Physical activity can improve several metabolic risk factors associated with cardiovascular disease (CVD) and is associated with a lower risk of CVD mortality. We sought to evaluate the extent to which metabolic risk factors mediate the association between physical activity and CVD mortality and whether physical activity provides protective effects against CVD mortality in healthy adults and those with metabolic risk factors. A sample of 10,261 adults from the Third National Health and Nutrition Examination Survey with public-access mortality data linkage (follow-up 13.4 ± 3.9 years) was used. Physical activity was assessed by questionnaire and classified into inactive, light, and moderate/vigorous activity categories. Metabolic risk factors (dyslipidemia, type 2 diabetes mellitus, obesity, hypertension, inflammation, and insulin resistance) were categorized using clinical thresholds. After adjusting for basic confounders, engaging in light or moderate/vigorous physical activity was associated with a lower risk of CVD mortality (p <0.05). Adjustment for each risk-factor set only slightly attenuated this relation. When all risk-factor sets were added to the model simultaneously, light (hazard ratio 0.72, 0.62 to 0.84) and moderate/vigorous (hazard ratio 0.72, 0.62 to 0.85) activity remained at lower risk of CVD mortality. In addition, physical activity provided protective effects for CVD mortality in healthy subjects and those with metabolic risk factors in isolation or in clusters. In conclusion, physical activity was associated with a lower risk of CVD mortality independent of traditional and inflammatory risk factors. Taken together these results suggest that physical activity may protect against CVD mortality regardless of the presence of metabolic risk factors.
Physical activity is a therapeutic option often prescribed to patients as a means of improving metabolic risk factors, which include high-density lipoprotein (HDL) cholesterol, excess adiposity, high blood pressure, and glucose metabolism and control. Accordingly, physical activity is also associated with a decreased risk of cardiovascular disease (CVD) mortality, which may in part be influenced by these improvements in metabolic risk factors. The purpose of the present analysis is to examine the extent to which metabolic risk factors mediate the association between leisure time physical activity and CVD mortality. This will further be examined by investigating the protective effects of physical activity intensity on CVD mortality risk in metabolically healthy adults and in those with individual or clusters of metabolic risk factors.
Methods
The study sample was obtained from the Third National Health and Nutrition Examination Survey (NHANES III), which is a nationally representative cross-sectional survey of the United States conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention. Data were collected from 1988 through 1994 in 33,994 subjects ≥2 months old using a multistage stratified probability cluster design. Complete details of the study design and procedures are reported elsewhere. Data from the public-access mortality linkage file were used with follow-up through December 31, 2006. Mortality status was determined primarily through probabilistic matching with the National Death Index death certificate records. The National Death Index has a high sensitivity and specificity generally exceeding 95%. In addition, a calibration study using the NHANES I Follow-Up Survey reported that the National Death Index correctly classified the vital status of 98.5% of participants in the cohort. Underlying cause of death was based on the International Statistical Classification of Diseases, Ninth Revision for deaths occurring from 1988 through 1998 and the International Statistical Classification of Diseases, 10th Revision for deaths occurring from 1999 through 2006. Death was classified for cardiovascular causes from codes 390 to 398, 401 to 404, 410 to 414, 420 to 438, 440 to 448, and 451 to 459 from the ninth revision and codes 100 to 109, 111, 113, 125 to 151, 160 to 178, and 180 to 199 from the tenth revision. All study participants gave their informed written consent before participation in the examination and the study protocol was approved by the National Center for Health Statistics. Subjects were excluded if they were pregnant (n = 196), were <20 or >90 years of age (n = 15,190), had a body mass index (BMI) <18.5 kg/m 2 (n = 1,123), or had missing data for age, mortality follow-up, metabolic risk factors, or physical activity (n = 7,224). This left a final sample of 10,261 men and women.
Participants completed a questionnaire on whether they participated in leisure-time physical activity and, if so, the frequency of their participation in the following activities: walking, jogging, bicycling or bicycling on an exercise bicycle, swimming, aerobics or aerobic dancing, other dancing, calisthenics or exercises, gardening or yard work, and weight lifting. Subjects could also list up to 4 additional activities. Information on duration of physical activity was not assessed. A metabolic equivalent (MET) value was assigned to each activity based on the standardized coding scheme by Ainsworth et al. Physical activity was classified into inactive, light, moderate, and vigorous using standardized methods consistent with American College of Sports Medicine guidelines. Vigorously active was defined as engaging ≥3 times/week in activities with a MET value of 6 for subjects >60 years old and a MET value of 7 for subjects <60 years old. Moderately active was engaging ≥5 times/week in activities of which ≤2 could be considered vigorous. Lightly active was engaging in activity that was considered neither moderate nor vigorous. Inactive was defined as engaging in no leisure-time physical activity.
Age, gender, income (<$10,000, $10,000 to $29,999, $30,000 to $39,999, ≥$40,000), ethnicity (non-Hispanic White or non-White), smoking status (current smoker, former smoker, or never smoked), alcohol intake (≥3 or <3 drinks/day), dietary fat intake (>30% or ≤30%), medications (lipid, blood pressure, diabetes), and a self-reported physician diagnosis of hypertension, heart attack, stroke, and congestive heart failure were assessed by questionnaire. Participants were classified as having CVD if a physician previously told them they had a heart attack, stroke, or congestive heart failure. BMI was calculated using measured height and weight.
Blood was drawn from the antecubital vein after ≥6-hour fast for assessment of triglycerides, HDL cholesterol, total cholesterol, plasma glucose, serum insulin, glycated hemoglobin, and C-reactive protein. Blood pressure was assessed by a physician or trained interviewer after the participant had been seated quietly for 5 minutes. Homeostasis model assessment was used to evaluate insulin resistance.
Metabolic risk factors were grouped into risk-factor sets based on their pathophysiologic effects. Dyslipidemia was defined as triglycerides ≥2.06 mmol/L, HDL cholesterol <1.04 mmol/L for men and <1.29 mmol/L for women, or total cholesterol ≥6.0 mmol/L. Hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or a physician diagnosis of hypertension. Inflammation was defined as C-reactive protein ≥0.3 mg/dl and insulin resistance was defined as homeostasis model assessment ≥2.5. Type 2 diabetes mellitus was defined as glycated hemoglobin ≥6.5% or a fasting plasma glucose ≥7.0 mmol/L and obesity was defined as a BMI ≥30 kg/m 2 .
Baseline characteristics were reported stratified by physical activity and were weighted to be representative of the Unites States population. Because of the small number of deaths in the vigorous category, moderate and vigorous activity groups were collapsed. Differences in participant characteristics among physical activity levels were assessed using 1-way analysis of variance and chi-square tests. Hazard ratios (HRs) for CVD mortality risk were estimated by Cox proportional hazard regressions. No violation of the proportional hazards assumption was observed. No significant gender or age interactions were found; therefore analyses were conducted with men and women of all ages together. The basic multivariate model was adjusted for age, gender, income, ethnicity, smoking status, alcohol intake, high dietary fat intake, and pre-existing CVD. To examine the extent to which metabolic risk factors mediate the association between CVD mortality and physical activity, each risk-factor set (dyslipidemia, type 2 diabetes mellitus, obesity, hypertension, inflammation, and insulin resistance) was added to the basic model individually and then simultaneously with inactive as the referent. Further analysis examined CVD mortality risk in subjects stratified by physical activity and number of metabolic risk factors present. Analyses were performed using SAS 9.2 (SAS Institute, Cary, North Carolina).
Results
Baseline characteristics stratified by physical activity group are presented in Table 1 . Most of the population reported engaging in light (42.1%) or moderate/vigorous (35.7%) activity. Active subjects were younger, were more likely to be of a high socioeconomic status, be nonsmokers, and have a more favorable metabolic profile than inactive subjects. During follow-up of 13.4 ± 3.9 years there were 2,433 deaths, 1,095 of which were due to cardiovascular causes. Unadjusted survival curves for each physical activity category are shown in Figure 1 . Unadjusted mortality rates per 1,000 person-years were 13.2, 6.2, and 7.6 in the inactive, light, and moderate/vigorous physical activity groups, respectively.
| Variable | Physical Activity Intensity | ||
|---|---|---|---|
| Inactive (n = 2,077) | Light (n = 4,431) | Moderate/Vigorous (n = 3,753) | |
| Age (years) | 49.9 ± 0.6 | 43.7 ± 0.5 ¶ | 45.2 ± 0.6 ¶ |
| Men | 34.0% | 48.2% ¶ | 54.0% ¶ |
| White | 64.6% | 79.5% ¶ | 80.2% ¶ |
| Black | 14.3% | 9.2% ¶ | 9.3% ¶ |
| Hispanic | 8.4% | 4.7% ¶ | 3.8% ¶ |
| Other ethnicity | 12.7% | 6.6% ¶ | 6.7% ¶ |
| Income | |||
| <$10,000 | 25.1% | 13.9% ¶ | 13.6% ¶ |
| $10,000–$29,000 | 43.1% | 34.5% ¶ | 34.2% ¶ |
| $30,000–$39,000 | 12.7% | 16.0% | 14.1% |
| ≥$40,000 | 19.2% | 35.6% ¶ | 38.1% ¶ |
| Smoking status | |||
| Never | 44.7% | 45.2% | 45.2% |
| Former | 23.5% | 25.4% | 30.8% ¶ |
| Current | 31.8% | 29.4% | 24.0% ¶ |
| Pre-existing cardiovascular disease | 10.3% | 4.0% ¶ | 5.4% ¶ |
| Hypertension | 31.9% | 22.6% ¶ | 21.3% ¶ |
| Type 2 diabetes mellitus | 7.6% | 5.5% ¶ | 4.1% ¶ |
| Obesity ⁎ | 28.2% | 24.5% | 18.0% ¶ |
| Hypercholesterolemia † | 24.6% | 19.2% ¶ | 18.0% ¶ |
| Low high-density lipoprotein cholesterol ‡ | 40.4% | 39.4% | 33.3% ¶ |
| Hypertriglyceridemia § | 18.5% | 18.2% | 16.0% |
| Homeostasis model assessment–estimated insulin resistance ‖ | 40.0% | 33.8% ¶ | 26.2% ¶ |
| C-reactive protein ≥3 mg/L | 37.9% | 29.6% ¶ | 23.4% ¶ |
⁎ Body mass index ≥30 kg/m 2 .
† Total cholesterol ≥6.0 mmol/L.
‡ High-density lipoprotein cholesterol <1.04 mmol/L for men and <1.29 mmol/L for women.
‖ Homeostasis model assessment ≥2.5.
Dyslipidemia (HR 1.16, 1.03 to 1.32), type 2 diabetes mellitus (HR 1.66, 1.42 to 1.94), hypertension (HR 1.39, 1.23 to 1.57), inflammation (HR 1.30, 1.15 to 1.47), and insulin resistance (HR 1.22, 1.09 to 1.38) were significantly associated with CVD mortality after adjustment for age, gender, income, ethnicity, smoking status, alcohol intake, high dietary fat intake, and pre-existing CVD. Surprisingly, obesity alone was not found to be significantly associated with CVD mortality (HR 1.12, 0.97 to 1.30) and adjustment for physical activity did not substantially attenuate the results (data not shown). The more metabolic risk factors that were present, the higher the risk for CVD mortality; subjects with ≥3 metabolic risk factors had the highest CVD mortality risk (HR 1.70, 1.35 to 2.15) compared to subjects free of metabolic risk factors. After adjustment for all other CVD metabolic risk factors and physical activity, type 2 diabetes mellitus (HR 1.53, 1.30 to 1.81), hypertension (HR 1.33, 1.17 to 1.50), and inflammation (HR 1.22, 1.08 to 1.38) remained significantly associated with CVD mortality.
As presented in Table 2 engaging in light (HR 0.72, 0.62 to 0.84) or moderate/vigorous (HR 0.72, 0.61 to 0.84) activity was similarly associated with a lower risk for CVD mortality after adjustment for covariates (basic model). Adjustment for each risk-factor set resulted in little to no attenuation of the inverse relation between CVD mortality and physical activity. When all risk-factor sets were added simultaneously to the basic model, similar results were observed and no further attenuation of the HR was observed.
