Child-to-Adult Body Mass Index and Height Trajectories: A Comparison of 2 British Birth Cohorts

Abstract

Markers of growth and changes of body mass index (BMI) are associated with adult chronic disease risk. To better understand such associations, the authors examined the 1946 (n ≈ 5,300) and 1958 (n ≈ 17,000) British birth cohorts to establish how child-to-adult height and BMI have changed across generations. Individuals born in 1958 were no heavier at birth than those born in 1946, but they were taller in childhood by about 1 cm on average, grew faster thereafter, and were 3–4 cm taller by adolescence. The 1958 cohort achieved adult height earlier and were taller by 1 cm, an increase that was entirely due to their longer leg length. BMI trajectories diverged from early adulthood, with a faster rate of BMI gain in the 1958 cohort than in the 1946 cohort, although the mean BMI at 7 years and rate of childhood gain had not shown an increase. By midadulthood, the 1958 cohort had on average a greater BMI (1–2 kg/m²), larger waist (6–7 cm) and hip (5 cm) circumferences, and a higher prevalence of obesity (25.1% vs. 10.8% in males and 23.7% vs. 14.8% in females). Changes in height and adiposity over a relatively short period of 12 years suggest the likelihood of opposing trends of influences on later disease risk in these populations.

It is now recognized that growth and development at different life stages are associated with adult chronic disease (1). In particular, increased body mass index (BMI) has an important influence on adult morbidity and mortality, notably for cardiovascular disease, diabetes, and some cancers (2–5). In many populations, the prevalence of obesity continues to increase across all ages. Because the obesity epidemic will have affected different generations at different life stages, individuals who are obese in adult life are likely to have had different weight gain trajectories. Older generations may have gained weight rapidly in adulthood after a relatively slower gain in childhood or have had a steady weight gain over a longer period, while in younger generations obesity may have commenced at an early age in childhood and continued through to adult life. Yet, it is unclear how duration of exposure to a high BMI and BMI gain affect health in the longer term. A longer duration of adiposity may increase the risk of type 2 diabetes (6, 7), although the role of prolonged obesity is still unclear (8).

Because different BMI trajectories are likely to have implications for morbidity and mortality in later life, it is important to understand the development of BMI throughout the life course and how BMI trajectories have changed across recent generations experiencing the obesity epidemic at different life stages. Investigating BMI trajectories in separate cross-sectional samples is likely to underestimate the rate of BMI increase with age, because of the secular increase in BMI over time (9). Moreover, individual trajectories cannot be obtained from cross-sectional data that are collected from different individuals who were born and have grown up at different times. The relation between a single measure of BMI in adulthood and disease outcome is also likely to be affected by selective survival and reverse causality in old age. Hence, BMI trajectories should be investigated in longitudinal studies, ideally before disease onset. Where possible, trends in other markers of adiposity, such as central adiposity, should be considered given its strong independent association with risks for diabetes and cardiovascular disease (10, 11).

Trends in other components of physical development, such as birth weight, height, and leg length, are also of interest because of their associations with adult chronic disease. For example, low birth weight and short stature have been found to be associated with cardiovascular disease and mortality in later life (12–14). Recent evidence suggests that leg length, the component of height most sensitive to early environment (15, 16), has a stronger association than total height with risk of cardiovascular mortality and insulin resistance in adulthood (13, 17).

Before investigating whether growth and BMI trajectories have a differing impact on disease risk in different generations, our aim here is to establish the extent to which child-to-adult BMI and height have changed across 2 British cohorts, born in 1946 and 1958. Previous comparisons of these cohorts reported modest increases in the prevalence of childhood obesity (18) and height (19). Here, we extend the comparisons to longer periods of the life span to age 53 years (1946 cohort) and 45 years (1958 cohort). We aim to 1) determine the extent to which a range of growth parameters have changed across the 2 cohorts, including birth weight, child-to-adult height and adult leg length, overweight/obesity from childhood to adulthood, and adult central adiposity; and 2) compare how BMI trajectories have changed across these generations.

MATERIALS AND METHODS

Study samples

The 1946 British birth cohort includes children from a socially stratified sample of legitimate, single births (n = 5,362) in 1 week of March 1946 in England, Wales, and Scotland. Cohort members were followed up on 22 occasions from birth to age 53 years. At 53 years, 3,035 participants provided information (20). Contact was not attempted for the 1,979 individuals who were living abroad (11% of the original cohort), had previously refused (12%), were untraced since the last contact at 43 years (5%), or had died (9%).

The 1958 British birth cohort includes all children born in 1 week of March 1958 in England, Wales, and Scotland. About 17,000 livebirths were followed up on 8 occasions, birth to age 45 years. Immigrants to Britain born during the week were incorporated into the follow-ups at 7, 11, and 16 years (n = 920). At 45 years, 11,971 cohort members were invited to participate in a medical assessment by a trained nurse (21); 9,377 participants provided information. Contact was not attempted for the 5,549 individuals who were living abroad (7% of the total cohort), had died (6.7%), or had previously refused or with whom there was no contact since childhood (16.2%). To ensure comparability of the 2 cohorts, we restricted the 1958 cohort sample to legitimate, singleton births and to nonimmigrants.

Measures

The birth weight of each individual was extracted from birth records within a few weeks of delivery for the 1946 cohort (to the nearest quarter of a pound) and was measured for the 1958 cohort (to the nearest ounce). To be consistent with the 1946 cohort, the birth weights for stillbirths or deaths within a month (n = 594) were excluded from the 1958 cohort.

In the 1946 cohort, height and weight were measured to the nearest inch or pound at 7, 11, and 15 years; to the nearest 0.5 cm and 0.5 kg at 36 and 43 years; and to the nearest millimeter and 0.1 kg at 53 years; these measurements were self-reported at 20 and 26 years of age. For females who were pregnant, the prepregnancy weight at 26 years (n = 63) and the current weight at 20 (n = 107), 36 (n = 30), and 43 (n = 10) years were recorded. Height and weight in the 1958 cohort were measured to the nearest inch and pound at 7, 11, and 16 years; to the nearest centimeter and 0.1 kg at 33 years; and to the nearest millimeter and gram at 45 years; measurements were self-reported at 23 years of age. For females who were pregnant, the prepregnant weight at 23 years and current weight at 33 years (n = 229) were recorded. For both cohorts, self-reported prepregnancy weights were used, but measurements during pregnancy were excluded. Trunk length, represented by sitting height, was measured (to the nearest 0.1 cm) at 43 and 53 years for the 1946 cohort and at 45 years for the 1958 cohort. Leg length (standing height − sitting height) was calculated. Birth weight and all heights and weights were converted to grams, centimeters, and kilograms, respectively.

BMI (kg/m²) was calculated at each age, from 7 years to adulthood, for both cohorts. Overweight and obesity in adulthood were defined as BMI ≥25 and BMI ≥30 kg/m² as recommended by the World Health Organization (22), and in childhood, they were defined by using age- and sex-specific international standards (23) corresponding to cutoffs at age 18 years. Waist and hip circumferences were measured in millimeters according to a standardized protocol at 43 and 53 years of age for the 1946 cohort and at 45 years of age for the 1958 cohort. Pregnant women were excluded.

Statistical analysis

Measurements were not always taken at the same exact ages in the 2 cohorts; the mean ages during childhood were 7.05, 10.86, and 14.54 years for the 1946 cohort and 7.36, 11.46, and 15.85 years for the 1958 cohort. These discrepancies could affect comparisons of childhood measurements across the 2 studies. Thus, we centered height and weight at 7, 11, and 15.5 years for all individuals by using predictions from linear regression models that assumed a linear age trend over short periods.

To determine whether growth parameters have changed between the 2 generations, we calculated the mean birth weight, height, weight, and BMI at each age from childhood to adulthood and the leg and trunk lengths and waist and hip circumferences in adulthood separately for each cohort. The mean height and prevalence of overweight/obesity were plotted against age to visualize the patterns of development throughout the life course. Because the 2 cohorts are independent samples, differences in mean height, weight, and BMI in childhood and adulthood (43–45 years) were examined by using 2-sample t tests.

Cross-sectional comparison of BMI was hindered by differences in the time points of measurement. Therefore, we estimated BMI trajectories from the exact ages at measurement. Because graphical representation of the mean BMI by age suggested a faster growth rate in childhood than in adulthood, we applied piecewise linear models with random coefficients (24–26) to the repeated BMI measures of the 1946 cohort (7, 11, 15, 20, 26, 36, 43, and 53 years) and the 1958 cohort (7, 11, 16, 23, 33, and 45 years). Such models allow individuals with incomplete outcome data to be included in analyses.

Let BM_Iij be the observed j th BMI for the ith individual, t_ij be the exact age of each measurement, and e_ij be independent error with a N(0, σ²) distribution. For each cohort and gender, we fitted 2 curves, one for “childhood” and one for “adulthood,” where knot t₀ is the age at which the time growth pattern changes. The indicator represents I_ij = 0 (age ≤ t₀) and I_ij = 1 (age > t₀). The model of a linear childhood trend and a quadratic adult trend is specified as follows:

with a random intercept (b_oi), fixed linear coefficients for childhood and adult age (b₁,b₂), and a quadratic coefficient for adult age (b₃, nonzero for females of the 1946 cohort). Knots were chosen at 20 years for males and 16 years for females on the basis of the profile likelihood functions. Estimated BMI trajectories are represented by the mean BMI at 7 years and the rates of BMI growth in childhood and adulthood.

The piecewise linear models were fitted in MLwiN (27). The distribution of BMI, especially in adulthood, is slightly skewed. We repeated analyses using log-transformed BMI, but the results did not alter our conclusions. Thus, the results using untransformed BMI are presented. In all analyses, the 1946 cohort sample was weighted to account for its stratified-sample design (20).

RESULTS

There was little difference in the mean birth weight, although there were a higher prevalence of low birth weight and a lower prevalence of very higher birth weight in the 1958 cohort compared with the 1946 cohort (i.e., 4.8% with <2,500 g and 1.7% with >4,500 g for the 1958 cohort compared with 4.3% and 2.5%, respectively) (P < 0.001). There were a small increase in the mean weight (0.4–0.5 kg) between cohorts at 7 years, no increase at 11 years, but then a greater increase at 15.5 years (6.6 kg for boys and 2.4 kg for girls). Thus, the weight gain was faster for the younger cohort during adolescence (11–15.5 years) compared with the 1946 cohort. Subsequently, in adulthood, the younger cohort were significantly heavier, by 7.1 kg (males) and 4.8 kg (females), at 43–45 years (Table 1).

Table 1.

Observed Mean for Anthropometric Measures for the 1946 and 1958 British Cohorts

	Age, years	Male				Female
		1946 Cohort a		1958 Cohort		1946 Cohort a		1958 Cohort
		Mean (Standard Deviation)	No.	Mean (Standard Deviation)	No.	Mean (Standard Deviation)	No.	Mean (Standard Deviation)	No.
Birth weight, gb		3,441 (562)	2,798	3,421 (515)	7,817	3,299 (520)	2,529	3,284 (495)	7,374
Weight, kg	7c	22.9 (2.9)	2,067	23.3 (3.5)	6,448	22.3 (3.4)	1,933	22.8 (3.9)	6,009
	11c	34.0 (5.8)	2,065	33.3 (6.8)	5,813	34.6 (7.3)	1,895	34.6 (7.7)	5,514
	15.5c	51.2 (9.5)	1,899	57.8 (10.2)	5,016	51.7 (9.1)	1,718	54.1 (8.5)	4,738
	20	70.7 (9.0)	1,862			57.6 (8.5)	1,759
	23			72.8 (10.4)	5,514			58.3 (9.0)	5,524
	26	73.8 (10.1)	1,823			59.1 (9.3)	1,782
	33			80.2 (13.0)	4,954			65.4 (13.3)	5,089
	36	76.7 (11.5)	1,640			62.2 (11.4)	1,657
	43–45d	79.3 (12.2)	1,618	86.4 (14.5)	4,146	66.5 (13.2)	1,619	71.3 (15.1)	4,137
	53	83.9 (13.5)	1,452			71.8 (14.6)	1,501
Height, cm	7c	119.8 (5.6)	2,147	121.1 (5.6)	6,438	118.9 (5.7)	1,993	120.0 (5.8)	5,998
	11c	140.1 (6.7)	2,082	140.8 (6.8)	5,861	140.1 (7.2)	1,913	141.1 (7.4)	5,533
	15.5c	165.2 (8.9)	1,913	169.4 (7.9)	5,039	157.9 (6.3)	1,726	160.8 (6.2)	4,746
	20	176.4 (6.6)	1,878			162.1 (6.3)	1,763
	23			177.4 (7.0)	5,568			162.3 (6.7)	5,579
	26	176.7 (6.5)	1,826			161.8 (6.4)	1,786
	33			176.9 (6.9)	6,351			162.6 (6.5)	6,630
	36	175.0 (6.5)	1,637			161.7 (6.1)	1,654
	43–45d	174.8 (6.5)	1,618	176.2 (6.7)	4,192	161.8 (5.9)	1,611	162.7 (6.2)	4,196
	53	174.2 (6.4)	1,453			161.2 (5.9)	1,508
Leg length, cm	43–45d	82.9 (5.5)	1,588	84.1 (4.7)	4,130	75.3 (4.8)	1,578	76.5 (4.2)	4,062
	53	82.5 (5.2)	1,435			74.5 (5.0)	1,479
Trunk length, cm	43–45d	91.9 (4.1)	1,588	92.1 (3.8)	4,131	86.6 (4.0)	1,585	86.3 (3.5)	4,062
	53	91.8 (4.4)	1,435			86.6 (4.1)	1,479
BMI, kg/m2	7c	15.9 (1.3)	2,057	15.8 (1.7)	6,293	15.7 (1.6)	1,920	15.8 (2.0)	5,847
	11c	17.2 (2.1)	2,050	16.7 (2.5)	5,764	17.5 (2.7)	1,887	17.3 (2.9)	5,469
	15.5c	19.6 (2.4)	1,881	20.1 (2.7)	4,998	20.7 (3.1)	1,700	20.9 (2.9)	4,709
	20	22.7 (2.5)	1,829			21.9 (2.9)	1,735
	23			23.1 (2.9)	5,486			22.1 (3.2)	5,504
	26	23.6 (2.9)	1,822			22.6 (3.2)	1,782
	33			25.6 (3.8)	4,947			24.6 (4.8)	5,086
	36	25.0 (3.2)	1,632			23.8 (4.1)	1,648
	43–45d	26.0 (4.1)	1,617	27.8 (4.3)	4,145	25.4 (4.9)	1,608	27.0 (5.5)	4,137
	53	27.6 (4.0)	1,452			27.7 (5.5)	1,496
Waist circumference, cm	43–45d	92.3 (9.7)	1,604	98.5 (11.2)	4,255	78.4 (11.4)	1,603	85.6 (12.9)	4,266
	53	98.1 (10.7)	1,454			86.2 (13.1)	1,507
Hip circumference, cm	43–45d	100.7 (7.0)	1,608	105.8 (7.5)	4,255	100.8 (10.4)	1,607	105.3 (11.3)	4,263
	53	104.2 (7.0)	1,453			106.3 (11.0)	1,505

Abbreviation: BMI, body mass index.

^aWeighted for sample stratification.

^bExcluding stillbirths or deaths within the first month from the 1958 cohort.

^cWeight, height, and BMI centered at 7, 11, and 15.5 years.

^dMeasures were taken at 43 years in the 1946 cohort and at 45 years in the 1958 cohort.

Members of the 1958 cohort were significantly taller than those of the 1946 cohort by 1.3 cm (boys) and 1.1 cm (girls) at 7 years; by 0.7 cm (boys) and 1 cm (girls) at 11 years; and by 4.2 cm (boys) and 2.9 cm (girls) at 15.5 years. Height growth was faster for the younger cohort, both in early childhood (before 7 years) and during adolescence (11–15.5 years) compared with the 1946 cohort (Table 1). Height gain after 15.5 years was less in the 1958 cohort, suggesting that they had achieved adult height earlier than the 1946 cohort did (Figure 1). The average increase in adult height of 1.4 cm (men) and 0.9 cm (women) between cohorts was due entirely to increases in mean leg length (Table 1).

Figure 1.

Height trajectories for the 1946 and 1958 British cohorts. Childhood height was centered at 7, 11, and 15.5 years; observed adult height (measured at 43 years in the 1946 cohort and 45 years in the 1958 cohort) was plotted at 20 years for males and at 18 years for females.

Figure 2 shows observed means and estimated trajectories for BMI in the 2 cohorts. As expected, BMI gain was significantly faster in childhood than in adulthood (P < 0.05) in both populations. BMI gain showed a linear trend in childhood in both cohorts and in adulthood in the 1958 cohort but a quadratic trend in adulthood for females in the 1946 cohort (Figure 2). In a comparison of the estimated BMI trajectories of the cohorts, among males, the mean BMI at 7 years was slightly lower in the 1958 cohort (15.55 vs. 15.73 kg/m²) (Table 2), although the rate of BMI gain was similar to that for the earlier generation born in 1946. Among females, there was no difference in the mean BMI at 7 years, but girls born in 1958 gained BMI at a slower rate in childhood than did those born in 1946 (0.588 vs. 0.660 kg/m² per year) (Table 2). In adulthood, the trajectories of BMI differed between cohorts and between genders. For males, the rate of adult BMI gain was 0.149 (95% confidence interval: 0.143, 0.155) kg/m² per year in the 1946 cohort, increasing significantly to 0.213 (95% confidence interval: 0.209, 0.217) kg/m² per year in the 1958 cohort (Table 2), such that by 45 years the mean BMI of the 1958 cohort was greater by 1.8 kg/m² (Figure 2). For females, the rate of BMI gain remained constant throughout adulthood (0.216 kg/m² per year) for the younger cohort, but for those born in 1946 there was a quadratic trend, such that the rate of BMI gain increased with age (Table 2). The rate of gain was slower in the older than the younger cohort from early adulthood, such that the difference in mean BMI widened from 0.18 kg/m² at 23 years to 1.19 kg/m² at 45 years (Figure 2). Therefore, the gender difference in mean BMI at 45 years was significantly greater in the 1958 cohort (1.1-kg/m² difference between men and women) compared with the 1946 cohort (0.5-kg/m² difference) (Table 2).

Figure 2.

Estimated body mass index (BMI) trajectories and observed mean BMI at different ages for the 1946 and 1958 British cohorts. The knot is at 20 years for males (childhood is 7–<20 years, adulthood is 20–53 years for the 1946 cohort and 20–45 years for the 1958 cohort) and at 16 years for females (childhood is 7–≤16 years, adulthood is >16–53 years and >16–45 years, respectively).

Table 2.

Parameter Estimates (Fixed Effects) for the 1946 and 1958 British Cohorts From a Piecewise Linear Model^a

	1946 Cohort		1958 Cohort
	Coefficient	95% Confidence Interval	Coefficient	95% Confidence Interval
Males
Intercept	12.069	11.969, 12.169	11.913	11.850, 11.976
Age (childhood, <20 years)	0.523	0.515, 0.531	0.519	0.513, 0.525
Age (adult, ≥20 years)	0.149	0.143, 0.155	0.213	0.209, 0.217
Mean BMI at 7 years	15.73	15.67, 15.79	15.55	15.50, 15.59
Mean BMI at 45 years	26.26	26.11, 26.41	28.08	27.96, 28.19
Mean BMI at 53 years	27.46	27.27, 27.64	29.92	29.78, 30.07
Females
Intercept	10.880	10.735, 11.025	11.311	11.231, 11.391
Age (childhood, ≤16 years)	0.660	0.646, 0.674	0.588	0.580, 0.596
Age (adult, >16 years)	−0.034	−0.053, −0.014	0.216	0.212, 0.220
Age2 (adult, >16 years)	0.0029	0.0026, 0.0032
Mean BMI at 7 years	15.50	15.43, 15.57	15.43	15.38, 15.48
Mean BMI at 45 years	25.78	25.57, 25.99	26.97	26.82, 27.12
Mean BMI at 53 years	27.86	27.60, 28.12	28.70	28.52, 28.88

Abbreviation: BMI, body mass index.

^aKnot is at 20 years for males and at 16 years for females in both cohorts.

Figure 3 shows that, in both cohorts, the prevalence of overweight/obesity changed little during childhood but increased rapidly in adulthood, with a faster rate of increase in the 1958 cohort. Prevalence rates during childhood were similar in both cohorts, with the exception of 7 years, when girls born in 1958 had a higher prevalence of overweight (9.5% vs. 7.6%) and obesity (2.8% vs. 1.1%). A divergence in prevalence of overweight started from the late 20s (25–29 years), while for obesity, the divergence started in the early 20s (20–25 years) (Figure 3). By midadulthood, 49.7% of men and 32.8% of women born in 1958 were overweight compared with 47.9% and 27.4%, respectively, of those born in 1946; 25.1% of men and 23.7% of women in the 1958 cohort were obese compared with 10.8% and 14.8%, respectively, in the 1946 cohort (Figure 3). Finally, the mean waist and hip circumferences in midadulthood increased significantly between cohorts, with the 1958 cohort having 6.2 cm (men) and 7.2 cm (women) larger waists. The differences in hip circumference were 5.2 cm (men) and 4.8 cm (women) (Table 1).

Figure 3.

Observed prevalence of overweight and obesity for the 1946 and 1958 British cohorts.

DISCUSSION

Our study highlights both differences and similarities in growth characteristics in 2 British cohorts born only 12 years apart. We found no evidence that the younger cohort members were heavier at birth than were the older cohort members. However, the younger cohort was taller by about 1 cm in childhood and, subsequently, had faster growth, such that these members were taller by 3–4 cm during adolescence. Because of the longer period of growth in the older cohort, the difference in adult height was about 1 cm, and this was entirely due to an increase in leg length in the younger cohort. Correspondingly for weight, the younger cohort gained weight at a faster rate from adolescence and continued to do so throughout adult life. As a result of differences in both height and weight gain, BMI trajectories differed between the cohorts. Participants in the 1958 cohort gained BMI more rapidly after early adulthood than those born in 1946 despite similar rates of BMI gain in childhood for males and slower rates for females. By midadulthood, individuals in the younger cohort had a larger BMI (1–2 kg/m²) and waist (6–7 cm) and hip (5 cm) circumferences compared with the older cohort. They also had a higher prevalence of obesity (25.1% vs. 10.8% in men and 23.7% vs. 14.8% in women). These changes are substantial over the relatively short period of 12 years between cohorts.

Secular trends in height and obesity at specific ages are well documented (28), but evidence on trends in growth trajectories is sparse. To our knowledge, this is the first study to compare BMI and height growth in large national cohorts, from birth throughout childhood, adolescence, and across adult life. Comparable measures of adult leg and trunk lengths and central adiposity provide additional information on how growth has changed between these 2 populations. Methodologically, piecewise regression models with random coefficients included individuals with incomplete data and provided estimates for mean BMI at different ages and separate slopes at different life stages. Childhood height was centered at 7, 11, and 15.5 years, so that childhood and adult height and growth between these life stages could be compared across cohorts. However, the height and BMI trajectories were based on a small number of widely spaced measures on individuals and, thus, do not fully capture the true underlying patterns of change. Nonetheless, the trajectories presented illustrate the magnitude of height and BMI differences at specific points in development. It is also possible that survival bias could have affected the results, although the number of deaths by midadulthood was relatively small and unlikely to greatly affect average BMI curves.

The interrelated growth parameters included in the study are important because each has been related to later chronic disease. Notably, low birth weight has been found to be associated with cardiovascular disease (12); short stature, short leg length, and impaired postnatal growth have been associated with cardiovascular disease (13, 14, 17, 29) and diabetes (17, 30), and tall stature has been associated with breast cancer (31). High BMI and excessive BMI gain at different life stages are associated with cardiovascular disease and its risk factor (3, 32), diabetes (7), and some cancers (33). However, not all evidence relating growth markers to health outcomes is consistent. For example, BMI in childhood was found to be unrelated to cardiovascular disease in adulthood in the Aberdeen children of the 1950s study (34). Comparing markers of growth and growth trajectories is of interest in this context.

Obesity in children has increased dramatically over the past few decades in most developed countries (35). Yet, in the 2 cohorts shown here, there was little increase in childhood BMI as demonstrated in our study and a previous report (18). However, the obesity epidemic during the 1980s has affected both generations since childhood, leading to excessive BMI gain in the 1958 study after early adulthood, compared with the 1946 study. The gap in BMI between the 2 generations widened steadily throughout adult life in men, while in women the widening occurred in early adulthood and the difference was stable thereafter. Thus, our findings are consistent with those of other studies (36), in which younger cohorts exhibited a faster rate of increase in overweight/obesity during adult life and became obese at younger ages than did their predecessors. The marked increase in overweight/obesity (from 58.7% to 74.8% in males and from 42.2% to 56.5% in females) over a 12-year period is comparable with findings from the Health Survey for England for the corresponding periods (from 56.6% to 72.1% and from 43.7% to 56.2%, respectively) (Dr. Denise Howel, Newcastle University, personal communication, 2007).

Given the higher prevalence of adult obesity and excessive weight gain in the 1958 cohort compared with their predecessors born in 1946 and the associations between these characteristics and cardiovascular disease, cancer, and diabetes risk (3, 7, 32, 33), we might expect trends in these diseases to mirror trends for BMI. However, our study also demonstrates increases in height and leg length, indicating improvements in the early life environment (16), which would tend to lessen the risks for cardiovascular disease although not for cancer. The increase in adult height between the 2 cohorts, by 1.2 cm per decade for men and 0.8 cm for women, is consistent with a rise of approximately 1 cm per decade previously reported for young adults in the same period (37). The effects of these opposing influences on cardiovascular disease and diabetes risk over time are uncertain. Although the number of people diagnosed with diabetes has increased worldwide (38), cardiovascular disease mortality in the United Kingdom has been declining since the 1970s, and research suggests that this is due to both declining incidence rates, as a result of changes in risk factors (39), and a reduction in case fatalities as a result of improved treatment (40). At a population level, a 1-cm increase in leg length has been associated with an estimated decreased risk in cardiovascular disease of approximately 4% (29) and in type II diabetes of 3.2% (30), while a 1-cm increase in waist circumference has been associated with an increased relative risk of a cardiovascular disease event of 2% (41). Thus, the large increases in adult BMI and levels of obesity over a relatively short interval between generations born in 1946 and 1958 raise questions about future trends for mortality and morbidity, particularly in cohorts who experienced the obesity epidemic in childhood. The 2 cohorts studied here have experienced the obesity epidemic at different stages of their lives in their 30s and 20s, respectively, with possible differences in effects on cardiovascular disease risks. Based on the estimated age trend for the younger cohort (7–45 years), the mean BMI will reach 30 kg/m² for men and 29 kg/m² for women by age 53 years, and thus the average male population will be obese. With the relatively low level of childhood obesity (0.4%–2.8%) in these 2 early cohorts compared with today's children (4.6% for boys and 6.8% for girls aged 5–10 years in 2002–2003) (42), if the obesity epidemic continues at its current pace, the level of obesity will increase substantially in the future adult population. These trends may have a negative impact on mortality and morbidity, although further studies are required to address the possibility that the impact of BMI trajectories on mortality and disease risk has changed over time.

Glossary

Abbreviation

BMI

body mass index