The Effect of Cardiorespiratory Fitness and Obesity on... : Medicine & Science in Sports & Exercise (original) (raw)
Obesity is rising at an alarming rate in the United States (24), whereas the levels of physical activity remain stable and low (12). The pattern of these health behaviors could manifest later as disease, such as cancer. From 1992 to 1998, cancer incidence rates declined in men but not women, whereas cancer death rates declined in both women and men (18). National cancer rates continue to be higher among blacks as compared with whites, with a 17% excess rate among women and a 40% excess rate among men (13). According to Friedenreich (14), some of the main biologic mechanisms by which physical activity reduces cancer risk include modifications of endogenous sex and metabolic hormone levels, and growth factors, decreased body fat, and possibly enhanced immune function.
Only four published epidemiologic studies from two cohorts have examined the effect of fitness on all-cause cancer mortality (4,7,19,20). The association between obesity and cancer is reported from the first and second Cancer Prevention Studies (11,16,17,21). We are not aware of any epidemiologic study that has examined the interaction between cardiorespiratory fitness and obesity on all-cause cancer mortality. In a comprehensive review of physical inactivity and mortality, a need for more evidence from prospective observational cohorts on fitness or physical activity, obesity, and mortality was recommended (6). There is especially a need for studies that include women and older persons. Therefore, the purpose of this study was to determine the effect of fitness, obesity, and the interaction between fitness and obesity on all-cause cancer mortality for women and men. These associations were also explored for physical activity and its interaction with obesity on cancer mortality.
METHODS
This research used data from the Lipids Research Clinics (LRC) First Prevalence Studies and the Mortality Follow-up Studies. Participants were from eight geographically diverse centers in the United States: Cincinnati, OH; Houston, TX; Iowa City, IA; La Jolla, CA; Minneapolis, MN; Oklahoma City, OK; Palo Alto, CA; and Seattle, WA. Recruitment used a standardized protocol in target populations defined by occupational groups, households in geographic zones, households with membership in a particular medical plan, or students and parents within school districts (22). Participants were sampled from defined groups but were not necessarily representative of the local populations. The procedures followed for this study were in accordance with the ethical standards of the Institutional Review Board and participants provided written informed consent.
The methodology and study design of the LRC Prevalence and Mortality Follow-Up Study is reported elsewhere (28,31). A two-stage procedure was used with selected participants from a brief first visit, participating in a more extensive second visit. Participants in the second visit consisted of a 15% random sample of all visit 1 participants and 100% of those with elevated plasma lipids. The response rate for both strata of the sample was 85%. It was during the second examination that fitness measures were obtained, and this examination provided the baseline measures for this study. The two examinations took place between 1972 and 1976, and the median time between an individual’s two visits was 96 d.
Vital status follow-up.
Deaths were obtained by annual follow-up contacts, mostly by phone, with the cohort up to the end of 1987. At the end of 1987, vital status was known on 99.6% of the cohort. After 1987, annual follow-up contacts were discontinued; follow-up was conducted by searching the National Death Index (1988–91) and the Epidemiology Research Index (1992–98). For this study, vital status information was complete through 1998. Using the Kaplan-Meier estimator, the median follow-up time was 24.9 yr. Cause of death was ascertained by nosologist’s coding of the death certificates for the entire follow-up period. International Classification of Disease version 9 (ICD-9) codes 140–171 and 174–239 identified all-cause cancer deaths. We excluded skin cancer deaths from our definition (ICD-9 172–173).
Measurement of fitness and physical activity.
Cardiorespiratory fitness, now called “fitness” for the remainder of this report, was assessed as the time to produce predicted maximal heart rate based on age and training during a standardized treadmill test (25). Participants were told to refrain from eating for 2 h before testing, and most tests were performed in the morning. The test was conducted according to the Bruce protocol (25). Seven, 3-min stages were used in which the speed and inclination were increased in a stepwise fashion as follows: stage 1, 1.7 miles per hour (mph) and 10% inclination; stage 2, 2.5 mph and 12% inclination; stage 3, 3.4 mph and 14% inclination; stage 4, 4.2 mph and 16% inclination; stage 5, 5.0 mph and 18% inclination; stage 6, 5.5 mph and 20% inclination; and stage 7, 6.0 mph and 22% inclination.
The ECG was monitored continuously and blood pressure was measured at the end of each stage. Heart rate was monitored continuously and was also recorded at the end of each stage, or earlier if the participant stopped during a stage. The test was stopped when participants reached 90% of their predicted maximal heart rate, based on age and physical training (9,25). In this study, physical training was determined using two questions on physical activity: (1) “Do you regularly engage in strenuous exercise or hard physical labor?” Yes or no (2). If Yes: “Do you exercise or labor at least three times a week?” Yes or no. Physical activity was categorized as: (1) very active: individuals reporting strenuous exercise or hard physical labor ≥ 3 times per week; (2) moderately active: individuals reporting strenuous exercise or hard physical labor <3 times per week; or (3) inactive: individuals reporting no strenuous exercise or hard physical labor. Physical training was determined in LRC by whether the participant was classified as very active (trained) versus moderately active or inactive (not trained).
The exercise test was terminated early if the participant was unable to continue because of chest pain, fatigue, dyspnea, or leg pain or because of abnormalities in the ECG (≥1 mm horizontal ST-segment change, major arrhythmias, or conduction defects), a decrease in systolic blood pressure, technical difficulties, or if subjects were unwilling to continue. Otherwise, the test was stopped when the participant attained 90% of predicted maximal heart rate and either maintained it for 1 min, maintained it to the end of the stage, or exceeded the target heart rate by 8 bpm, whichever occurred first (25). For these analyses, fitness was quantified as the duration of the exercise test in minutes.
Other measurements.
At the second visit, a detailed examination was conducted which included an interview, physical exam, graded exercise test, and collection of plasma samples. Height and weight were measured with the participant wearing light clothing and no shoes. Height was measured to the nearest 0.5 cm using a headboard and a vertical rule fixed to a wall. Weight was measured to the nearest 0.1 kg using a balance scale. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg·m−2).
Education was categorized as less than high school graduate, high school graduate, or more than high school. Cigarette smoking was categorized as current >20 cigarettes·d−1, current ≤20 cigarettes·d−1, former, or never. Participants were questioned on the type and amount of different types of alcoholic beverages consumed in the past 7 d and average grams of alcohol intake per day were calculated. For statistical modeling, alcohol intake was categorized as >20, >10–20, 1–10, or 0 g·d−1. Dietary intake was assessed with a 24-h recall, and Keys score was calculated as described by Anderson et al. (3) and was treated continuously in statistical models. For women, we adjusted for self-reported menopause (yes, no).
Statistical analyses.
From the eight LRC Prevalence Study sites mentioned earlier, participants less than 30 yr of age were excluded because mortality follow-up was not done on those participants. Starting with N = 3880 women and N = 4803 men, we excluded two participants over the age of 75 yr. Because the association between BMI and mortality is likely different in blacks compared with whites (26), we chose not to combine data from ethnic groups. The number of minority individuals examined was too small to study them here separately (N = 532). To reduce confounding from preexisting illness (23), participants who died in the first 4 yr of follow-up (N = 165) and participants with a BMI less than 18.5 kg·m−2 (N = 136) were excluded.
Because heart rate response to exercise was used as an indicator of fitness, we excluded participants taking medication that may alter heart rate (N = 172) and participants with inconsistent heart rates (N = 22). We excluded 222 participants with a positive exercise test (e.g., indicating possible cardiovascular disease) and 342 participants due to contraindications for participating in the exercise test (e.g., aortic stenosis, congestive heart failure, excessive blood pressure at rest, R-on-T type premature ventricular contractions, ventricular tachycardia, parasystolic focus, atrial flutter, atrial fibrillation, and congenital heart disease) (2). If the duration of the graded exercise test was less than 1 min (N = 311), participants were excluded since a steady state for exercise was not reached. Thirty-four participants who were missing data on fitness, height, weight, smoking, Keys score, or alcohol consumption were excluded. Thus, the analysis sample included 5475 participants: 2585 women and 2890 men.
To account for the sampling scheme, the data were treated as a stratified random sample, with two strata: hyperlipidemics and normolipidemics (including borderline hyperlipidemics). Associations between fitness, BMI, and cancer mortality were examined using stratified Cox proportional hazards models, with the sampling strata (hyperlipidemics and normolipidemics) as the stratifying variable. These procedures enabled us to draw inferences to those screened at visit 1 (31). All hazard ratios (HR) were reexamined among nonsmokers (i.e., former and never smokers), in order to account for the potential of confounding by smoking. Quintile cut points for fitness and BMI were calculated using fitness and BMI results in participants who were recruited as part of the random sample. Because of the higher proportion of hyperlipidemics in the sample relative to the population, cancer mortality rates were calculated by averaging across lipid strata using the inverse of the sampling probability as the weight. SAS version 6.12 (Cary, NC) was used to conduct all analyses.
RESULTS
Descriptive statistics of study sample.
Table 1 shows descriptive information on the LRC sample included for analyses. Most participants stopped the exercise test beyond 90% of predicted maximal heart rate. In women, the median percent predicted maximal heart rate achieved was 95.6% (interquartile range 92.3–99.4%) and in men the median was 96.2% (interquartile range 93.0–99.4%). The mean time on the treadmill was 7.25 min for women and 9.47 min for men. MET values were extrapolated from published exercise intensities (2,8) and yielded corresponding values for these group means of 8.44 METs for women and 10.75 METs for men. These values represent the capacity required to walk on the treadmill at the corresponding speed and grade. For the sample, 86% of women and 67% of men were in the lowest physical activity category, not reporting vigorous physical activity (e.g., inactive).
Description of analysis sample at the second examination (1972–76) by gender, LRC Prevalence Study.
The mean BMI in this sample for women (25.0 kg·m−2) and men (26.8 kg·m−2) fell in the overweight range (25.0–29.9 kg·m−2). Among women, 58.2% were normal weight (18.5–24.9 kg·m−2), 27.4% were overweight (25.0–29.9 kg·m−2), and 14.4% were obese (≥30 kg·m−2). Among men, 31.1% were normal weight, 52.8% were overweight, and 16.1% were obese. The correlation between BMI and fitness was modest: r = −0.21 in women, −0.10 in men.
Cancer deaths and death rates across fitness and BMI categories.
The most common cancer sites among women were lung (N = 29), breast (N = 20), colon/rectal (N = 15), and ovarian/uterine (N = 15). For men, the most common cancer sites were lung (N = 74), prostate (N = 17), and colon/rectal (N = 15). The number of cancer deaths and the age-adjusted cancer death rates for each quintile of fitness and BMI by gender are shown in Table 2, overall and among nonsmokers. In some strata, the number of deaths became small, especially in the highest quintiles of fitness. Overall at each quintile of fitness and BMI, cancer death rates were generally higher for men relative to women, except for the highest fitness quintile.
Adjusted* cancer death rates by quintiles of fitness and BMI and by gender overall and among nonsmokers, LRC Prevalence Study.
Proportional hazards modeling on fitness, obesity, and cancer mortality.
A J- or U-shaped association has been found between BMI and mortality in many studies (23). Because of this, we examined the shape of the BMI and cancer mortality association, as well as the shape of the fitness and cancer mortality association. Examination of BMI and fitness with quadratic terms and quadratic spline terms (32) in separate models for women and men showed no improvement in the prediction of cancer mortality over the model with just the linear term. Therefore, only the results using BMI and fitness treated linearly are shown here.
Table 3 shows associations between fitness and cancer mortality by gender, before and after adjusting for continuous BMI. Among women, although the most fit two quintiles tended to have the lowest cancer mortality risk according to the HR, especially when reducing the sample to nonsmokers, none of the relationships reached statistical significance. For men, the risk of cancer mortality tended to be lowest in the highest fitness quintile, even when reducing the sample to nonsmokers. The HR changed little for fitness when further adjusting for BMI for both women and men.
Adjusted* hazard ratios (95% confidence intervals; CI) for cancer mortality by quintiles of fitness and BMI and by gender overall and among nonsmokers, LRC Prevalence Study.
Table 3 also shows associations between BMI and cancer mortality by gender, with and without adjustment for continuous fitness. The highest risk of cancer mortality tended to occur in the highest BMI quintile among women. Among men, BMI quintiles were not related to cancer mortality overall. The HR for BMI changed little when further adjusting for fitness for both women and men.
Based on our findings in Table 3, we collapsed across quintiles and compared the lowest quintile versus the other four quintiles and the highest quintile versus the other four quintiles for both fitness and BMI (Table 4). For both women and men being in the lowest quintile of fitness (hypothesized highest risk) relative to the other four quintiles did not meaningfully change the hazard for cancer mortality overall or among nonsmokers. However, cancer mortality was generally reduced for men in the highest quintile of fitness (hypothesized lowest risk) relative to the other four quintiles.
Adjusted* hazard ratios (95% confidence intervals; CI) for cancer mortality by gender overall and among nonsmokers, LRC Prevalence Study.
For women, being in the highest quintile (hypothesized highest risk) relative to the other four quintiles of BMI significantly increased cancer mortality risk overall, which was attenuated among nonsmokers only. For men, being in the highest quintile of BMI relative to the other four quintiles was not related to cancer mortality risk overall. The estimate in nonsmokers was below one; however, the confidence intervals were wide. Additionally, there appeared to be no relationship for women or men with cancer mortality when comparing the lowest quintile (hypothesized lowest risk) to the other four quintiles of BMI, overall and among nonsmokers. Further adjustment of BMI by fitness and further adjustment of fitness by BMI did not meaningfully change these results.
Interactions between fitness and obesity were tested keeping both measures continuous in the proportional hazard models. There were no significant interactions between fitness and obesity predicting cancer mortality overall (P = 0.37 women, P = 0.46 men) or among nonsmokers (P = 0.65 women, P = 0.26 men). Table 5 displays the interactions between fitness and BMI, using quintile categories from Table 4. We hypothesized that individuals in both the first BMI quintile and the fifth fitness quintile would be at lowest risk, and those in the fifth BMI quintile and first fitness quintile would be at the highest risk. There was some support for the former of the two hypotheses.
Adjusted* hazard ratios (95% confidence intervals; CI) on the interaction between fitness and BMI for cancer mortality by gender, LRC Prevalence Study.
Physical activity and cancer mortality.
We also explored whether physical activity predicted cancer mortality, adjusting for obesity, age, education, smoking, alcohol, Keys score, and menopause (women only). Among women, when compared with those who were not active, being classified as very active (HR = 1.32; 95% CI, 0.50–3.47) or moderately active (HR = 1.18; 95% CI, 0.67–2.10) was not significantly related to cancer mortality. Similarly among men, when compared with those who were not active, being classified as very active (HR = 1.33; 95% CI, 0.74–2.37) or moderately active (HR = 0.97; 95% CI, 0.70–1.35) was not significantly related to cancer mortality. There were no significant interactions between physical activity categories and continuous BMI predicting cancer mortality in either women (P = 0.34 very active, P = 0.57 moderately active) or men (P = 0.49 very active, P = 0.10 moderately active). The results were similar when restricting to only nonsmokers.
DISCUSSION
Fitness and physical activity and risk of cancer mortality.
Using the LRC Prevalence Study cohort, we found that the risk of cancer mortality was lower in the most fit quintile relative to the other four quintiles of fitness for men (HR 0.47; 95% CI, 0.27–0.81), adjusting for multiple confounders including BMI. The reduction in risk remained when reducing the sample to nonsmokers (i.e., former and never smokers), although some risk estimates included the null value. The reduction in risk also remained when further adjusting for BMI. However, cancer mortality was not significantly reduced for women comparing the most fit quintile relative to the other four quintiles of fitness (HR 0.87; 95% CI, 0.53–1.41).
The size of the associations of fitness with cancer mortality was generally similar in LRC women and men compared with those observed in other studies. In this study, the relative hazard comparing the highest quintile to the lowest quintile of fitness for cancer mortality was 0.86 (95% CI, 0.49–1.50) for women and 0.41 (95% CI, 0.22–0.74) for men, controlling for multiple covariates including BMI. The relationship between fitness and cancer mortality was reported from the Aerobics Center Longitudinal study in an earlier study (7) and later updated (19,20). Based on 44 female (N = 7080) and 179 male (N = 25,341) cancer deaths, the risk of cancer mortality was 0.47 (95% CI, 0.18–1.22) for women and 0.36 (95% CI, 0.21–0.61) for men comparing the highest quintile of fitness to the lowest quintile, adjusted for age, exam year, smoking, chronic illness, and ECG abnormalities (19). In the Canada Health Survey Mortality Follow-up Survey, fitness was assessed at home by using a submaximal step test (4). The risk of cancer mortality increased across three levels of decreasing fitness levels, adjusting for age, gender, smoking, and alcohol consumption. However, the estimates included the null value, based on only 32 cancer deaths for women and men.
We did not observe a relationship between our measure of physical activity and cancer mortality, despite the apparent relationship with fitness. This could be due to our crude estimate of physical activity, based on a two-item assessment. It is possible that in this case cardiorespiratory fitness may have been a better marker of long-term physical activity. In the Aerobics Center Longitudinal Study (19), a reduction in cancer mortality risk was observed with higher levels of physical activity among men but not women, although the confidence intervals were large and imprecise. In the Canada Health Survey, the risk of cancer mortality was not related to physical activity levels (4), but their estimates were also imprecise. Based on a comprehensive review by Thune and Furberg (27) (which included the two studies just mentioned), of 17 observational studies of physical activity and all-cause cancer mortality, 10 studies indicated a significant reduced effect of leisure or occupational activity, whereas 6 studies suggested a reduced effect and 1 study showed an increased risk. A weaker effect on this association was generally found in women as compared with men. Since then, an update was published for the British Regional Heart study showing a reduced risk for cancer in the most active compared to the least active group of men (29).
Differences across studies may be due to the timing of measurement (15). There may be periods in life when attaining certain levels of physical activity or fitness are more important in preventing cancer than in other time periods. The time periods in life when physical activity may result in decreased endogenous sex hormones have not been established (14). Unfortunately, in the LRC Prevalence Study, fitness and physical activity were only measured once. Studies documenting the changing course of physical activity and fitness over time can further our understanding of these associations.
Obesity and risk of cancer mortality.
In this study, when controlling for multiple confounders, the risk of cancer mortality was generally lower for women but not for men, comparing the lowest BMI quintile with the highest quintile. These results were not meaningfully changed when further adjusting for fitness. Our results can be compared to several studies from the Cancer Prevention Studies. In the first Cancer Prevention Study of approximately 750,000 women and men enrolled during 1959–72, cancer mortality was higher among those 40% or more overweight (16,21). The cancer mortality ratios were consistently higher in overweight women than overweight men. In the second Cancer Prevention Study of approximately 1.5 million women and men enrolled in 1982, the relationship between BMI and the risk of death from cancer was fairly linear, showing no elevation in risk among women and men in the lowest BMI category (11,17). Similar to our results and to the results from the first cohort, risk of cancer mortality was higher at a given category of BMI for women as compared with men.
Interactions between fitness and obesity on cancer mortality.
We investigated, but did not find, significant interactions between fitness and obesity for cancer mortality among women or men. This null finding remained when examining the interaction between physical activity and obesity on cancer mortality across gender. No other epidemiologic reports provide information on the interaction between fitness and obesity, but one study does provide estimates for the interaction between physical activity and obesity on cancer mortality. For women and men in the National Health and Nutrition Examination I cohort, the risk of all-cause cancer mortality increased as physical activity declined (3 levels) within BMI strata < 22.0 and 22.0–26.0 kg·m−2 (1). However, at higher levels of BMI (>26.0 kg·m−2), physical activity was not related to cancer mortality in women or men.
Limitations.
The LRC cohort was not a representative sample of the U.S. population, but nevertheless, the cohort was drawn from diverse groups and represents the clinics, work sites, geographic locations, school districts, or membership within a medical plan from which they were sampled. Our measure of fitness measure was excellent for an epidemiologic study, whereas the measures of adiposity and physical activity were less precise. The correlation between BMI and percent body fat is approximately 0.7 in adults (30). An additional limitation of this work was the statistical power, because many of the risks and interactions documented here were assessed with relatively large confidence intervals. The cut points we chose to form categories of fitness and BMI could have a large effect on the magnitude of the risks observed. We a priori chose our quintile cutpoints based on the random sample strata chosen for LRC. Also, we did not have an adequate number of cancer cases to examine deaths from certain types of cancer separately. Furthermore, prevalent cancer was not collected at baseline. To account for this limitation, we excluded cancer mortality events that occurred during the first 4 yr of follow-up.
CONCLUSIONS
In conclusion, these data indicate that higher fitness levels may reduce the risk of cancer mortality for men. Additionally, not being obese may reduce the risk of cancer mortality among women. Interactions between fitness and obesity on cancer mortality were not identified. A working group for the International Agency for Research on Cancer of the World Health Organization recently published a summary on physical activity and weight control in the context of cancer prevention (5). They recommended that prevention of overweight and obesity should begin early and individuals should be encouraged to perform regular physical activity. The American Cancer Society also recently updated their guidelines on physical activity and nutrition (10), supporting a physically active lifestyle. Taking these steps will reduce the risk of cancer, as well as cardiovascular disease, diabetes, and total mortality (5).
This research was supported by a grant from the Centers for Disease Control and Prevention (no. U48/CCU409660).
REFERENCES
1. Albanes, D., A. Blair, and P. Taylor. Physical activity and risk of cancer in the NHANES I population. Am. J. Public Health 79: 744–750, 1989.
2. American College of Sports Medicine. ACSM’s Guidelines for Exercise Testing and Prescription. Philadelphia: Lippincott, Williams & Wilkins, 2000.
3. Anderson, J., D. Jacobs, Jr., N. Foster, et al. Scoring systems for evaluating dietary pattern effect on serum cholesterol. Prev. Med. 8: 525–537, 1979.
4. Arraiz, G., D. Wigle, and Y. Mao. Risk assessment of physical activity and physical fitness in the Canada Health Survey Mortality Follow-up Study. J. Clin. Epidemiol. 45: 419–428, 1992.
5. Bianchini, F., R. Kaaks, and H. Vainio. Weight control and physical activity in cancer prevention. Obes. Rev. 3: 5–8, 2002.
6. Blair, S., and S. Brodney. Effects of physical inactivity and obesity on morbidity and mortality: current evidence and research issues. Med. Sci. Sports Exerc. 31: S646–S662, 1999.
7. Blair, S., H. Kohl III, R. Paffenbarger, Jr., D. Clark, K. Cooper, and L. Gibbons. Physical fitness and all-cause mortality: a prospective study of healthy men and women. JAMA 262: 2395–2401, 1989.
8. Bruce, R. Exercise testing for ventricular function. N. Engl. J. Med. 296: 671–675, 1977.
9. Bruce, R., L. Fisher, M. Cooper, and G. Gey. Separation of effects of cardiovascular disease and age on ventricular function with maximal exercise. Am. J. Cardiol. 34: 757–763, 1974.
10. Byers, T., M. Nestle, A. Mctiernan, et al., for the American Cancer Society 2001 Nutrition and Physical Activity Guidelines Advisory Committee. American Cancer Society guidelines on nutrition and physical activity for cancer prevention. CA Cancer J. Clin. 52: 92–119, 2002.
11. Calle, E., M. Thun, J. Petrelli, C. Rodriguez, and C. Heath, Jr.. Body mass index and mortality in a prospective cohort of U.S. adults. N. Engl. J. Med. 341: 1097–1105, 1999.
12. Centers for Disease Control and Prevention. Physical activity trends: United States, 1990–1998. MMWR 50:166–169, 2001.
13. Devesa, S. D., Grauman, W. Blot, G. Pennello, R. Hoover, and J. Faraumeni, J r. Atlas of Cancer Mortality in the United States, 1950–94. Rockville, MD: National Institutes of Health, National Cancer Institute, 1999.
14. Friedenreich, C. Physical activity and cancer prevention: from observational intervention research. Cancer Epidemiol. Biomarkers Prev. 10: 287–301, 2001.
15. Friedenreich, C. Physical activity and cancer: lessons learned from nutritional epidemiology. Nutr. Rev. 59: 349–357, 2001.
16. Garfinkel, L. Overweight and cancer. Ann. Intern. Med. 103: 1034–1036, 1985.
17. Garfinkel, L., and S. Stellman. Mortality by relative weight and exercise. Cancer 62: 1844–1850, 1988.
18. Howe, H. L., P. A. Wingo, M. J. Thun, et al. Annual Report to the Nation on the Status of Cancer (1973 Through 1998), featuring cancers with recent increasing trends. J. Natl. Cancer Inst. 93: 824-4-42, 2001.
19. Kampert, J., S. Blair, C. Barlow, and H. Kohl III. Physical activity, physical fitness, and all-cause and cancer mortality: a prospective study of men and women. Ann. Epidemiol. 6:452–457, 1996.
20. Lee, C., and S. Blair. Cardiorespiratory fitness and smoking-related total cancer mortality in men. Med. Sci. Sports Exerc. 34: 735–739, 2002.
21. Lew, E., and L. Garfinkel. Variations in mortality by weight among 750,000 men and women. J. Chron. Dis. 32: 563–576, 1979.
22. Lipid Research Clinics Program Epidemiology Committee. Plasma lipid distributions in selected North American populations: the LRC Program Prevalence Study. Circulation 60:427–439, 1979.
23. Manson, J., M. Stampfer, C. Hennekens, and W. Willett. Body weight and longevity. JAMA 257: 353–358, 1987.
24. Mokdad, A., M. Serdula, W. Dietz, B. Bowman, J. Marks, and J. Koplan. The spread of the obesity epidemic in the United States, 1991–1998. JAMA. 282: 1519–1522, 1999.
25. Sheffield, L., W. Haskell, G. Heiss, et al. Safety of exercise testing volunteer subjects: the Lipid Research Clinics’ Prevalence Study experience. J. Cardiac Rehabil. 2: 395–400, 1982.
26. Stevens, J. Obesity and mortality in African Americans. Nutr. Rev. 58: 346–358, 2000.
27. Thune, I., and A. Furberg. Physical activity and cancer risk: dose-response and cancer, all sites and site-specific. Med. Sci. Sports Exerc. 33: S530–S550, 2001.
28. US-USSR Steering Committee. The pathogenesis of atherosclerosis: collaborative US-USSR Study on the prevalence of dyslipoproteinemia and ischemic heart disease in American and Soviet populations. Am. J. Cardiol. 40:260–268, 1977.
29. Wannamethee, S., A. Shaper, and M. Walker. Physical activity and risk of cancer in middle-aged men. Br. J. Cancer 85: 1311–1316, 2001.
30. Willett, W. Anthropometric measures and body composition. In: Nutritional Epidemiology, B. MacMahon (Ed.). New York: Oxford University Press, 1990, pp. 217–244.
31. Williams, O., S. Stinnett, L. Chambless, et al. Populations and methods for assessing dyslipoproteinemia and its correlates: the Lipid Research Clinics Program Prevalence Study. Circulation 73: 4–11, 1986.
32. Witte, J., and S. Greenland. A nested approach to evaluating dose-response and trend. Ann. Epidemiol. 7: 188–193, 1997.
Keywords:
OBESITY; NEOPLASMS; EPIDEMIOLOGY; PHYSICAL FITNESS; LEISURE ACTIVITIES; SURVIVAL ANALYSIS
©2003The American College of Sports Medicine