Cardiorespiratory fitness (CRF) is widely accepted as an important reversible cardiovascular risk factor. In the present study, we examined the nonmodifiable and modifiable determinants of CRF within a large healthy Caucasian population of men and women. The study included 20,239 patients presenting to Cooper Clinic (Dallas, Texas) for a comprehensive medical examination from 2000 through 2010. CRF was determined by maximal treadmill exercise testing. Physical activity categories were 0 metabolic equivalent tasks (METs)/min/week (no self-reported moderate or vigorous intensity physical activity), 1 to 449 METs/min/week (not meeting physical activity guideline), 450 to 749 METs/min/week (meeting guideline), and ≥750 METs/min/week (exceeding guideline). Linear regression modeling was used to determine the most robust clinical factors associated with achieved treadmill time. Age, gender, body mass index (BMI), and physical activity were the most important factors associated with CRF, explaining 56% of the variance (R 2 = 0.56). The addition of all other factors combined (current smoking, systolic blood pressure, blood glucose, high-density and low-density lipoprotein cholesterol, health status) were associated with CRF (p <0.05) but additively only improved R 2 by 2%. There was a significant interaction between BMI and physical activity on CRF, such that normal-weight (BMI <25 kg/m 2 ) subjects achieved higher CRF for a given level of physical activity compared to obese subjects (BMI ≥30 kg/m 2 ). Percent body fat, not lean body mass, was the key factor driving this interaction. In conclusion, BMI was the most important clinical risk factor associated with CRF other than nonmodifiable risk factors age and gender. For a similar amount of physical activity, normal-weight subjects achieved a higher CRF level compared to obese subjects. These data suggest that obesity may offset the benefits of physical activity on achieved CRF, even in a healthy population of men and women.
Cardiorespiratory fitness (CRF) is a strong predictor of cardiovascular and all-cause mortality and provides prognostic value beyond traditional risk factor assessment. Because CRF is a composite of several nonmodifiable and modifiable clinical risk factors, understanding the importance of these factors on CRF estimates is critical when assigning recommendations concerning healthy lifestyles. The primary goal of the present study was to determine the key factors that explain CRF in generally healthy men and women. Second, we explored the interplay between 2 known modifiable factors (physical activity and body mass index [BMI]) on CRF. Third, we assessed whether meeting the American College of Sports Medicine and American Heart Association guidelines for physical activity was associated with higher CRF estimates in normal-weight, overweight, and obese men and women. To achieve this goal, we used the Cooper Center Longitudinal Study (CCLS), the largest United States population database of CRF estimates with concomitant risk factor capture.
Methods
The CCLS is a prospective study composed of patients who received preventive medical examinations at the Cooper Clinic in Dallas, Texas. Most were referred by their personal physicians, employers, or were self-referred for examination. The present study included 20,329 Caucasian men and women, 20 to 90 years of age, who completed a comprehensive medical examination from 2000 through 2010 and achieved ≥85% of their maximum predicted heart rate during exercise treadmill testing. We excluded subjects seen at the Cooper Clinic before 2000 to avoid potential confounding owing to temporal population shifts in cardiovascular risk factors (i.e., smoking, BMI). We also excluded nonwhite subjects to improve our internal validity and to make direct comparisons between CCLS and National Health and Nutrition Examination Survey (NHANES) fitness estimates among non-Hispanic whites. All subjects provided written informed consent to participate in research.
Personal medical history, body composition, laboratory measurements, and assessment of CRF by maximal exercise treadmill tests were performed at the first visit to the Cooper Clinic. Information regarding age, gender, ethnicity, and medical history were obtained by questionnaires. A history of hypertension, current smoking, diabetes, or other pre-existing medical condition was verified by a physician during the clinic visit. Health status was a categorical value assigned as normal or abnormal, with abnormal defined as a personal history of heart attack, stroke, diabetes, hypertension or cancer, or an abnormal electrocardiogram at rest of during exercise based on ST segment abnormalities, as previously described.
BMI was calculated from measured weight and height during the clinic visit. BMI was treated as continuous variable and categorized as normal weight (<25 kg/m 2 ), overweight (≥25 to <30 kg/m 2 ), and obese (≥30 kg/m 2 ). Waist circumference was measured by trained staff and reported in centimeters. Skinfold measurements at 7 sites or underwater weighing measurements were used to estimate percent body fat. Lean body mass was calculated by the following equation: lean body mass = (body weight [kilograms] × [1 – {percent body fat/100}]). Blood pressure at rest was recorded using standard auscultatory methods after the patient had been seated for 5 minutes. Systolic and diastolic blood pressures were recorded at the first and fifth Korotkoff sounds, respectively. Fasting venous blood samples were obtained and plasma concentrations of lipids and glucose were determined with automated bioassays in the Cooper Clinic laboratory that meet quality control standards of the Centers for Disease Control and Prevention Lipid Standardization Program.
Physical activity was assessed by self-reported participation in recreational or leisure-time activities during the previous month. For each activity, number of sessions per week (frequency) and average duration per session were reported. From these data, we converted frequency and duration to minutes of activity per week. Each activity value was then weighted by multiplying minutes of activity by an estimated metabolic equivalent task (MET) value yielding MET minutes per week. MET values for physical activity were based on average intensity of each activity using the compendium of physical activities developed by Ainsworth et al. METs per minute per week groups were created based on American College of Sports Medicine guidelines. Categories were 0 METs/min/week (no self-reported moderate or vigorous intensity physical activity), 1 to 449 METs/min/week (not meeting guideline for physical activity), 450 to 749 METs/min/week (meeting guidelines), and ≥750 METs/min/week (exceeding guideline for physical activity). An alternative definition of physical activity was also explored, categorizing subjects into 4 groups: 0 = no organized physical activity; 1 = nonrunning activities; 2 = 0 to 10 miles/week of running; 3 = 11 to 20 miles/week running; and 4 = >20 miles/week of running.
CRF was determined using a maximal treadmill exercise test and a modified Balke protocol as previously described. Treadmill speed was set initially at 88 m/min. The grade was 0% during the first minute, 2% during the second minute, and increased 1% each minute until 25 minutes. After 25 minutes, the grade did not change and speed was increased 5.4 m/min each minute until test termination. Time achieved on a treadmill has previously been shown to be highly correlated with measured maximal oxygen uptake in men and women. For the present study, CRF was defined (1) as a continuous variable expressed as total time achieved on a treadmill and (2) categorized into quintiles of CRF. METs (1 MET = 3.5 ml oxygen uptake per kilogram of body weight per minute) were estimated from the final treadmill speed and grade.
To examine the distribution of baseline characteristics according to categories of CRF, chi-square tests to compare proportions and analysis of variance to compare means were used. Pearson correlation coefficient was used to determine the correlation between CRF and BMI. Multiple linear regression models using stepwise selection (forward) were employed to determine factors related to CRF, defined as total time achieved on an exercise treadmill test. Initial variables entering the model were based on knowledge of previous factors related to CRF. The final multivariable model included age, gender, BMI, physical activity, smoking, glucose, high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterol, health status, and systolic blood pressure. Waist circumference, percent body fat, and lean body mass were substituted for BMI in exploratory models. Test for interaction between BMI and physical activity categories were assessed in age-adjusted models stratified by gender.
Results
The average age of the cohort was 47 ± 9.9 years and 66% were men. Women and men <30 years old achieved a mean CRF level of 10.9 ± 2.1 and 12.3 ± 2.4 METs, respectively. Of those 30 to 39 years old, the mean CRF value was 10.6 ± 1.9 METs for women and 12.2 ± 2.1 METs for men. MET levels were 9.9 ± 1.8 and 11.7 ± 2.1 for 40- to 49-year-old women and men and 9.0 ± 1.7 and 10.7 ± 1.9 for 50- to 59-year-old women and men, respectively. For subjects ≥60 years old, women and men achieved MET levels of 8.0 ± 1.5 and 9.6 ± 1.9, respectively. Converting METs to maximum oxygen consumption (1 MET = 3.5 ml oxygen uptake per kilogram of body weight per minute), CRF estimates were within 1 to 2 ml/kg/min of the population-based NHANES (1999 to 2004) sample for non-Hispanic white women and men (for comparable age groups studied).
The relation between clinical risk factors and CRF, categorized into quintiles and reported in METs, is presented in Table 1 . Age was inversely associated with fitness, with subjects in the lowest CRF quintile on average 10 years older than those in the highest CRF quintile. Male gender was more prevalent with increasing CRF quintile, with 88% of highly fit subjects being men (p <0.001). BMI was strongly related to CRF level achieved, such that women and men in the highest CRF quintile had a mean BMI of 21.0 ± 2.1 and 25.3 ± 2.5 kg/m 2 , respectively, whereas women and men in the lowest CRF quintile had a mean BMI of 27.5 ± 5.5 and 32.5 ± 5.7 kg/m 2 , respectively (p <0.001 for men and women). Forty-two percent of subjects in the lowest CRF quintile reported not taking part in any form of organized physical activity compared to 6.9% of highly fit subjects. This resulted in a larger percentage of highly fit subjects meeting the guidelines for physical activity compared to subjects in the lowest quintile (p <0.001). Other risk factors such as smoking, hypertension, diabetes, and LDL cholesterol levels were negatively associated with CRF category, whereas HDL cholesterol was positively associated with CRF level (p <0.001 for all). Women and men in the highest CRF quintile on average had an HDL cholesterol level of 71.5 ± 15.6 and 52.8 ± 12.4 mg/dl, respectively.
| Fitness Quintiles | |||||
|---|---|---|---|---|---|
| Quintile 1 (n = 3,968) | Quintile 2 (n = 3,995) | Quintile 3 (n = 4,174) | Quintile 4 (n = 4,059) | Quintile 5 (n = 4,133) | |
| METs | <9.0 | 9.0–10.2 | 10.3–11.2 | 11.3–12.5 | ≥12.6 |
| Age (years) | 53.2 (10.6) | 49.5 (9.5) | 46.8 (8.9) | 45.2 (8.7) | 42.7 (7.9) |
| Men | 1,564 (39.4%) | 2,174 (54.4%) | 2,900 (69.5%) | 3,197 (78.7%) | 3,624 (87.7%) |
| Body mass index (kg/m 2 ) | |||||
| Women | 27.5 ± 5.5 | 23.9 ± 3.4 | 22.8 ± 3.0 | 21.7 ± 2.5 | 21.0 ± 2.1 |
| Men | 32.5 ± 5.7 | 29.7 ± 4.0 | 28.1 ± 3.2 | 26.8 ± 2.9 | 25.3 ± 2.5 |
| Reports no organized physical activity | 1,653 (41.7%) | 1,098 (27.5%) | 919 (22.0%) | 572 (14.1%) | 283 (6.9%) |
| Meets guideline for physical activity ⁎ | 1,834 (46.2%) | 2,452 (61.4%) | 2,935 (70.3%) | 3,246 (80.0%) | 3,747 (90.7%) |
| Current smoker | 484 (12.8%) | 458 (12.1%) | 546 (13.7%) | 508 (13.2%) | 388 (10.0%) |
| Self-reported hypertension | 1,085 (29.1%) | 761 (20.1%) | 616 (15.6%) | 475 (12.3%) | 298 (7.6%) |
| Self-reported diabetes | 180 (4.9%) | 76 (2.0%) | 51 (1.3%) | 25 (0.7%) | 15 (0.4%) |
| Systolic blood pressure (mm Hg) | 124.8 ± 16.6 | 121.2 ± 14.9 | 120.2 ± 14.0 | 120.1 ± 13.5 | 119.1 ± 12.2 |
| Diastolic blood pressure (mm Hg) | 82.7 ± 10.2 | 81.8 ± 10.1 | 81.6 ± 9.7 | 81.4 ± 9.7 | 79.8 ± 8.9 |
| Glucose (mg/dl) | 101.0 ± 22.9 | 96.9 ± 15.7 | 95.8 ± 12.6 | 95.0 ± 11.2 | 93.6 ± 9.7 |
| Low-density lipoprotein cholesterol (mg/dl) | 120.1 ± 34.0 | 118.5 ± 33.1 | 119.1 ± 33.9 | 118.9 ± 32.0 | 114.3 ± 30.4 |
| High-density lipoprotein cholesterol (mg/dl) | |||||
| Women | 61.9 ± 16.8 | 66.9 ± 16.4 | 67.1 ± 15.3 | 69.3 ± 15.6 | 71.5 ± 15.2 |
| Men | 43.8 ± 11.7 | 44.4 ± 10.4 | 46.3 ± 11.3 | 48.9 ± 11.6 | 52.8 ± 12.4 |
⁎ American College of Sports Medicine and American Heart Association guidelines for physical activity. Moderate intensity activity = ≥5 times for 30 min/week or ≥450 METs/min/week.
Next, we determined the contribution of modifiable and nonmodifiable clinical risk factors on CRF. Age and gender explained 14% and 13% of the variance in CRF (p <0.001), respectively, with younger age and male gender associated with greater CRF achieved. BMI was a key modifiable risk factor for CRF, such that with each increment of kilogram per meter squared, there was a decrease of 30 seconds on the treadmill (p <0.001). This is illustrated in Figure 1 demonstrating a negative correlation between treadmill time and BMI for women (r = −0.53, p = 0.001; Figure 1 ) and men (r = −0.56, p <0.001; Figure 1 ). In addition to BMI, physical activity was strongly associated with CRF (p <0.001). Overall, 4 clinical factors, age, gender, BMI, and physical activity, contributed significantly to CRF explaining 56% of the variability in the model. Addition of other clinical factors such as smoking, glucose, HDL and LDL cholesterol, health status, and systolic blood pressure were associated with CRF (p <0.05) in the multivariable model but improved R 2 by only 2%. Of note, substituting waist circumference for BMI in the final multivariable model did not change the model characteristics (R 2 = 0.58).
