Background:

Adult obesity prevalence is influenced by rates of weight gain or loss among individual persons, but few studies have measured individual weight change in large populations. Changes in weight may not coincide with changes in the lipid accumulation product (LAP), a continuous index derived from waist circumference and triglycerides concentration for estimating excess lipids.

Design and measurements:

Descriptive report of longitudinal changes from US studies that included body mass index (BMI, kg/m²) and LAP.

Subjects:

A total of 16 763 white and black adults studied between 1989 and 1996 in three observational cohorts (Coronary Artery Risk Development in Young Adults, Atherosclerosis Risk in Communities Study and Cardiovascular Health Study).

Results:

The means of individual annual changes in BMI were positive for young adults, but the mean changes were reduced at older ages (P for linear trend <0.001), becoming negative after 73–83 years old. These mean, individual changes in BMI, specific to sex and age, were approximately 0.1 U/year greater than those reported from previous, representative, longitudinal measurements collected in the United States between 1971 and 1984. Mean, individual annual changes in LAP were strongly positive before middle age. For men, the annual LAP changes were reduced at older ages (P linear trend <0.05). For women, they were greater at older ages (white women, P<0.001) or remained unchanged (black women, P>0.3). With increasing age, there was a greater proportion of persons whose positive LAP change was accompanied by simultaneous BMI change that was negative or zero.

Conclusions:

These longitudinal observations made during 1989–1996 suggest greater annual changes in BMI compared to an adult cohort studied during 1971–1984. As estimated by LAP, adults of all ages tended to accumulate excess lipids, including circumstances in which they lost weight.

Introduction

Cross-sectional surveys of the US population have documented a continuing rise in the prevalence of overweight and obesity for nearly every adult age group.^{1, 2} These surveys leave no doubt that the population's average relative weight—commonly described as the body mass index (BMI, weight (in kg) divided by height (in m) squared)—is greater now than it was four decades ago. To supplement this recent prevalence information it would be useful also to know the patterns by which adults individually experienced weight gain or loss over time. However, the latest population-based study of longitudinal, measured, weight change among US adults was the follow-up to the first National Health and Nutrition Examination Survey (NHANES-I) that ended its data collection in 1984.³

In the absence of a more recent, nationwide, representative study, we here describe longitudinal, observational data assembled from three US cohorts that were developed to investigate cardiovascular risk: the study of Coronary Artery Risk Development in Young Adults (CARDIA), the Atherosclerosis Risk in Communities Study (ARIC) and the Cardiovascular Health Study (CHS). Our report presents the cohorts' experience of annual BMI changes among individual adults observed at two points in the period between 1989 and 1996. Although the three cohorts were recruited from just a few communities, the distributions of their baseline BMIs when stratified by age, sex and racial group indicate a close similarity to the comparable distributions of BMI derived from a cross-sectional sampling of US adults conducted during 1988–1994, NHANES-III.

From the perspective of predicting chronic diseases, we recognized that changes in BMI are not necessarily the same as changes in cardiometabolic risk. Nor are changes in BMI equivalent to changes in the amount or location of adipose tissue. None of these large cohorts obtained images or regional dual-energy X-ray absorptiometry from which to estimate specific changes in the adverse volume of truncal (more specifically, intraabdominal) fat⁴ or changes in the possibly protective volume of lower limb fat.⁵ The cardiovascular cohorts did, however, make repeated measures of waist circumference and fasting plasma triglycerides. From these two variables it was possible for each participant at each encounter to obtain the lipid accumulation product (LAP), an easily calculated, continuous index proposed for the estimation of accumulated excess lipids.^{6, 7} Dichotomous markers ('hypertriglyceridemic waist' or 'enlarged waist with elevated triglycerides') similar to the continuous LAP index are associated with prevalent coronary artery disease^{8, 9} and with established cardiometabolic risk factors.^{10, 11, 12}

Public health interest in obesity control requires a clear understanding of how contemporary experiences of individual weight change may differ from weight change experiences in earlier or different settings. We have therefore compared our longitudinal BMI data from 1989 to 1996 with longitudinal BMI data reported in NHANES-I approximately 15 years earlier. Clinical interest in the prevention of chronic disease requires that we consider the meaning of change in weight as possibly contrasted with change in excess lipids. Accordingly, from the three cardiovascular cohorts we have provided descriptive information that contrasts the individual annual changes in BMI with the simultaneous individual annual changes of LAP.

Methods

Our longitudinal cohort data came from public-use, limited-access, datasets furnished by the National Heart Lung and Blood Institute (Table 1). To position our measurements for all three cohorts in a similar range of calendar years, we took as our longitudinal baselines the third examination of CARDIA (1990–1991; examined at ages 23–35 years), the second examination of ARIC (1990–1992; ages 48–67 years) and the first examination of CHS (1989–1990; ages 65–89 years). All three of these cardiovascular cohorts were intentionally inclusive of the white and black communities; their similar eligibility criteria excluded only persons with physical or mental disabilities that would prevent full participation. We restricted our analytic samples to 16 763 white and black adults whose data included measured heights, weights, waist circumferences and fasting plasma triglycerides concentrations. We excluded users of hypolipidemic medication and women who were pregnant from our analyses. Displayed in Table 1 are the anthropometric protocols used by each study for arriving at BMI and waist circumference.

Table 1 - Participants selected from three community-based, longitudinal studies: CARDIA, ARIC and CHS.

Full table

The LAP index was developed earlier using population-based data to estimate adult minimum values for waist circumference, that is, the circumference of abdomens that would contain the least amount of adipose tissue (65 cm for men, 58 cm for women). As suggested by two-dimensional plots of increasing waist circumference and triglycerides concentration with aging (see http://www.biomedcentral.com/1471-2261/5/26), we accepted the following definitions:⁶

We calculated each participant's annualized changes of BMI and LAP, and characterized all the participants by their age in the middle of the observation interval ('mid age'), that is, their age at baseline plus half of their observation interval. (Baseline age was available in 1-year categories for CARDIA and ARIC but only in 2-year categories for CHS.) To model the effects of age on annualized change in BMI and in LAP we considered the independent variable to be each participant's mid age, and we constructed separate linear regression models (with and without a term for the square of mid age) for white and black men and women. We also divided our analytic sample into six age strata defined by the participants' age at the first examination included in the studies (Table 1). For each of the resulting 24 subcohorts (age sex race) we assigned a subcohort 'approximate age' that represented the subcohort's mean baseline age plus half its mean interval to the follow-up examination, and we computed the subcohort's mean, median and percentiles (5, 25, 75 and 95) of annualized change in BMI and in LAP.

For comparison with the earlier longitudinal US survey (data through 1984) that reported changes in BMI at approximate ages 35, 45, 55, 65 and 75 years old,³ we re-analyzed our data without regard to the six age-defined strata described above. This comparison analyzed only the 7459 participants whose mid age was 32–37 years (from CARDIA: 864 white and 555 black participants), 52–57 years (ARIC: 2164 white and 666 black participants), 62–67 years (ARIC: 1950 white and 407 black participants) or 72–77 years (CHS: 819 white and 34 black participants).

We assessed the representativeness of our data by comparing the baseline distribution of BMI for each of the 24 research subsamples with the estimated distribution of BMI of the corresponding national subpopulation (geometric mean and its 95% confidence interval) derived from an independent, cross-sectional, contemporaneous survey (NHANES-III) that also included measurements of waist circumference and the concentration of fasting triglycerides. For this purpose, the subsamples included eligible research participants at baseline who may have been omitted from the subcohort at the next included examination visit due to pregnancy, initiation of hypolipidemic medication or loss to follow-up. We used NHANES-III sampling weights along with SUDAAN software (Release 9.0, Research Triangle Institute, Research Triangle, NC, USA) to account for the unequal selection probabilities in NHANES-III (for example, clustered design, planned oversampling and differential nonresponse).¹³ It was not informative to compare the distributions of LAP between those in our longitudinal cohorts and the ones obtained in NHANES-III because NHANES-III employed nonequivalent protocols for its waist measurements (different anatomical landmarks)^{14, 15} and its assays for triglycerides used serum rather than plasma.¹⁶

Results

Longitudinal changes in BMI

For both sexes the annual changes in BMI were greatest in young adults, except for white women, where the highest change was in the group aged 54 years (Table 2 and Figure 1). Among men the median annual change in BMI at ages 29 years was about +0.2 U/year (+0.18 for white men, +0.25 for black men), and this median value was lower for each successively older male subcohort. The incidence of positive change in BMI (percent of persons with annual change greater than zero) ranged from 77% at ages 29 down to 46% at ages 79 for men in the combined racial groups. The estimated changes in BMI (curvilinear models regressed on mid age and the square of mid age) reached zero at age 77 years for white men and nearly zero (0.04 U per year) at age 73 for black men. The median annual changes in BMI for white men at ages 64, 71 and 79 years respectively were +0.13, +0.07 and -0.02 U per year (Figure 1). For black men at the same ages the median annual changes in BMI were +0.05, +0.06 and +0.07 U per year, but their two oldest subcohorts provided unreliable estimates due to their very small sample size.

Figure 1.

Distributions of annual change in body mass index (BMI, kg/m² per year) by each subcohort's sex, approximate age and racial group (white (W) or black (B)). The points displayed are at percentiles 5 (), 25, 50, 75 and 95 (X). The estimates for subcohorts of older blacks may be unreliable because of small numbers (^*n=32, ^**n=22, ^†n=58, ^††n=37).

Full figure and legend (110K)

Table 2 - Mean values for annual change in BMI (kg/m²/year) and annual change in LAP (cm mmol/l/year) by sex, racial group and subcohort.

Full table

Among white women, the median annual changes in BMI at ages 29, 35 and 54 years rose from +0.16 to +0.22 and then to +0.27 U of BMI per year (Figure 1). For black women, in contrast, the corresponding median changes in BMI began at +0.38 and then dropped to +0.27 and +0.23 U per year. Women of both racial groups showed a clear decline in annual change in BMI at ages above 54 years. The incidence of positive change for women (combined races) ranged 76% at ages 29 to 50% at 79. The estimated changes in BMI (curvilinear models regressed on mid age and square of mid age) reached zero at age 79 years for white women and 83 years for black women. The median annual changes in BMI at ages 64, 71 and 79 years for white women were +0.19, +0.07 and 0.00 U of BMI per year (Figure 1). For black women at the same ages the median changes in units of BMI per year were +0.12, +0.07 and +0.07, but their two oldest subcohorts may have provided unreliable estimates due to small sample size.

When compared to the annual changes in BMI reported³ among US adults in the longitudinal NHANES-I follow-up conducted from 1971–1975 to 1981–1984, our more recent estimates of annual change in BMI for white men and white women were consistently at least 0.1 U per year greater for mid ages 32–37, 52–57, 62–67 and 72–77 years (Figure 2). For the black men and women, our more recent annual changes in BMI likewise exceeded the NHANES-I follow-up report by an amount ranging from 0.07 to 0.26 U of BMI per year.

Figure 2.

Comparison of the US annual changes in body mass index (BMI) reported by a longitudinal cohort centered around 1978 (denoted by triangles, ; all racial groups included) ³ with the US annual changes in BMI observed for white population (small squares, ) and black population (large squares, ) in cohorts centered around 1993. In the more recent cohorts there were no participants with mid ages of 42–47 years, and there were too few black participants with mid ages 72–77 for reliable estimates.

Full figure and legend (86K)

Longitudinal changes in LAP

The annual changes in LAP were predominantly positive across the 24 subcohorts (Table 2 and Figure 3). For men (combined racial groups), the incidence of positive change in LAP ranged from 74% at ages 35 years down to 54% at ages 79 years. The interquartile range of values for change in LAP among the younger men (29 and 35 years old) was approximately 3 cm mmol/l/year, while this range for men at ages 54 and older was between 6 and 8 cm mmol/l/year.

Figure 3.

Distributions of annual change in lipid accumulation product (LAP, cm mmol/l/year) by each subcohort's sex, approximate age and racial group (white (W) or black (B)). Displayed points are at percentiles 5 (), 25, 50, 75 and 95 (X). The estimates for subcohorts of older blacks may be unreliable due to small numbers (^*n=32, ^**n=22, ^†n=58, ^††n=37).

Full figure and legend (116K)

For women, the incidence of a positive change in LAP was consistently about 70% for ages from 29 years to 54 years, and in the older subcohorts it dropped no lower than 66% (at ages 64 years). The interquartile range of values for change in LAP among the younger women (29 and 35 years old) was about 2 cm mmol/l/year, a narrower range than that observed for men at the same ages. Even so, the interquartile range for women at ages 54 and 64 years was between 7 and 8 cm mmol/l/year, and it then rose to values from 9 to 10 cm mmol/l/year at ages 71 and 79 years.

Population estimates of change in LAP are unavailable from NHANES-I because that survey did not obtain waist circumferences or fasting triglycerides.

Discordance of BMI and LAP changes

Although BMI and LAP values were well correlated in cross-sectional comparisons (Table 3, r range from 0.62 to 0.75), we noted weaker correlations between longitudinal rates of BMI change and LAP change (r range from 0.28 to 0.45). Viewed dichotomously, most participants were concordant with regard to their change in BMI and change in LAP (that is, they either gained in both indices or did not gain in both indices). With age, the concordance rate (both sexes combined) declined from 79 to 61% (Figure 4). As the sum of the two discordant fractions rose with age from 21 to 39%, the fraction of participants that gained LAP but not weight rose from 8 to 26%, while the fraction that gained weight but not LAP was relatively stable between 13 and 18%. Among participants who experienced a positive change in LAP, the subset of those who simultaneously experienced a zero or negative change in weight rose from 11% at ages 29 years to 42% at ages 79 years. We also found that among adults who did not gain weight at ages 71 and 79 years about half (40% of men and 58% of women) nevertheless experienced a gain in LAP.

Figure 4.

Percent of each sample (by approximate age, combined racial groups) experiencing four mutually exclusive combinations of longitudinal change: no gain of weight or lipid accumulation product (LAP, concordant), gain of weight but not LAP (discordant), gain of LAP but not weight (discordant) and gain of both weight and LAP (concordant).

Full figure and legend (112K)

Table 3 - Pearson's correlation of Ln BMI, Ln LAP, Ln triglycerides and Ln waist circumference (cross-sectional comparisons) and changes in each of these variables (longitudinal comparisons) by sex and cohort.

Full table

Comparability of the baseline subsamples with the estimated US population

Among the estimates in NHANES-III that corresponded to the 24 baseline subsamples in this report, the lowest estimate for geometric mean BMI was 23.1 [95% confidence interval 22.2–24.0] calculated for white women at ages 23–29 years. The highest NHANES-III estimate for BMI was 29.7 [27.9–31.8] for black women at ages 58–67 years. For 19 of the 24 cardiovascular baseline subsamples the observed geometric mean BMI was within the 95% confidence interval estimated from NHANES-III for the corresponding national population (P>0.05 for 19 t-tests).

Three of our white baseline groups had BMI geometric means that were lower than the corresponding national confidence intervals; they were CARDIA men at 30–35 years (BMI of 25.3 vs [25.5–27.4], t-test P=0.017), ARIC women at 48–57 years (26.4 vs [26.6–28.2], P=0.031) and CHS men at 73–89 years (25.5 vs [25.8–26.8], P=0.008). In contrast, two of our black baseline groups had BMI geometric means that were higher than the corresponding national confidence intervals; they were ARIC men at 48–57 years (27.9 vs [24.8–27.1], P=0.002), and ARIC women at 48–57 years (30.6 vs [27.9–30.3], P=0.046).

Discussion

Longitudinal surveys of measured weight change among adults in representative populations are uncommon. Scandinavian studies centered on the years 1972 and 1990 have demonstrated longitudinal patterns in which the rates of weight change were highest for the youngest participants (approximate ages 28–30 years), declined gradually toward zero for older ages and became negative at ages above 60 or 70 years old.^{17, 18} In the United States, the 10-year follow-up experience of NHANES-I was centered on the year 1978.³ An analysis of that study reported that the population mean annual changes in BMI (weighted estimates including all racial groups) at an approximate age of 35 years were +0.09 per year [95% confidence interval 0.07–0.11] for men and +0.13 [0.11–0.15] for women (Figure 2). The mean annual changes described in the NHANES-I follow-up study reached negative values after the approximate age of 55 for men and slightly older for women. They declined further to -0.11 [-0.09 to -0.13] for men and -0.17 [-0.15 to -0.19] for women at the approximate age of 75.

Our report based on cardiovascular research cohorts centered 15 years later than the previous US longitudinal study cannot provide estimates for the full US population because it has no sampling weights and includes only two race-ethnic groups. Nevertheless, we found that the sex- and age-specific mean annual changes in BMI for white or black subpopulations were greater than the previous rates (for the full population) by approximately 0.1 U of BMI per year (Figure 2). Furthermore, in our recent cohorts the estimated mean annual change in BMI did not decline to zero until at least age 73 (for black men) or 83 (for black women). That is, the recent positive rates of change in BMI persisted with aging at least 18 years longer through life than was previously noted.

Our report also describes longitudinal changes in LAP, an index that attempts to estimate the physiological burden of accumulated excess lipids. Although the ability of LAP to predict disease outcomes remains to be established, its component measures taken together (waist circumference and triglycerides) have been shown to have a close correlation with metabolic dysregulation.^{19, 20, 21} In one prospective study, an enlarged waist with elevated triglycerides predicted accelerated cardiovascular illness better than the more complex metabolic syndrome.²² The LAP index, although it requires a fasting venipuncture, could be a useful obesity estimator for assessment of persons (for example, amputees) whose weight or height may be difficult to interpret.

Unlike the pattern of declining rates of annual change in BMI at older ages, the annual change in LAP appeared to show no consistent decline with age. Thus, the eldest white men, although their average annual change in BMI approached zero or less, had a mean annual change in LAP that was positive and similar to the annual change in LAP for white men at ages 29 years (Table 2). The eldest white women, with an annual change in BMI of approximately zero, had a mean (and median) annual change in LAP that was substantially greater than the annual change in LAP for white women at ages 29 or 35 years (Table 2 and Figure 3). This pattern emphasizes that substantial portions of the elderly population are not gaining weight but nevertheless may continue to accumulate excess lipids. It is likely that these persons are losing salutary lean mass (primarily skeletal muscle and bone)^{23, 24} or subcutaneous adipose tissue in the lower extremities. Diminished lean mass or lower body subcutaneous adipose tissue is associated with an increased risk for cardiometabolic disease.^{25, 26, 27, 28}

This report is limited by the absence of data on racial-ethnic groups other than white and black population who were predominantly non-Hispanic. The prevalence of obesity may be especially high in some Hispanic, Asian and Pacific Islander, and Native American groups,²⁹ and without their information it is difficult to draw conclusions applicable to the entire US population. Our analyses specific to white and black groups, however, permit comparisons to earlier data from Scandinavia (based almost entirely on white population ) and to US surveys from earlier decades when Hispanics and Asians were smaller segments of the US population.

Our report is further limited by the lack of information on adults less than 23 years old or between 41 and 47 years old gaps that resulted from our strategic restriction to longitudinal cohort data collected only since 1989. It is likely that US adults less than 23 years old (with the possible exception of white women) gain weight at higher annual rates than do adults at any subsequent age.³⁰ Our exclusion from analysis of persons reporting the use of hypolipidemic medication is unlikely to have had any major impact on our findings since the prevalence of such usage by US adults in the years 1988–1994 was less than 3%.⁶

Our community-based observations regarding adults' longitudinal changes in BMI are compatible with the rising prevalence in the United States of overweight and obesity during the final decades of the twentieth century. In the absence of data from the earlier periods regarding accumulated excess lipid, we cannot determine the degree to which recent increases in overweight may reflect preserved healthy mass as opposed to increments in adverse lipid storage. Such distinctions between changes in weight and changes in accumulated excess lipids may be clinically important, especially among older adults for whom weight loss or even weight stability might signal the disappearance of healthy mass^{23, 24} while their burden of excess lipids continues to rise.

References

1. Kuczmarski RJ, Flegal KM, Campbell SM, Johnson CL. Increasing prevalence of overweight among US adults. The National Health and Nutrition Examination Surveys 1960–1991. JAMA 1994; 272: 205–211. | Article | PubMed | ISI | ChemPort |
2. Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, Flegal KM. Prevalence of overweight and obesity in the United States, 1999–2004. JAMA 2006; 295: 1549–1555. | Article | PubMed | ISI | ChemPort |
3. Williamson DF, Kahn HS, Remington PL, Anda RF. The 10-year incidence of overweight and major weight gain in US adults. Arch Intern Med 1990; 150: 665–672. | Article | PubMed | ISI | ChemPort |
4. Despres JP. Intra-abdominal obesity: an untreated risk factor for Type 2 diabetes and cardiovascular disease. J Endocrinol Invest 2006; 29: 77–82. | PubMed | ChemPort |
5. Okura T, Nakata Y, Yamabuki K, Tanaka K. Regional body composition changes exhibit opposing effects on coronary heart disease risk factors. Arterioscler Thromb Vasc Biol 2004; 24: 923–929. | Article | PubMed | ISI | ChemPort |
6. Kahn HS. The 'lipid accumulation product' performs better than the body mass index for recognizing cardiovascular risk: a population-based comparison. BMC Cardiovasc Disord 2005; 5: 26. | Article | PubMed |
7. Kahn HS. The lipid accumulation product is better than BMI for identifying diabetes: a population-based comparison. Diabetes Care 2006; 29: 151–153. | Article | PubMed | ISI |
8. St-Pierre J, Lemieux I, Vohl MC, Perron P, Tremblay G, Despres JP et al. Contribution of abdominal obesity and hypertriglyceridemia to impaired fasting glucose and coronary artery disease. Am J Cardiol 2002; 90: 15–18. | Article | PubMed | ISI | ChemPort |
9. St-Pierre J, Lemieux I, Perron P, Brisson D, Santure M, Vohl MC et al. Relation of the 'hypertriglyceridemic waist' phenotype to earlier manifestations of coronary artery disease in patients with glucose intolerance and type 2 diabetes mellitus. Am J Cardiol 2007; 99: 369–373. | Article | PubMed | ISI | ChemPort |
10. Lemieux I, Almeras N, Mauriege P, Blanchet C, Dewailly E, Bergeron J et al. Prevalence of 'hypertriglyceridemic waist' in men who participated in the Quebec Health Survey: association with atherogenic and diabetogenic metabolic risk factors. Can J Cardiol 2002; 18: 725–732. | PubMed | ISI |
11. Kahn HS, Valdez R. Metabolic risks identified by the combination of enlarged waist and elevated triacylglycerols. Am J Clin Nutr 2003; 78: 928–934. | PubMed | ISI | ChemPort |
12. Bloomgarden ZT. 2nd International Symposium on Triglycerides and HDL: metabolic syndrome. Diabetes Care 2005; 28: 2577–2584. | Article | PubMed |
13. Mohadjer L, Montaquila J, Waksberg J, Bell B, James P, Flores-Cervantes I et al. National Health and Nutrition Examination Survey III, Weighting and Estimation Methodology. National Center for Health Statistics: Hyattsville, MD, 1996. (Accessed 2-21-2007, at http://www.cdc.gov/nchs/data/nhanes/nhanes3/cdrom/NCHS/
    MANUALS/WGT_EXEC.PDF ).
14. National Center for Health Statistics. National Health and Nutrition Examination III: Body Measurements (Anthropometry). Westat Inc., Rockville, MD, USA, 1988. Accessed 2-21-2007, at http://www.cdc.gov/nchs/data/n
    hanes/nhanes3/cdrom/NCHS/MANUA
    LS/ANTHRO.PDF ).
15. Chumlea NC, Kuczmarski RJ. Using a bony landmark to measure waist circumference (letter). J Am Diet Assoc 1995; 95: 12. | Article | PubMed | ISI | ChemPort |
16. Laboratory Methods Committee LRCP. Cholesterol and triglyceride concentrations in serum/plasma pairs. Clin Chem 1977; 23: 60–63. | ISI |
17. Rissanen A, Heliovaara M, Aromaa A. Overweight and anthropometric changes in adulthood: a prospective study of 17 000 Finns. Int J Obes 1988; 12: 391–401. | PubMed | ISI | ChemPort |
18. Droyvold WB, Nilsen TI, Kruger O, Holmen TL, Krokstad S, Midthjell K et al. Change in height, weight and body mass index: longitudinal data from the HUNT Study in Norway. Int J Obes 2006; 30: 935–939. | Article | ISI | ChemPort |
19. Kriketos AD, Furler SM, Gan SK, Poynten AM, Chisholm DJ, Campbell LV. Multiple indexes of lipid availability are independently related to whole body insulin action in healthy humans. J Clin Endocrinol Metab 2003; 88: 793–798. | Article | PubMed | ISI | ChemPort |
20. Katsuki A, Sumida Y, Urakawa H, Gabazza EC, Murashima S, Maruyama N et al. Increased visceral fat and serum levels of triglyceride are associated with insulin resistance in Japanese metabolically obese, normal weight subjects with normal glucose tolerance. Diabetes Care 2003; 26: 2341–2344. | Article | PubMed | ISI | ChemPort |
21. van Wijk JP, Halkes CJ, Erkelens DW, Castro CM. Fasting and daylong triglycerides in obesity with and without type 2 diabetes. Metabolism 2003; 52: 1043–1049. | Article | PubMed | ISI | ChemPort |
22. Tanko LB, Bagger YZ, Qin G, Alexandersen P, Larsen PJ, Christiansen C. Enlarged waist combined with elevated triglycerides is a strong predictor of accelerated atherogenesis and related cardiovascular mortality in postmenopausal women. Circulation 2005; 111: 1883–1890. | Article | PubMed | ISI | ChemPort |
23. Gallagher D, Ruts E, Visser M, Heshka S, Baumgartner RN, Wang J et al. Weight stability masks sarcopenia in elderly men and women. Am J Physiol Endocrinol Metab 2000; 279: E366–E375. | PubMed | ISI | ChemPort |
24. Hughes VA, Roubenoff R, Wood M, Frontera WR, Evans WJ, Fiatarone Singh MA. Anthropometric assessment of 10-y changes in body composition in the elderly. Am J Clin Nutr 2004; 80: 475–482. | PubMed | ISI | ChemPort |
25. Tanko LB, Bagger YZ, Alexandersen P, Christiansen C, Larsen PJ. Peripheral adiposity and cardiovascular risk (letter). Circulation 2003; 108: e164. | Article | PubMed | ISI |
26. Sakai Y, Ito H, Egami Y, Ohoto N, Hijii C, Yanagawa M et al. Favourable association of leg fat with cardiovascular risk factors. J Intern Med 2005; 257: 194–200. | Article | PubMed | ISI | ChemPort |
27. Snijder MB, Visser M, Dekker JM, Goodpaster BH, Harris TB, Kritchevsky SB et al. Low subcutaneous thigh fat is a risk factor for unfavourable glucose and lipid levels, independently of high abdominal fat. The Health ABC Study. Diabetologia 2005; 48: 301–308. | Article | PubMed | ISI | ChemPort |
28. Van Pelt RE, Jankowski CM, Gozansky WS, Schwartz RS, Kohrt WM. Lower-body adiposity and metabolic protection in postmenopausal women. J Clin Endocrinol Metab 2005; 90: 4573–4578. | Article | PubMed | ISI | ChemPort |
29. Kumanyika SK. Obesity in minority populations: an epidemiologic assessment. Obes Res 1994; 2: 166–182. | PubMed | ChemPort |
30. Lewis CE, Jacobs Jr DR, McCreath H, Kiefe CI, Schreiner PJ, Smith DE et al. Weight gain continues in the 1990s: 10-year trends in weight and overweight from the CARDIA study. Am J Epidemiol 2000; 151: 1172–1181. | PubMed | ISI | ChemPort |

Acknowledgements

We thank the investigators of the CARDIA, ARIC and CHS cohorts along with the staff at the National Heart Lung and Blood Institute who released the limited-access, longitudinal datasets. We are also grateful for the extraordinary efforts of the field workers, laboratory personnel and statisticians who collected and processed the information in NHANES-III.