Paper

International Journal of Obesity (2004) 28, 120–128. doi:10.1038/sj.ijo.0802480 Published online 21 October 2003

Vigorous exercise and the population distribution of body weight

P T Williams1

1Life Sciences Division, Lawrence Berkeley Laboratory, Donner Laboratory, Berkeley, CA, USA

Correspondence: Dr PT Williams, Life Sciences Division, Lawrence Berkeley Laboratory, 1 Cyclotron Road, Donner Laboratory, Berkeley, CA, 94720, USA. E-mail: ptwilliams@ibl.gov

Received 31 October 2002; Revised 19 July 2003; Accepted 2 August 2003; Published online 21 October 2003.

Top

Abstract

BACKGROUND: While the benefits of vigorous exercise on body weight and regional adiposity are well established, whether these benefits affect equally the highest and lowest portions of the weight distribution have not been previously reported. The impact of exercise on the more extreme body weights and body circumferences is clinically important because these values represent individuals at greatest health risk.

Method: Self-reported weights and body circumferences from a cross-sectional sample of 7288 male and 2326 female runners were divided into five strata, according to the distances run per week and within each stratum the 5th, 10th, 25th, 50th, 75th, 90th and 95th percentiles were determined. Least-squares regression was then employed at each percentile to determine the dose–response relationship between running distance and adiposity as determined by body mass index (BMI) and self-reported circumferences of the waist, hip and chest.

RESULTS: Per kilometer run per week, the associated decline for BMI was three-fold greater at the 95th than at the 5th percentile in men, and six-fold greater at the 95th than the 5th percentile in women (all P<0.001). Reported waist circumference also declined more sharply at the 95th percentile than at the 5th percentile in men (-0.13plusminus0.02 vs -0.06plusminus0.01 cm per km/week) and women (-0.18plusminus0.04 vs -0.05plusminus0.01 cm per km/week). In women, both hip and chest circumferences declined more sharply per kilometer run at the 95th percentile than at the 5th percentile.

Conclusions: These results are consistent with the hypothesis that running promotes the greatest weight loss specifically in those individuals who have the most to gain from losing weight. Comparisons based on average BMI or average body circumferences are likely to underestimate the health benefits of running because of the J-shaped relationship between adiposity and mortality. Whether the observed cross-sectional associations are primarily due to exercise-induced weight loss or self-selection remains to be determined.

Keywords:

exercise, running, physical activity, body mass index, waist circumference, hip circumference, chest circumference

Top

Introduction

In contrast to their lower-mileage counterparts, runners who chart greater weekly run in distances have lower weights and narrower waists, hips and chests. 1,2,3,4,5 Individual responses in weight loss due to exercise, however, may vary considerably.3,4,5

Medical research relies on the average to efficiently summarize data from small to moderate size samples.6,7 The average is expected to be estimated with much greater precision than estimates of the 10th or 90th percentiles.8 Yet the more extreme values are generally most interesting because they represent individuals at greatest or least health risk.9

An average response to an intervention (eg, diet, exercise, drug) is often the appropriate statistic to use when individual responses are not dependent upon initial levels. However, when individual responses are based on initial levels, using the average may mask important information. For example, mean plasma high-density lipoprotein (HDL)-cholesterol concentrations have been reported to increase significantly with exercise training in sedentary men. However, the increase was three-fold greater in the men who started with a baseline HDL-cholesterol greater than or equal to 48 mg/dl compared to those whose starting HDL-cholesterol was 37 mg/dl or less.10

This paper examines the relationships of running distance to the percentile distribution of body mass index (BMI) and body circumferences (waist, hip, chest) in the National Runners' Health Study. Specifically, it tests whether cross-sectional relationships of adiposity with running distance were the same at the 5th, 10th, 25th, 50th, 75th, 90th and 95th percentiles of fatness. The study's large sample size provides the statistical power required for estimating percentiles precisely. The results suggest a greater decline in adiposity per mile run at higher percentiles. If these cross-sectional relationships are reflective of the changes in adiposity associated with exercise training, then the expected benefits from vigorous exercise will depend strongly on initial levels, although causality remains to be determined from longitudinal data.

Top

Methods

The design and subject characteristics of the National Runners' Health Study are described in detail elsewhere.1,2 Briefly, all participants received a two-page questionnaire, distributed nationally at races and to subscribers of the nation's largest running magazine (Runners' World, Emmaus, PA, USA). This questionnaire solicited information on: demographics (age, race, education); running history (age when the participant began running at least 12 miles per week, average weekly distance run, best marathon and 10-km race times during the past 5 y); weight history (greatest and current weight, weight when the participant started running, least weight as a runner, circumferences of the chest, waist and hips); diet (vegetarianism and the current weekly consumption of alcohol, red meat, fish, fruit; vitamin C, vitamin E and aspirin); current and past cigarette use; prior history of heart attacks and cancer; and medications taken for high blood pressure, hypothyroid, hypercholesterolemia or diabetes. The average number of kilometers run per week was calculated by averaging the reported yearly distances of the preceding 5 y. BMI was calculated as the weight in kilograms divided by height in meters squared. We estimate that 15–19% of the subjects returned our questionnaire (uncertainty due to an unknown number of race participants who received our questionnaire and undeliverable addresses).

The current report focuses on the self-reported body weights, heights and body circumferences from 7288 male and 2326 female nonvegetarian, nonsmoking runners without prior history of heart disease or cancer. Self-reported body circumferences of the waist, hip and chest were in response to the question 'Please provide, to the best of your ability, your body circumference in inches' without further instruction. Waist circumferences were reported by 6763 men and 2164 women, hip circumferences by 3673 men and 2109 women, and chest circumferences by 5946 men and 2154 women. Self-reported height and weight from the questionnaire have been found previously to correlate strongly with their clinic measurements (unpublished correlation in 110 men were r=0.96 for both). Self-reported body circumferences are somewhat less precise as indicated by their correlations with self-reported circumferences on a second questionnaire (waist: r=0.84; hip: r=0.79; chest: r=0.93) and with their clinic measurements (waist: r=0.68; hip: r=0.63; chest: r=0.77). Chest circumference has not been frequently used as a measure of adiposity, however, others have reported that chest circumference provides a measure of upper body obesity that exhibits relationships to plasma leptin levels that were not apparent for waist or hip measurements,11 and that endurance-oriented physical activity significantly decreases chest diameter.12 Thoracic fat has also been related to low-density lipoprotein levels.13

Statistical analyses

Least-squares regression was used to estimate the apparent change in adiposity by reported running distance at the 5th, 10th, 25th, 50th, 75th, 90th and 95th percentiles. Specifically, the data were divided into five categories by running distance (less than or equal to16, 16–32, 32–48, 48–64 and 64 or more km/week), and within each distance category the BMIs corresponding to these percentiles were determined. The average weekly running distances were also calculated within each of the five distance categories. We then applied simple linear regression to the five bivariate observations consisting of the average distance run (independent variable) and the BMI or circumference at the ith percentile (eg, i=5th, 10th, 25th, 50th, 75th, 90th and 95th), for each distance category (dependent variable).

Since the usual underlying statistical assumptions presumably do not apply for percentiles (particularly those representing the tails of the distribution), the standard errors and significance levels were calculated using bootstrap resampling.14 Bootstrap estimates were created as follows: (1) within each of the five distance categories, sampling with replacement was used to create a bootstrap data set of distance run and BMI; (2) average distance run and BMI corresponding to the 5, 10, 25, 50, 75, 90 and 95 percentiles for the bootstrap sample were determined; (3) least-squares regression was applied to estimate at each percentile the apparent change in BMI per kilometers run across the five distance categories; (4) steps 1–3 were repeated 10,000 times. This yielded 10,000 regression slopes (one for each bootstrap sample). The average and the standard deviation of the 10,000 regression slopes provide the bootstrap estimate of the regression slope and its standard error at the ith percentile. The bootstrap samples also provide an estimate of statistical significance. Two-tailed significance levels were calculated as 2*minimum (p, 1-p), in which p is the proportion of times that the 10,000 bootstrap slopes were less than zero.

If running causes the same change in BMI regardless of whether the individual is lean or overweight, then the regression slopes for the 5th, 10th, 25th, 50th, 75th, 90th and 95th percentiles will be parallel. Nonparallel regression slopes could indicate that running affects various portions of the adiposity population distribution differently. Bootstrap resampling was used to estimate the difference between two regression slopes (eg, the 75% slope minus the 25% slope) and the corresponding standard error. Bootstrap resampling was also used to test whether the slopes increased or decreased progressively from 5 to 95% of the adiposity distribution. (Nearly identical results were obtained for a contrast that increased linearly with the percentile or with the ordinal sequence of the seven percentiles). Bootstrap estimates and standard errors for differences in regression slopes, and linear contrasts across regression slopes were again based on 10 000 bootstrap samples. Two-tailed significance levels were calculated as 2*minimum (p, 1-p), where p is the proportion of times that the difference in slopes at the ith and jth percentile, or the linear contrast across slopes (ie, the slope of the slopes) was less than zero.

Top

Results

Table 1 presents the characteristic of the male and female runners by weekly distance run. Age and height were largely unrelated to running distance. Runners who ran further each week weighed less, had lower BMI, narrower circumferences of the waist, hip and chest, weighed less when they began running, and had run more years.


Body mass index

Figure 1 displays the 5, 10, 25, 50, 75, 90 and 95th percentiles for men's BMI stratified by weekly distance run. The figure presents the significance of the difference in BMI across the five distance categories (ie, by 16 km increments in running distance). The figure shows that the median BMI (ie, 50th percentile) became significantly smaller for each 16 km increment in weekly running distance (eg, median BMI was significantly smaller in men who ran 16–31.9 km/week than men who ran under 16 km/week, significantly smaller in men who ran 32–41.9 km/week than men who ran either 16–31.9 km/week or under 16 km/week,..., and significantly smaller in men who ran over 64 km/week than in any distance category under 64 km/week). In general, there were significant declines in BMI with every 16 km/week increment in running distance for both the leanest (5% of the sample) and heaviest runners (95% of the sample), and for all percentiles in between (only two exceptions being the differences between 16–31.9 km/week and under 16 km per week at the 10 and 95th percentiles). Figure 2 displays a similar pattern for women; however, there are slightly fewer significant comparisons, which is most likely the consequence of the smaller sample size.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Distribution of BMI by reported distance run per week in male runners (N=7288). Regression slopes and their standard error (listed on the right) were determined by least squares and bootstrap resampling, respectively. All regression slopes were significant at P<0.001. Significant differences (p<0.05) from the 0 to 16 km (*), 0 to 16 and 16 to 32 km (†), 0 to 16, 16 to 32 and 32 to 48 km (‡), 0 to 16, 16 to 32, 32 to 48 and 48 to 64 km (§) are designated for each percentile.

Full figure and legend (43K)

Figure 2.
Figure 2 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Distribution of BMI by reported distance run per week in female runners (N=2326). Regression slopes and their standard error (listed on the right) were determined by least squares and bootstrap resampling, respectively. All regression slopes were significant at P<0.001. Significant differences (p<0.05) from the 0 to 16 km (*), 0 to 16 and 16 to 32 km (†), 0 to 16, 16 to 32 and 32 to 48 km (‡), 0 to 16, 16 to 32, 32 to 48 and 48 to 64 km (§) are designated for each percentile.

Full figure and legend (44K)

The calculated rate (plusminuss.e.) of decline per reported kilometer run for the median and the other percentiles is estimated by the regression coefficients presented on the right of Figures 1 and 2 and in Table 2. For example, the 50th percentile of the men's BMI distribution declined -0.029 kg/m2 per km run, with a standard error of 0.002. The men's and women's BMI also declined significantly with running distance at all other percentiles. Table 2 shows that for both sexes, the declines at the 5, 10, 25, 50, 75, 90 and 95% were each significant at P <0.001.


Figure 1 also tests whether the regression slopes are the same at all percentiles. In men, the slope at the 95th percentile of BMI was twice as great as the slope at the median (50th percentile) and over three-fold greater than the slope at the 5th percentile of BMI. In women, the slope at the 95th percentile was three times greater than the slope at the median, and six-fold greater than the slope at the 5th percentile of BMI. In both sexes, the slopes became progressively weaker for the 95th, 90th, 75th and 50th percentiles of BMI, and the overall trend for progressively steeper slopes with increasing percentiles was significant at P<0.001.

Tables 3 and 4 provide more detailed comparisons between the regression slopes at the 5, 10, 25, 50, 75, 90 and 95% (ie, all 21 pairwise comparisons between the seven regression slopes). The tables show that nearly all of the pairwise comparisons were significantly different from each other (one exception in men: the 5th vs the 10th percentile; three exceptions in women: the 5th vs the 10th and 25th percentiles, and the 25th vs the 50th percentile). These analyses suggest that the decline in body weight associated with running distance could depend upon whether an individual's weight is low or high relative to others in the population. Specifically, Figures 1 and 2 show the steepest regression slopes are at the highest percentiles, and that the slopes become progressively more gradual (closer to zero) with decreasing percentiles through the 5th.



Body circumferences

As expected, self-reported circumferences of the waist and hips exhibited associations with weekly distance run that were consistent with those observed for BMI. Specifically, Tables 2, 3 and 4 show: (1) that waist and hip circumferences declined significantly in association with weekly distance run for all percentiles between the 5th and 95th; (2) that the declines were substantially greater above the median than below the median. Compared to the 5th percentile, the decline per kilometer run at the 95th percentile was twice as great in men and two to three times as great in women for waist and hip circumferences. In men, chest circumference declined in association with weekly distance run at all percentiles, although there was no conclusive evidence (ie, P=0.08) that the decline was different across percentiles. In women, weekly running distance was not associated with differences in chest circumference below the median. Chest circumferences above the median did decline in association with running distance. The rate of decline was significantly greater in women having the greatest chest circumferences (95th percentile) than in women with smaller chests.

Standard regression slopes by starting running weight

We also examined the standard least-squares regression slope of adiposity vs running distance when the samples were divided by quartiles of starting BMI (i.e., when starting to run at least 12 miles per week). The quartiles of starting BMI were <22.62 (ie, those who were leanest when they started running), 22.62–24.41, 24.42–26.47 and greater than or equal to26.48 kg/m2 (ie, those who were heaviest when they started to run) in men and <19.64, 19.64–20.93, 20.94–22.46, and greater than or equal to22.47 kg/m2 in women. In support of the results of Figures 1 and 2, the regression slopes were significantly more negative (steeper decline) in runners who were initially heaviest. In men, the corresponding slopes (from leanest to heaviest quartile) per km run/week (plusminuss.e.) were -0.018plusminus0.002, -0.025plusminus0.002, -0.029plusminus0.002 and -0.049plusminus0.003 kg/m2 for BMI; -0.054plusminus0.006, -0.066plusminus0.006, -0.081plusminus0.006 and -0.113plusminus0.008 cm for waist circumference; -0.055plusminus0.009, -0.063plusminus0.011, -0.082plusminus0.012 and -0.096plusminus0.012 cm for hip circumference; and -0.040plusminus0.008, -0.047plusminus0.008, -0.052plusminus0.008 and -0.074plusminus0.009 cm for chest circumference. In women, the corresponding slopes were -0.006plusminus0.003, -0.022plusminus0.003, -0.023plusminus0.003 and -0.050plusminus0.007 kg/m2 for BMI; -0.042plusminus0.011, -0.066plusminus0.015, -0.059plusminus0.014 and -0.134plusminus0.021 cm for waist circumference; -0.053plusminus0.011, -0.073plusminus0.014, -0.054plusminus0.012 and -0.121plusminus0.019 cm for hip circumference; and -0.016plusminus0.008, -0.028plusminus0.012, -0.035plusminus0.011 and -0.086plusminus0.017 cm for chest circumference.

Compressed weight distributions among longer-distance runners

The sharp drop in BMI and body circumferences per kilometer run above the median and the more gradual decline in BMI and circumferences per km below the median leads to a more compressed body-weight distribution among long-distance runners vis-à-vis short-distance runners. The effect is apparent for the standard deviation, which is a commonly employed measure of dispersion. The standard deviation for BMI in men was 3.09 for those running under 16 km/week; 2.53 for 16–31.9 km/week, 2.23 for 32–47.9 km/week; 2.18 for 48–63.9 km/week; and 1.94 for those running over 64 km/week. Table 1 shows that the compression is most pronounced for BMI and waist in both sexes, and in chest circumference in women.

Top

Discussion

The statistical average and standard deviation are the most frequently used parameters for describing the relationship of adiposity to other variables in a population. For example, that average BMI and average waist, hip and chest circumferences have been previously reported to all decrease incrementally with reported distance run per week in men and women.1,2 The present report shows that the relationship of adiposity to the quantity of vigorous exercise is more complex than previously described. Specifically, the declines in weight and body circumferences per kilometer run are greatest at the highest percentiles of body weight and become gradually weaker for the leaner percentiles. Thus, the declines in mean BMI and body circumferences per kilometer run may be disproportionately due to large reductions in the most-overweight individuals. This conclusion is supported by another analysis that compared the regression slopes between adiposity and running distance stratified by BMI when starting to run. The steepness of the regression slope increased progressively with starting BMI.

The steeper (ie, more negative) regression slopes for BMI and body circumference per km run at the 95th percentile compared to the 50th percentile could be due to exercise-induced weight loss, effects due to self-selection, or both. There may be a greater susceptibility to exercise-induced weight loss at higher BMIs and progressively greater resistance to losing weight as adiposity approaches the median. Alternatively, if the leaner body weights of longer-distance runners are due to self-selection rather than weight loss, then the selective pressure against heavier body types may primarily involve the most extreme values and become progressively weaker as adiposity approaches the median. However, at least some of the leanness of higher mileage runners is likely to be the direct consequence of running since: (1) clinical trials show sedentary men assigned to a program of running experience significance weight loss relative to controls4,5 and (2) differences between runners current weights and greatest lifetime weights are correlated significantly with average weekly distance run (the leanest runners are also the ones who have lost more weight since starting to run).1

Running may promote the greatest weight loss specifically among those individuals who have the most to gain by losing weight. The relationship of BMI to morbidity and mortality is nonlinear.15,16 Several studies report a U-shaped relationship between mortality and adiposity,15,16 while others report accelerated increase in mortality as BMI is increased.17 A 14-y follow-up of over one million men and women revealed that among lifetime nonsmokers, the nadir for risk was between 23.5 and 24.9 kg/m2 in men and between 22.0 and 23.4 kg/m2 in women.18 In men, the risk for all cause mortality increased 30% as BMI decreased to 18.5 kg/m2 or increased to 30 kg/m2 from the nadir, accelerating thereafter to 170% higher mortality for BMI greater than or equal to40 kg/m2. In women, the risk for all cause mortality increased 35% as BMI decreased less than or equal to18.5 kg/m2 or increased to 30 kg/m2 from the nadir, and then accelerating to 90% higher mortality as BMI increased from 30 and 40 kg/m2. Among nonsmokers, the U-shaped distribution seemed to arise from the combination of excess mortality due to causes other than cardiovascular disease and cancer among very lean men and women, and a J-shaped increase in cardiac mortality from the lowest to the highest BMI quartiles. These relationships are substantially modified by prior history of disease and smoking.

The apparent differences in the effect of running distance on adiposity in overweight and normal-weight men and women were also observed for circumferences of the waist and hip (both sexes) and chest (women only). In men, the waistline reduction per kilometer run was 60% greater at the 95th percentile than at the 50th percentile. Several studies suggested that anthropometrical estimates of abdominal adiposity (presumably intra-abdominal fat) increase the risks for morbidity and mortality, independent of total body fat.19,20,21 A J-shaped distribution (ie, accelerated mortality) has also been reported for intra-abdominal fat.22

Prior studies have not generally focused on testing whether the relationship of adiposity to physical activity differs by percentiles of the population. This may in part reflect the particular requirements for these analyses, namely: (1) a large-enough sample to estimate the 5th, 10th, 25th, 50th, 75th, 90th and 95th percentiles with sufficient precision for hypothesis testing (ie, a larger sample size is required to estimate the 5th or 95th percentiles with the same accuracy as the mean) and (2) the ability to create bootstrap estimates to compare trends across percentiles (because of the nonnormality of the residuals and the lack of independence for the trends measured at different percentiles within the same sample). The analyses presented in this report show that the sample mean (or least-squares estimates) is insufficient for describing the associations of adiposity with vigorous exercise. The differences in adiposity between men and women who run only a few kilometers per week and those who run substantially greater distances are expected to translate into even greater reductions in mortality than suggested by averages because of the J-shaped relationship between adiposity and mortality.

There are important limitations to the analyses presented. The questionnaire did not include any questions about intentional vs unintentional weight loss, and the change in intensity or volume during the history of running and body mass, and so it is not possible to assess how these variables may have affected the analyses. The steeper regression slopes for BMI at the higher percentiles of adiposity may reflect the increase in energy expenditure per kilometer run as body weight is increased. There were no direct measurements of fat mass and fat-free mass for these runners so it cannot be determined how body composition might have been affected by running distance. The inverse association between running distance and adiposity is likely to at least partially reflect self-selection, although weight loss is a well-established consequence of exercise, and it has been previously reported that the runners in this cohort who ran the most miles were the furthest below their greatest lifetime weight.1,3,4,5 Self-reported body circumferences of the waist, hip and chest were in response to the question 'Please provide, to the best of your ability, your body circumference in inches' without further instruction. The relationships between circumference and running distance are expected to be weakened by different perception of where waist, hip and chest circumferences lie. However unless the perceived location varies symmetrically in relation to running distance, this subjectivity is unlikely to produce the relationships in the tables and figures. Moreover, the associations with waist, hip, and chest are entirely consistent with those observed for BMI, which is less subjective.

Proof of causality requires a randomized, controlled trial, which this study is not. Nevertheless, the large number of subjects studied at exercise levels generally unattainable in intervention studies provides a unique opportunity for assessing the relationship between vigorous physical activity and the population distribution of adiposity. These relationships reported here are consistent with the hypothesis that running affects adiposity differently, depending upon whether the person is lean or fat relative to others in the population.

Top

References

  1. Williams PT. Relationship of distance run per week to coronary heart disease risk factors in 8,283 male runners. The National Runners' Health Study. Arch Intern Med 1997; 157: 191–198. | Article | PubMed | ISI | ChemPort |
  2. Williams PT. High density lipoprotein cholesterol and other risk factors for coronary heart disease in female runners. N Engl J Med 1996; 334: 1298–1303. | Article | PubMed |
  3. Williams PT, Wood PD, Haskell WL, Vranizan K. The effects of running mileage and duration on plasma lipoprotein levels. JAMA 1982; 247: 2674–2679. | Article | PubMed | ISI | ChemPort |
  4. Wood PD, Stefanick ML, Dreon DM, Frey-Hew HB, Garay SC, Williams PT, Superko HR, Fortmann SP, Acbers JJ, Vranizan KM. Changes in plasma lipids and lipoproteins in overweight men during weight loss through dieting as compared with exercise. N Eng J Med 1988; 319: 1173–1179.
  5. Wood PD, Stefanick ML, Williams PT, Haskell WL. The effects on plasma lipoproteins of a prudent weight-reducing diet, with or without exercise, in overweight men and women. N Engl J Med 1991; 325: 461–466. | PubMed | ISI | ChemPort |
  6. Rothwell PM. Can overall results of clinical trials be applied to all patients? Lancet 1995; 345: 1616–1619. | Article | PubMed |
  7. Bailey KR. Generalizing the results of randomized clinical trials. Control Clin Trials 1994; 15: 15–23. | Article | PubMed |
  8. Yusuf S, Wittes J, Probstfield J, Tyroler HA. Analysis and interpretation of treatment effects in subgroups of patients in randomized clinical trials. JAMA 1991; 266: 93–98. | Article | PubMed | ChemPort |
  9. Ioannidis JP, Lau J. The impact of high-risk patients on the results of clinical trials. J Clin Epidemiol 1997; 50: 1089–1098. | Article | PubMed | ChemPort |
  10. Williams PT, Stefanick ML, Vranizan KM, Wood PD. The effects of weight loss by exercise or by dieting on plasma high-density lipoprotein (HDL) levels in men with low, intermediate, and normal-to-high HDL at baseline. Metabolism 1994; 43: 917–924. | Article | PubMed | ChemPort |
  11. Sudi KM, Gallistl S, Tafeit E, Moller R, Borkenstein MH. The relationship between different subcutaneous adipose tissue layers, fat mass and leptin in obese children and adolescents. J Pediatr Endocrinol Metab 2000; 13: 505–512. | PubMed |
  12. Houmard JA, McCulley C, Roy LK, Bruner RK, McCammon MR, Israel RG. Effects of exercise training on absolute and relative measurements of regional adiposity. Int J Obes Relat Metab Disord 1994; 18: 243–248. | PubMed |
  13. Pouliot MC, Despres JP, Moorjani S, Tremblay A, Lupien PJ, Nadeau A, Theriault G, Bouchard C. Computed tomography-measured trunk fat and plasma lipoprotein levels in nonobese women. Metabolism 1989; 38: 1244–1250. | PubMed |
  14. Efron B. The Jackknife, the bootstrap and other resampling plans. Society for industrial and applied mathematics. Society for Industrial and Applied Mathematics: Philadelphia, PA; 1982. pp 1–92.
  15. Tsugane S, Sasaki S, Tsubono Y. Under- and overweight impact on mortality among middle-aged Japanese men and women: a 10-y follow-up of JPHC study cohort I. Int J Obes Relat Metab Disord 2002; 26: 529–537. | Article | PubMed |
  16. Troiano RP, Frongillo Jr EA, Sobal J, Levitsky DA. The relationship between body weight and mortality: a quantitative analysis of combined information from existing studies. Int J Obes Relat Metab Disord 1996; 20: 63–75. | PubMed | ChemPort |
  17. Jonsson S, Hedblad B, Engstrom G, Nilsson P, Berglund G, Janzon L. Influence of obesity on cardiovascular risk. Twenty-three-year follow-up of 22025 men from an urban Swedish population. Int J Obes Relat Metab Disord 2002; 26: 1046–1053. | Article | PubMed | ChemPort |
  18. Calle EE, Thun MJ, Petrelli OM, Rodriguez C, Heath CW. Body-mass index and mortality in a prospective cohort of US adults. N Engl J Med 1999; 341: 1097–1105. | Article | PubMed | ISI | ChemPort |
  19. Després JP, Moorjani S, Lupien PJ, Tremblay A, Nadeau A, Bouchard C. Regional distribution of body fat, plasma lipoproteins, and cardiovascular disease. Arteriosclerosis 1990; 10: 497–511. | PubMed | ISI | ChemPort |
  20. Kissebah AH, Krakower GR. Regional adiposity and morbidity. Physiol Rev 1994; 74: 761–811. | PubMed | ISI | ChemPort |
  21. Lakka HM, Lakka TA, Tuomilehto J, Salonen JT. Abdominal obesity is associated with increased risk of acute coronary events in men. Eur Heart J 2002; 23: 706–713. | Article | PubMed | ISI |
  22. Oppert JM, Charles MA, Thibult N, Guy-Grand B, Eschwege E, Ducimetiere P. Anthropometric estimates of muscle and fat mass in relation to cardiac and cancer mortality in men: the Paris Prospective Study. Am J Clin Nutr 2002; 75: 1107–1113. | PubMed | ISI | ChemPort |
Top

Acknowledgements

Supported by Grant HL-058621 from the National Heart Lung and Blood Institute, and was conducted at the Lawrence Berkeley Laboratory (Department of Energy DE-AC03-76SF00098 to the University of California).

Top

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated

Extra navigation

.

natureevents

ADVERTISEMENT