Journal Logo


Relationship between leisure-time physical activity and selected dietary variables in the Worcester Area Trial for Counseling in Hyperlipidemia


Author Information
Medicine & Science in Sports & Exercise: September 1997 - Volume 29 - Issue 9 - p 1199-1207
  • Free


Physiologic risk factors and some health behaviors have been observed to cluster together, suggesting that individuals who practice one beneficial health behavior may practice several beneficial behaviors(4,10,14,15). This presents a challenge for epidemiologic research that attempts to identify and estimate the effect of putative etiologic factors contributing to the development of disease, often using the simplifying assumption that they are independent of one another. Both Leisure-time Physical Activity (LTPA) and various dietary factors have been independently associated with several chronic diseases, including coronary heart disease and cancers of the prostate, colon, and breast(2,18,27,28,31,37,39). Some, but not all, of these investigations have controlled for the effects of both LTPA and dietary factors. Because relatively little is known about the strength of the association between LTPA and dietary behaviors, it is unclear whether the lack of control for either one of these factors influenced the findings of some of these investigations. Furthermore, unless it is known how strongly and in what way LTPA and diet are related, it is unclear what type and level of analytic control would be necessary to estimate adequately the effect of either behavior in an investigation.

Several lines of evidence suggest that LTPA and dietary behaviors are related. At least three studies have been conducted using principal components analysis to examine the relationship between a number of health behaviors, including LTPA and diet. Each of these investigations report that LTPA and diet “loaded” on the same factor, suggesting that the two behaviors are related (19,40,43). Simoes et al. (36) report an inverse relationship between LTPA and fat intake in a population sample of the United States in examining data from the 1990 Behavioral Risk Factor Surveillance System. Eaton et al.(11) report that more physically active individuals participating in the Pawtucket Heart Health Program consumed less total fat and saturated fat than did inactive participants. These findings are consistent with several smaller investigations which have contrasted the diets of athletes and individuals who do not exercise regularly. Some of these studies report that athletes and inactive controls consumed an equal proportion of calories from fat sources (3,23). However, others report athletes consume fewer calories from fat sources(30,41).

Currently, there is relatively little information regarding the relationship between LTPA behavior and intake of fruits, vegetables, fiber, and various micronutrients. Pate et al. (29) recently found that physical inactivity was associated with a lack of fruit and vegetable intake among adolescents surveyed in the Youth Risk Behavior Survey. Sedentary participants in the Pawtucket Heart Health Program consumed fewer fruits and vegetables and a less nutrient dense diet compared with more active participants (11). In addition, data from the Health Professionals Follow-up and Nurses Health studies suggest a positive relationship between exercise participation and Vitamin E intake(34,39). Together, these data suggest some relation between LTPA participation and consuming a diet of greater micronutrient density.

In etiologic studies of chronic disease the potential for mixing (i.e., confounding) of LTPA and dietary exposures could create challenges with respect to analytic control, especially if the relationships are strong(17,35). Therefore, quantitative estimates of the degree to which these exposures are related should advance our understanding of the independent effects of both LTPA and dietary factors on various chronic diseases. It also could provide information regarding the degree of analytic control needed to account for each behavior in future etiologic studies. Additionally, knowledge of such relationships could inform the development and delivery of interventions aiming to change either behavior.


Subjects. The subjects were participants in the Worcester Area Trial for Counseling in Hyperlipidemia (WATCH, NHLBI # HL 44492-04). The WATCH study was designed to evaluate the effectiveness of two different methods of physician delivered nutrition intervention to reduce saturated fat intake and blood levels of low density lipoprotein cholesterol (LDL-C) among hyperlipidemic individuals (26). Participants in the WATCH study were 1,275 men and women aged 20-65 yr who were recruited because they were above the sex-, age-, and season-specific 75th percentile for LDL-C for adults in the Fallon Health Care System (25). They were recruited from the patient population at 12 practice sites of the Fallon Health Care System, a Health Maintenance Organization serving central Massachusetts, southern New Hampshire, northern Rhode Island, and northeastern Connecticut. Written consent was obtained for all participants. Recruitment into the WATCH study was completed between March 1991 and December 1993 and subjects were followed for 1 yr. A detailed description of the WATCH recruitment methodology is available elsewhere (26). Dietary and LTPA data collected at baseline in the WATCH study form the basis for the present analyses.

Dietary assessment. In the WATCH study, dietary data were collected using a 7-d diet recall (7DDR) and a single 24-h diet recall (24HR)(13,16). The 24HR data were originally to be used for the definitive group-level analysis of the overall effect of the WATCH because the 7DDR was untested in the context of evaluating an intervention at the beginning of the trial. In the present analyses the 24HR data were employed as a concurrent dietary assessment along with the 7DDR.

The 7DDR was developed specifically for the WATCH and was designed to quantify changes in lipid intake. This instrument is a structured self-assessment of 118 foods and 13 beverage items. Subjects report the portion size and frequency of consumption (number of times) for all food items consumed in the past week. Two-dimensional portion size models were provided to improve the reporting accuracy of the volume of food consumed. In addition, a 7-d grid is incorporated into the second page of the questionnaire (i.e., immediately before the questionnaire proper). The grid was used to assist subjects in prompting recall of breakfasts, lunches, dinners, and between meal snacks in the 7 d preceding the dietary assessment. Space was available to report foods not listed on the questionnaire. Typically, written-in foods contribute less than 1% of total food intake reported in this population.

The Nutrition Data System food data base (NDS 2.3) developed by the Nutrition Coordinating Center of the University of Minnesota, Minneapolis, MN was used to convert foods on the 7DDR to nutrient data(7). Foods identified during the assessment that were of a complex nutritional character (e.g., casseroles, stews) were considered“composite foods”. Foods of this type were converted to nutrient data using a combination of the appropriate NDS food codes that approximated the fat content of the composite food consumed.

An inter-method validation study of the 7DDR using a series of seven 24HR as a comparison measure found the 7DDR to provide a good estimate of total energy intake, lipid, cholesterol, and alcohol consumption in 27 women and 14 men (16). Correlations and beta coefficients in linear regression between the 7DDR and 24HR were: r = 0.72 and b = 0.97; r = 0.72 and b = 1.00; r = 0.69 and b = 0.55; r = 0.66 and b = 0.67 for total energy intake, total fat, cholesterol, and alcohol intake, respectively. The 7DDR produced micronutrient scores that were correlated to the 24HR to a lower degree (r = 0.26 to 0.68) (16).

Because the 7DDR was designed to measure lipid intake and therefore may provide less reliable micronutrient data, the available 24HR nutrient data were examined to provide additional information for micronutrient intake. Although nutrient data obtained from 24HR are limited in their ability to provide reliable estimates of habitual nutrient intake, they do provide good estimates of mean nutrient intake for groups of individuals(42). The 24HR was administered over the telephone on a randomly selected day of the week. In the interview the subjects were instructed to use “time-of-day” cues to enhance recall of meals, snacks, and beverages consumed on the calendar day before the interview. The interviewers used standardized probing techniques to aid in identifying the method of preparation, portion sizes, and the amount of condiments added to the foods consumed. Data obtained during the telephone interview were entered directly into the NDS data entry program and checked for internal consistency and completeness while the subject was still on the phone.

Leisure-time physical activity assessment. Information regarding participation in LTPA was obtained by questionnaire as a part of the 7DDR. Individuals were asked, “Do you exercise on a regular basis?”, and subjects who acknowledged exercising regularly then completed questions detailing the frequency (times per week), duration (minutes per session), and mode of exercise in which they participated. Individuals who did not answer the first question and provided no other LTPA information were assumed to engage in no explicit LTPA. Twelve activities were listed and space was provided for writing in unlisted activities. Subjects also provided information on the number of years they had been participating in LTPA (seeAppendix 1). The reliability and validity of this LTPA assessment has not been formally completed.

From these data, categories of LTPA in minutes per week(min·wk-1) were constructed. Individuals reporting a frequency of “less than once per week” were coded with a frequency of 0.5 sessions, and those reporting “more than four” were coded with 5 sessions per week. Individuals reporting “less than 10 min” per session were coded as 10 min, and those reporting “more than 40 min” were coded as 50 min per session. Otherwise, the midpoint of the duration interval was used for tabulation of min·wk-1 of LTPA by multiplying the recoded duration and frequency data together. Individuals who reported 0 to 29 min·wk-1 of LTPA were classified as“inactive,” and those reporting 30 min·wk-1 or more of LTPA each week were classified as “active.” Active subjects were further categorized into three levels of LTPA; 30 to 60, 61 to 120, and≥ 121 min·wk-1. LTPA energy expenditure per week (LTPA-EE, kcal·wk-1) was calculated using the LTPA min·wk-1 data, reported mode of activity (e.g. running, walking), and self-reported body mass. For these calculations, each activity reported was weighted with a metabolic equivalent (MET) intensity score using the compendium of physical activities compiled by Ainsworth et al. (1). Individuals who reported more than one activity were given the MET intensity score of the activity with the higher MET level. Estimation of LTPA energy expenditure in kcal assumed that resting metabolic rate (1 MET) was equal to 1 kcal·kg-1·h-1(1). Individuals also were categorized by level of intensity and length of their LTPA participation. LTPA intensity was classified as moderate (3 to 6 METs) and vigorous (≥6 METs). Length in years of LTPA participation was classified as short (<1 yr) and long (≥1 yr).

Statistical analysis. Mean and frequency distribution differences across the LTPA categories were evaluated for both males and females for the descriptive covariates using ANOVA and Chi-square testing. ANOVA was employed to compare the group means of the dietary data across the four LTPA levels. The dietary data were analyzed as both foods and nutrients. In unadjusted analyses, the dependent variable was the nutritional variable of interest and the independent variable was the LTPA category (four levels). Statistical testing was completed for the overall significance of the ANOVA and by comparing the mean of the inactive group to the means of each of the active LTPA groups using regression methods (Stata Software, Stata Corporation). This approach results in a large number of group mean comparisons (3 comparisons per food/nutrient (total 43) = 129 total comparisons). The type I error rate per food or nutrient is expected to be 14% (20), resulting theoretically in a total of 18 type I errors within the 129 mean comparisons. ANCOVA was employed to adjust the dietary and LTPA group comparisons for age, gender, educational attainment, and smoking status. Educational attainment was coded using three categories: 1) ≤ high school, 2) vocational, trade school, some college or an associate degree, or 3) a college degree, or greater. Smoking history was dichotomized into current smokers and nonsmokers. Differences between males and females in the relationship between LTPA and diet was assessed by examining the significance levels of the overall F tests for all foods and nutrients for both genders. A statistically significant finding (P ≤ 0.05) in one gender but not the other for a given food or nutrient was considered to constitute a gender difference in this relationship. The Macintosh-based statistical package Stata 4.0 (Stata Corporation, College Station, TX) was used in all analyses.

Several food categories were created by grouping food items from the 7DDR into logical categories based on the fat content of the item. For instance, the number of servings of meat (e.g. hamburgers, hot dogs, pork, beef, lamb) were collapsed together to calculate the total servings of meat consumed in the past week. The number of servings per week of high-fat sweets were tabulated by adding the total number of servings of donuts, cakes, cookies, ice cream, fruit pies, cream pies, and puddings (low-fat varieties excluded). Similarly, groupings of fried foods, starchy vegetables, vegetables, fruits, and high-and low-fat dairy products were tabulated. The distributions of the food variables were not normally distributed (e.g. skewness > 3 and/or kurtosis > 7), and accordingly, they were transformed to achieve approximately normal distributions. This was accomplished by adding the value“1” to each food variable (servings·wk-1) and then taking its natural log. This transformation resulted in the approximate normalization of the distributions of the tabulated food variables (e.g. skewness < 3 and kurtosis < 7).

The following nutrient data derived from the 7DDR were examined: total energy (kcal); grams of carbohydrate, protein, total fat; and the proportion of energy from carbohydrate, protein, total fat, saturated fat (SFA), polyunsaturated fat (PFA), and monounsaturated fat (MFA). In addition, cholesterol, dietary fiber, and selected vitamins and minerals were examined after they were adjusted for energy intake using the nutrient density method(e.g., U·1000 kcal-1) (42). Alcohol consumption was evaluated in grams per week and was found to be positively skewed. Therefore, alcohol consumption was transformed in the manner of the foods data to achieve an approximately normal distribution.


Subjects. Of the 1,275 individuals recruited into the WATCH study, 1,067 completed baseline 7DDR's. Eighteen (1.7%) of these subjects were excluded from the present analyses because their 7DDR energy intake was less than 500 or greater than 5000 kcal·d-1, leaving a sample size of 1,049. Individuals with incomplete age, gender, height, body mass, education, smoking, or LTPA data were also excluded (N = 130). Therefore, complete diet, LTPA, and covariate information were available for 919 (86.1%) WATCH subjects. Exclusion of subjects with missing covariate information did not materially influence the present results because unadjusted analyses using all 1,049 subjects were virtually identical to those which follow.

Mean age across the four LTPA categories was similar for each gender, except in the 61 to 120 min·wk-1 group (Table 1). In this LTPA group, the mean age among females was approximately 4 yr younger(P ≤ 0.05), compared with their inactive counterparts. Body mass(kg) tended to be lower with greater participation in LTPA and was significantly lower (P ≤ 0.05) among women participating in LTPA more than 60 min·wk-1. Body mass index (BMI) was significantly lower in males and females participating in LTPA more than 60 min·wk-1, compared to the inactive group. The proportion of smokers tended to decrease as LTPA participation increased for all subjects, but was statistically significant among females (X23df = 7.9, P= 0.05), and not males (X23df = 3.4, P = 0.33). Additionally, a greater educational attainment tended to be associated with increased participation in LTPA for both males and females (X26df = 11.9, P = 0.06 and X26df = 16.2, P = 0.01, respectively).

Foods. Eight of 12 food groups had statistically significant differences in mean food intake over LTPA levels (Table 2). Individuals who participated in LTPA for 30 min or more each week reported consuming fewer high-fat foods (meats, fried foods, high fat sweets, and 2% and 4% milk) and more servings of low-fat, micro-nutrient rich foods(fruits, vegetables, and low fat dairy products) than inactive subjects. For instance, inactive subjects reported a mean vegetable consumption of 4.6 servings·wk-1 which was significantly lower (P ≤ 0.05) than the mean vegetable consumption in the three successive LTPA categories (e.g. 5.7, 5.9, and 6.4 servings·wk-1, respectively). In general, adjustment for age, gender, education, and smoking did not alter the statistical differences observed in mean food intake across levels of LTPA(Table 2).

Macronutrients. Comparison of mean macronutrient intake across four levels of LTPA revealed no statistically significant differences(P ≤ 0.05) in total energy, carbohydrate, protein, or fat intake(Table 3). The lower consumption of total fat observed in the LTPA category 30-60 min·wk-1 in comparison with the inactive LTPA group was statistically significant in ANOVA but not after adjustment for age, gender, education, and smoking. The overall F test in ANOVA was statistically significant for four of six of the macronutrients expressed as a percent of total energy. Protein as a proportion of total energy intake was significantly greater among the 30-60 and ≥ 121 min·wk-1 LTPA categories compared with the inactive group. In contrast, the percent of energy from total fat and MFA was lower in each LTPA category compared with that in the inactive category, except in the 61-120 min·wk-1 category. Percent SFA was significantly lower in each active LTPA category. Intake of cholesterol was lower among the active group in comparison with the inactive group, but only significantly so in the 61- to 120-min·wk-1 LTPA category. Mean intake of dietary fiber was greater among two of the three active LTPA categories. Alcohol consumption was greater among the two upper LTPA categories compared with the inactive group and reached statistical significance among subjects in the ≥ 121 min·wk-1 category (Table 3). These ANOVA results were not altered materially by adjustment for age, gender, education, and smoking.

Micronutrients. Ten of 17 of the mean vitamin and mineral intakes were statistically significant across the LTPA groups (Table 4). Active individuals consumed a diet of a higher nutrient density compared with their inactive counterparts, and there were significant differences between the inactive and individual active LTPA groups for the minerals including calcium, potassium, magnesium, and the vitamins including beta-carotene, ascorbic acid, riboflavin, niacin, pantothenic acid, folic acid, and vitamins B6(Table 4). These results were not generally influenced by adjustment for age, gender, education, and smoking.

24-H Recall results. Analyses using the 24HR data generally confirm the previously reported 7DDR findings in a smaller sample of WATCH subjects with complete 24HR and covariate data (N = 756). Differences were significant between active and inactive individuals for percent of energy from total fat, iron, potassium, magnesium, vitamin B12, and cholesterol (P ≤ 0.05). There were marginally significant differences for percent of energy from SFA and MFA and for ascorbic, pantothenic, and folic acids (P > 0.05 to P≤ 0.10). Nonsignificant trends of increasing nutrient density with increasing min·wk-1 of LTPA were observed for copper, zinc, beta-carotene, retinol, riboflavin, thiamin, niacin, and vitamin B6(data not shown). While the large degree of variability related to the use of a single 24HR limits direct comparison with the 7DDR results, the consistency of the findings using a second dietary assessment suggests that the significant dietary differences observed using the 7DDR are not a result of idiosyncrasies of the 7DDR.

Gender differences. In general, relationships between LTPA and good dietary practices were observed in both males and females. However, there were a few gender differences in the LTPA-diet relationship (data not shown). In terms of the foods analyses, there were significant differences by LTPA category for meats, fried foods, and low-fat dairy products only among females, and there were significant LTPA findings for high-fat sweets, vegetables, and eggs among males, but not females. Nutrient intake differences over LTPA levels were observed for the variables percent of energy from protein and cholesterol only among males. In addition, differences were statistically significant for percent of energy from total fat and MFA, iron, retinol, thiamin, niacin, and B12 over LTPA levels for females, but not males.

Length and intensity of LTPA participation. Individuals participating in either moderate (3 to 6 METs) or vigorous (≥6 METs) intensity LTPA had dietary practices similar to each other, and both were significantly better than those of inactive subjects. Similarly, longer-term participants in LTPA (≥1 yr) did not have substantially different nutrient intakes than did shorter-term participants (<1 yr), and both were generally lower in fat and higher in vitamin density compared with inactive subjects(data not shown).


The present investigation found that individuals in the WATCH who participated in LTPA for at least 30 min·wk-1 consumed fewer servings of high-fat foods (e.g. meats, sweets, and fried foods), and more servings of micronutrient and fiber-rich foods (e.g. fruits and vegetables). In addition, these more physically active individuals consumed more grams of alcohol each week than did their inactive counterparts. The dietary differences between inactive and active subjects remained after controlling for total energy intake, age, gender, smoking, and educational attainment.

Our results are consistent with previous investigations that have used factor analysis to examine the mixing of broadly defined diet and physical activity behaviors (19,22,40,43). They also are consistent with two population-based studies that examined the relation between LTPA and the intake of fat (36) and fruits and vegetables (29). The present study extends these investigations by providing a more detailed description of the patterns of nutrient intake across a range of physical activity levels. Our findings are in contrast to some studies that have compared athletes to inactive controls (3,6,23). Given the consistency of our results with the population-based studies and several smaller studies(11,24,30,41), it seems likely that the lack of association between diet and LTPA in some previous investigations may be a result of at least two factors: 1) temporal changes in nutritional behavior in the last decade among active individuals, or 2) a lack of variation in dietary intake within small convenience samples that is available to larger population samples.

Contrary to our expectation, we found that reported total energy intake was not substantively greater among more active individuals in the WATCH population. Although many studies have observed a positive relationship between energy intake and LTPA participation(3,24,38,41), there are several potential reasons why we did not observe this relationship in the WATCH population. First, the physical activity assessment employed in the WATCH examined only leisure-time activities. Therefore, only a partial picture of the total daily energy expenditure was captured. It is possible that active individuals compensate for their LTPA energy expenditure by expending less energy in other portions of their day (12). Second, an under-reporting of dietary intake and over-reporting of LTPA is also conceivable. Hebert et al. (13) provide evidence that certain personality types are likely to under-report total energy intake based on their inclination to provide socially desirable responses. These same individuals also may be more likely to over-report LTPA participation. The net result of each of these biases could be an apparent equalization of energy intake over LTPA levels. Finally, it is possible that the inactive group in this population consumed an equal number of calories compared with those of active subjects. Inactive individuals in the WATCH had a greater BMI than more active WATCH subjects. It has been reported that individuals with more body mass expend more total energy each day than individuals with less body mass because their absolute resting energy expenditure is greater owing to their larger fat free mass, at an equal physical activity level(32,33). Consequently, subjects with a greater body mass may be expected to have larger total energy intake as compared to individuals with less body mass.

While we generally observed more healthy dietary patterns among active subjects, these patterns were not always consistent across the active LTPA categories, or between males and females. For instance, active subjects reporting LTPA participation of 60 to 120 min·wk-1, a slightly younger group in the WATCH study, were more likely to consume a diet resembling the inactive LTPA group (e.g.,% energy from total fat 38.7% vs 39.8%, respectively). However, there were specific nutrients for which this LTPA group consumed a diet that was consistent with the other active LTPA categories (e.g.% energy from SFA). Interestingly, individuals reporting only 30 to 60 min of LTPA each week consumed consistently the lowest fat and most micronutrient dense diet of any LTPA group in this population. It is plausible that the relatively low total caloric intake in this group had the effect of exaggerating their nutrient intakes which were adjusted for total energy intake. However, we think that the total food intake patterns presented inTable 2 are entirely consistent with the lower energy and fat intake and the more micronutrient dense diet for this group of subjects. We also observed some gender differences in the relationship between diet and LTPA. Active females consumed a diet more rich in iron and several micronutrients compared with inactive females, whereas men showed no differences between activity categories for some of these nutrients. Taken together, these findings demonstrate considerable variation in dietary behavior even among LTPA participants. Therefore, in etiologic investigations on the effect of LTPA on health outcomes, careful control of the specific foods or nutrients of interest may be required to control for dietary differences between sedentary and physically active groups, and even among active individuals.

The present study is limited by its restriction to individuals in the upper quartile of the LDL-C distribution and by the use of self-reported data. Individuals entered the WATCH study because they were determined to have an elevated LDL-C level and consented to entering a randomized trial. Based on their hyperlipidemic status, WATCH participants would be expected to consume a higher fat, less micronutrient rich diet, and to have a tendency to be overweight compared with healthy individuals.

Younger WATCH subjects had a higher prevalence of being overweight, whereas older subjects were comparable to the national average derived from prevalence estimates of “overweight” in the third National Health and Nutrition Examination Survey (NHANES III, e.g., BMI ≥ 27.8 and 27.3 kg·m2 for males and females, respectively)(21). The prevalence of “overweight” in WATCH compared with the NHANES data was 45% versus 30% and 47% versus 31% for males and females 20-49 yr of age, and 43% versus 42% and 48% versus 47% among males and females 50-69 yr of age, respectively (21). These data are consistent with the relatively high degree of inactivity in the WATCH population. Although there are differences in the ascertainment of LTPA in the WATCH (e.g., usual LTPA habits) and the Centers for Disease Control and Prevention's Behavioral Risk Factor Surveillance System (BRFSS, e.g., LTPA in past month) (8); approximately 70% of the WATCH population would be considered “sedentary” using the definition of little or no LTPA employed by the BRFSS (e.g. fewer than three sessions or less than 20 min per session). This figure is higher than the 46% of Massachusetts adults reported to be sedentary in 1992. 1 In terms of diet, the WATCH subjects consumed a diet of greater fat and SFA content (38.9 and 12.8% of total energy intake), compared to NHANES III (34.1 and 11.7% of total energy intake) for adults 20-69 yr of age(9). In fact, according to the 7DDR, subjects ate a diet with a fat content (as a proportion of energy intake), on average, about 1 SD above the population mean value.

Self-reported dietary and LTPA data could be influenced by reporting biases. The 7DDR minimized some potential biases by relying on a short period of recall that also asked about a specific time period (7 d), rather than asking about “usual” intake. The assessment of diet by two different methods (e.g., 7DDR and 24HR) in two different settings, each revealing the same dietary patterns across LTPA levels, lend strength to the findings which we have reported. The large number of statistical comparisons between the inactive and the individual active LTPA levels theoretically would result in 18 type I errors. Our finding that more than 50 of the comparisons were statistically significant suggest that the great majority of the statistical comparisons found to be significant were not a result of chance. Moreover, the consistency of our findings with previously published population studies argue against these limitations impacting the validity of our findings. Our finding of a relationship between LTPA and dietary behavior, even within a hyperlipidemic sample, a presumably more homogeneous group with respect to diet and LTPA, underscores the importance of considering each behavior as a potential confounder or effect modifier in etiologic research even within more homogeneous “high risk” groups.

In conclusion, we found that participation in LTPA was associated with the consumption of a diet that was lower in fat and cholesterol and higher in several micronutrients, fiber, and alcohol consumption. Because many of these dietary factors have been associated with outcomes of public health importance, these findings have implications for etiologic investigations on outcomes that have both LTPA and diet as risk factors. The fact that these behaviors are related to one another as well as to the same outcomes stresses the need to consider the role of effect modification (especially synergy) in future work relating physical activity or diet to specific risk factors or health outcomes. Finally, the present findings should be considered when targeting public health interventions for both diet and physical activity.


1Brooks, D. BRFSS Results for Massauchetts, 1992. personal communication, 1996.
Cited Here


1. Ainsworth, B., W. Haskell, A. Leon, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med. Sci. Sports Exerc. 25:71-80, 1992.
2. Bernstein, L., B. Henderson, R. Hanisch, J. Sullivan-Halley, and R. Ross. Physical exercise and reduced risk of breast cancer in young women. J. Natl. Cancer Inst. 86:1403-1408, 1994.
3. Blair, S., N. Ellsworth, W. Haskell, M. Stern, J. Farquhar, and P. Wood. Comparison of nutrient intake in middle-aged men and women runners and controls. Med. Sci. Sports Exerc. 13:310-315, 1981.
4. Blaxter, M. Health and Lifestyles. London and New York: Tavistock/Routledge, 1990, pp. 113-146.
5. Deleted in proof.
    6. Butterworth, D., D. Neiman, B. Underwood, and K. Linsted. The relationship between cardiorespiratory fitness, physical activity, and dietary quality. Int. J. Sports Med. 4:289-298, 1994.
    7. Buzzard, M., K. Price, and R. Warren. Considerations for selecting nutrient-calculation software: evaluation of the nutrient database.Am. J. Clin. Nutr 54:7-9, 1991.
    8. Casperson, C. and R. Merrit. Physical activity trends among 26 states, 1986-1990. Med. Sci. Sports Exerc. 27:713-720, 1994.
    9. Centers for Disease Control. Daily dietary fat and total food-energy intakes-third National Health and Nutrition Examination Survey, phase 1, 1988-1991. Morbidity and Mortality Weekly Report 43:116-125, 1994.
    10. Criqui, M., E. Barrett-Connor, M. Holdbrook, M. Austin, and J. Turner. Clustering of cardiovascular disease risk factors. Prev. Med. 9:525-533, 1980.
    11. Eaton, C., J. McPhillips, K. Gans, et al. Cross-sectional relationship between diet and physical activity in two southeastern New England communities. Am. J. Prev Med. 11:238-244, 1995.
    12. Goran, M. and E. Poehlman. Endurance training does not enhance total energy expenditure in healthy elderly persons. Am. J. Physiol. (Endocrinol. Metab.) 263:E950-E7, 1992.
    13. Hebert, J., L. Clemow, L. Pebert, I. Ockene, and J. Ockene. Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Int. J. Epidemiol. 24:389-398, 1995.
    14. Hebert, J. and G. Kabat. Differences in dietary intake associated with smoking status. Eur. J. Clin. Nutr 44:185-193, 1990.
    15. Hebert, J. and G. Kabat. Implications for cancer epidemiology of differences in dietary intake associated with alcohol consumption. Nutr. Cancer 15:107-119, 1991.
    16. Hebert, J., I. Ockene, L. Botelho, R. Luippold, P. Merriam, and G. Saperia. Development and validation of a seven-day diet recall. Presented at American Public Health Association 120th Annual Meeting, Washington, DC, 1992.
    17. Hennekens, C. and J. Buring. Epidemiology in Medicine. Boston, MA: Little Brown and Company, 1987, pp. 287-293.
    18. Hunter, D., J. Manson, G. Colditz, et al. A prospective study of vitamins C, E, and A and the risk of breast cancer. N. Engl. J. Med. 329:234-240, 1993.
    19. Kannas, L. The dimensions of health behavior among young men in Finland. Int. J. Health Educ. 14:146-151, 1981.
    20. Kirk, R. Experimental Design: Procedures for the Behavioral Sciences. Belmont, CA: Brooks/Cole Publishing Company, 1982.
    21. Kuczmarski, R., K. Flegal, S. Campbell, and C. Johnson. Increasing prevalence of overweight among US adults: the National Health and Nutrition Examination Surveys, 1960 to 1991. JAMA 272:205-211, 1994.
    22. Langlie, J. Interrelationships among preventive health behaviors: a test of competing hypotheses. Public Health Reports 94:216-225, 1979.
    23. Moore, C., H. Hartung, R. Mitchell, C. Kappus, and J. Hinderlitter. The relationship of exercise and diet on high-density lipoprotein cholesterol levels in women. Metabolism 32:189-196, 1983.
    24. Neiman, D., J. Butler, L. Pollett, and R. Lutz. Nutrient intake of marathon runners. J. Am. Diet. Assoc. 89:1273-1278, 1989.
    25. Ockene, I., J. Hebert, and D. Chiriboga.Seasonal variation of cholesterol levels. Presented at 3rd International Conference on Preventive Cardiology, Oslo, Norway, 1993.
    26. Ockene, I., J. Hebert, J. Ockene, P. Merriam, T. Hurley, and G. Saperia. Effect of training and a structured office practice on physician-delivered nutrition counseling: The Worcester-Area Trial for Counseling in Hyperlipedemia (WATCH). Am. J. Prev. Med. 12:252-258, 1996.
    27. Oliveria, S., H. Kohl, D. Trichopoulos, and S. Blair. The association between cardiorespiratory fitness and prostate cancer.Med. Sci. Sports Exerc. 28:97-104, 1996.
    28. Paffenbarger, R., R. Hyde, A. Wing, I. Lee, D. Jung, and J. Kampert. The association of changes in physical activity level and other lifestyle characteristics with mortality among men. N. Engl. J. Med. 328:538-545, 1993.
    29. Pate, R., G. Heath, M. Dowda, and S. Trost. Associations between physical activity and other health behaviors in a representative sample of U.S. adolescents. Am. J. Pub. Health 86:1577-1581, 1997.
    30. Pate, R., R. Sargent, C. Baldwin, and M. Burgess. Dietary intake of women runners. Int. J. Sports Med. 11:461-466, 1990.
    31. Pienta, K. and P. Esper. Risk factors for prostate cancer. Ann. Int. Med. 118:793-803, 1993.
    32. Poehlman, E., P. Arciero, M. Goran. Endurance exercise in aging humans: effects on energy metabolism. In: Exercise and Sport Science Reviews, J. Holloszy (Ed.). Baltimore: Williams & Wilkins, 1994, pp. 251-284.
    33. Ravussin, E., B. Burnand, Y. Schutz, E. Jequier. Twenty-four-hour energy expenditure and resting metabolic rate in obese, moderately obese, and control subjects. Am. J. Clin. Nutr 35:566-73, 1982.
    34. Rimm, E., M. Stampher, A. Ascherio, E. Giovannucci, G. Colditz, W. Willet. Vitamin E consumption and the risk of coronary heart disease in men. N. Engl. J. Med. 328:1450-2456, 1993.
    35. Rothman, K. Modern Epidemiology. Boston: Little Brown and Company, 1986, pp. 89-94.
    36. Simoes, E., T. Byers, R. Coates, M. Serdula, A. Mokdad, G. Heath. The association between leisure-time physical activity and dietary fat in American adults. Am. J. Pub Health 85:240-244, 1995.
    37. Slattery, M., M. Schumacher, K. Smith, D. West, N. Abd-Elghany. Physical activity, diet, and risk of colon cancer in Utah.Am. J. Epidemiol. 128:989-999, 1988.
    38. Smith, M., J. Mendez, M. Druckenmiller, P. Kris-Etherton. Exercise intensity, dietary intake, and high density lipoprotein cholesterol in young female competitive swimmers. Am. J. Clin. Nutr 36:251-255, 1982.
    39. Stampher, M., C. Hennekens, J. Manson, G. Colditz, B. Rosner, and W. Willet. Vitamin E consumption and the risk of coronary heart disease in women. New Eng. J. Med. 328:1444-1449, 1993.
    40. Tapp, J. and P. Goldenthal. A factor analytic study of health habits. Prev. Med. 11:724-728, 1982.
    41. Thompson, P., B. Lazarus, E. Culliname, et al. Exercise, diet, or physical characteristics as determinants of HDL-levels in endurance athletes. Artherosclerosis 46:333-339, 1983.
    42. Willett, W. Nutritional Epidemiology. New York: Oxford University Press, 1990, p. 64.
    43. Williams, A. and P. H. Wechsler. Interrelationships of preventive actions and other health areas. Health Services Reports 87:969-976, 1972.


    Do you exercise on a regular basis? ___ yes no ___; If you exercise regularly; How often do you exercise?

    ___ Less than once/wk

    ___ 2 times/wk

    ___ 3 times/wk

    ___ 4 times/wk

    ___ More than 4 times/wk


    How long do you exercise?

    ___ Less than 10 min per session

    ___ 10-20 min per session

    ___ 21-30 min per session

    ___ 31-40 min per session

    ___ More than 40 min per session

    What types of exercise do you perform?

    ___ Jogging/running

    ___ Brisk walking

    ___ Swimming

    ___ Bicycling

    Weight lifting

    ___ Rowing

    ___ Bowling

    ___ Golfing

    ___ Tennis/racquetball

    ___ Hiking

    ___ Aerobic dance

    ___ Basketball ________

    ___ Other

    How long have you been exercising regularly?

    ___ Less than 6 months

    ___ 6-12 months

    ___ 1-2 yr

    ___ 2-3 yr

    ___ More than 3 yr



    ©1997The American College of Sports Medicine