals with the changes in BMI in relation to number of life events experienced. During the first year of follow-up several subgroups of men and women who experienced many life events and several subgroups of men who experienced few life events showed a gain in body mass. After another year of follow-up this gain in body mass had disappeared in almost all subgroups; compared with dieting ran with an intermediate number of life events, dieting men with few life events gained +1.3 kg/m ²(95%CI:+0.0,+2.6) more body mass, whereas dieting men with many life events gained +1.8 kg/m ²(95%CI: +0.5,+3.0) more body mass. Thus, life events seem to have an impact on the etiology of overweight in men.

One life event in women is considered more specifically (Chapter 7); i.e. pregnancy. Women who breastfed their child for more than two months gained more body mass than was expected from aging. Nine months postpartum they were +0.6 kg/m ²(90%CI: +0.1,+1.0) heavier than expected. This difference was slightly smaller (not significantly) in women who breastfed their child for 0-2 months. Women who used bromocriptine to stop lactation lost body mass unexpectedly.

In a methodological study (Chapter 8) the impact of adjustment of the BMI for body diameters (i.e. knee, wrist, elbow, shoulder, pelvis and hip) on the prediction of body fatness was examined. In addition to what was explained by the BMI, 6% of the variation in body fatness was explained by the shoulder diameter in men, and also 6% by the knee diameter in women. This improvement in estimating body fatness by the diameters, used as continuous variables, was considered too small to justify frame categories being included in weight-height tables. Instead of frame size, fat distribution might be a more useful attribute in these tables.

In the last chapter (Chapter 9) several aspects relating to the validity of the mixed-longitudinal study are discussed. The main conclusions of this thesis are:
- Young adulthood seems to be a critical period in the development of moderate overweight, expecially in men.
- The age-related gain in BMI of overweight young adults and of young adults of normal weight is similar.
- Overweight subjects are unlikely to decrease their risk to health markedly by dieting as generally practiced, unless they persist. Dieting may be more effective in preventing overweight.
- The experience of many life events may play a part in the etiology of overweight in men, as does the experience of few life events.
- Women who breastfeed their child for more than two months may gain more body mass than expected from aging. - In weight-height tables the fat distribution might be a more useful attribute than frame size categories.