Abstract
Objectives
To compare the cross-sectional associations between cardiorespiratory fitness and echocardiographic measures of cardiac structure and function.
Background
Cardiorespiratory fitness (fitness) is inversely associated with heart failure risk. However, the mechanism through which fitness lowers heart failure risk is not fully understood.
Methods
We included 1,678 men and 1,247 women from the Cooper Center Longitudinal Study who received an echocardiogram from 1999—2011. Fitness was measured by Balke protocol and categorized into age-specific quartiles with quartile 1 representing low fitness. Cross sectional associations between fitness (in METs) and relative wall thickness (RWT), left ventricular end-diastolic diameter indexed to body surface area (LVEDD/BSA), left atrial volume indexed to body surface area (LA Vol/BSA), left ventricular systolic function (LVEF), and E/e′ ratio were determined using multivariable linear regression analysis.
Results
Higher levels of mid life fitness (METs) was associated with larger indexed LA volume (Men: β = 0.769, p <0.0001; Women: β = 0.879, p value = <0.0001) and indexed LVEDD (Men: β = 0.231, p <0.001; Women: β = 0.264, p <0.0001). Similarly, a higher level of fitness was associated with a smaller RWT (Men: β = -0.002; p = 0.04; Women: β = -0.005, p <0.0001) and E/e′ ratio (Men: β = -0.11; p = 0.003; Women: β = -0.13, p value = 0.01). However, there was no association between low fitness and LVEF (P=NS).
Conclusions
Low fitness is associated with a higher prevalence of concentric remodeling and diastolic dysfunction, suggesting that exercise may lower heart failure risk through its effect on favorable cardiac remodeling and improved diastolic function.
Keywords: diastolic dysfunction, echocardiography, exercise, heart failure, remodeling
Introduction
Higher levels of self-reported physical activity and measured fitness are associated with a lower risk for heart failure that is independent of established heart failure risk factors such as obesity, diabetes, and hypertension(1-4). While the potential mechanisms through which exercise might lower heart failure risk are not completely understood, multiple lines of evidence suggest that higher levels of exercise might have a direct effect on cardiac structure and function. In particular, individuals who report high levels of exercise across the lifespan have more compliant left ventricles than sedentary, age-matched controls (5,6). These findings suggest the hypothesis that higher levels of exercise may lower the risk for heart failure with preserved ejection fraction (HFPEF). Given the failure of numerous therapies for the treatment and prevention of HFPEF, this would represent an important observation that could have significant public health implications (7).
The associations between intense physical activity and fitness and physiologic cardiac remodeling and improved early diastolic filling are well established in elite athletes (8-10) but the effects of exercise on cardiac structure and function within the continuum of normal fitness levels are not known. Echocardiography represents a valuable potential intermediate phenotype that can provide important insights into cardiac structure and both systolic and diastolic function. The presence of pathologic ventricular remodeling patterns (11-13) have been identified as important intermediates in the pathway to heart failure. Ventricular remodeling is associated with increased volume and altered chamber geometry that develops in response to a myocardial injury and increased wall stress. Three patterns of LV remodeling have been indentified based on the measures of left ventricular (LV) mass index and relative wall thickness (RWT): concentric remodeling (normal LV mass index and increased RWT), eccentric hypertrophy (increased LV mass index and normal RWT), and concentric hypertrophy (increased LV mass index and increased RWT) (14). Similarly, subclinical systolic and diastolic dysfunction has been shown to be an important determinant of adverse cardiovascular outcomes (15-18). Specifically, subclinical systolic dysfunction carries increased risk of heart failure with reduced ejection fraction and asymptomatic diastolic dysfunction increases the risk of future HFPEF(15-18).
The purpose of this study was to characterize the association between measured cardiorespiratory fitness and cardiac structure and function in the Cooper Center Longitudinal Study. We hypothesized that higher fitness levels would be associated with a lower prevalence of diastolic dysfunction and a lower prevalence of concentric remodeling/hypertrophy. We further hypothesize that there would be no association between fitness and systolic function.
Methods
Study participants
The Cooper Center Longitudinal Study (CCLS) is an ongoing study derived from patients at the Cooper Clinic, a preventive health clinic in Dallas, TX, and has previously been well described(19,20). All participants are either self-referred to the clinic or are referred by their employer or personal physician. They are predominantly Caucasian and from the middle to upper socio-economic strata. For the present study, we included patients from the CCLS who received both a clinically indicated echocardiogram at the clinic between 1999 and 2011 and a standardized medical examination by a physician including a maximal treadmill exercise test. Echocardiograms were performed for a broad range of indications (Supplemental Table). This study was approved by both the Cooper Institute and the UT Southwestern Institutional Review Boards.
After excluding 14 participants with severe valvular disorders, we included 1,678 men and 1,247 women in the final study sample who received an echocardiogram within three months of their exam and measured fitness at the clinic. There was a low prevalence of low EF and abnormal stress echo in the study cohort and those subjects were included in the final analysis. Finally, because tissue Doppler was incorporated into the standard echocardiographic examination at the Cooper Clinic in 2003, 42% of the participants were missing tissue Doppler data. Therefore, we included 1,235 participants for the tissue Doppler analyses.
Data collection
Details of the clinical examination for CCLS participants have been reported previously (19,20). Medical history of diabetes, hypertension, coronary artery disease (CAD), smoking history, fasting blood glucose, blood pressure, and body mass index (calculated from weight and height, kilogram/meter2) were collected during the physical examination. Medications were extracted from the medical record. Hypertension was defined as either a systolic blood pressure >140, self reported hypertension, or use of antihypertensive drug. Diabetes was defined as the presence of a fasting blood sugar ≥126, self-reported diabetes, or use of anti-hyperglycemic drug.
Echocardiographic data
All echocardiograms were performed using a GE Vivid 7, and were read by a staff cardiologist at the clinic at the time the echocardiogram was done. The following variables were gathered from the resting echocardiograms, including the indication for the echocardiogram, left ventricular end diastolic diameter (LVEDD), left ventricular end systolic diameter (LVESD), posterior wall thickness (PWT), septal wall thickness (SWT), left atrial diameter, left ventricular ejection fraction (LVEF), valvular stenosis/regurgitation, mitral inflow Doppler (E wave, A wave, and deceleration time), and tissue Doppler (e′) of the lateral mitral annulus.
The following echocardiographic variables were defined in accordance with standard definitions: left ventricular ejection fraction (LVEF): (LV end diastolic volume − LV end systolic volume) ÷LV end diastolic volume, where both the LV end diastolic volume and LV end systolic volume were estimated using the modified Simpson's rule (biplane method of disks); LV mass (indexed to body surface area) [0.8 × {1.04[(LVEDD + PWT + SWT)3 − (LVEDD)3]} + 0.6 g]; relative wall thickness (RWT) [2 X PWT/LVEDD]. LA volume was determined by the biplane area-length method. LV filling pressures were estimated by calculating the ratio of the LV early diastolic filling wave (mitral inflow Doppler E wave velocity) to the early diastolic velocity of the lateral mitral valve annulus (lateral e′ wave velocity). An E/e′ ratio greater than 10 was used as a cut-point for diastolic dysfunction of indeterminate grade as this indicates elevated left ventricular filling pressure(16,21-23).
Remodeling patterns
Four LV remodeling pattern groups were created using standard definitions: Normal geometry was defined as a left ventricular mass index (LVMI) (gm/m2) ≤ 95 for women or ≤ 115 for men and a RWT ≤ 0.42; concentric remodeling, a LVMI ≤ 95 for women or ≤ 115 for men and a RWT > 0.42; eccentric hypertrophy, a LVMI > 95 for women or >115 for men and a RWT ≤ 0.42; and concentric hypertrophy, a LVMI > 95 for women or > 115 for men and a RWT > 0.42.
Exercise testing
As reported previously (19,20), fitness was measured in the CCLS cohort by a maximal treadmill exercise test using a Balke protocol. In this protocol, treadmill speed is set initially at 88 m/min. In the first minute, the grade is set at 0%, followed by 2% in the second minute, and an increase of 1% for every minute thereafter. After 25 minutes, the grade remains unchanged but the speed is increased by 5.4 m/min for each additional minute until the test is terminated. Participants were encouraged not to hold onto the railing and were given encouragement to exert maximal effort. The test was terminated by volitional exhaustion reported by the participant or by the physician for medical reasons. As described previously, the test time using this protocol correlates highly with directly measured maximal oxygen uptake (r = 0.92)(24).
Statistical analyses
In accordance with standard approaches to the analysis of fitness data, treadmill times were categorized into quartiles using age- and sex-specific thresholds of treadmill performance. This allows each participant to be categorized into one of four mutually exclusive fitness categories that is independent of age and sex, with quartile 1 as the lowest fit group and quartile 4 as the highest fit. In addition, treadmill times from the Balke protocol can also be used to estimate continuous measures of fitness [metabolic equivalents (MET's)](24).
Baseline characteristics and echocardiographic parameters collected were stratified by gender and then compared across quartiles of fitness using the Jonckheere-Terpstra trend test for all continuous variables and the Cochran-Armitage trend test for categorical variables. The association between fitness and echocardiographic parameters [indexed LA volume(LAVol/BSA), indexed LVEDD( LVEDD/BSA), RWT, EF, E/e′ ratio] were estimated using linear regression with fitness (METs) entered as a continuous independent variable in both age-adjusted and multivariable models adjusted for age, BMI, hypertension, and diabetes. All statistical analyses were performed using SAS for Windows (release 9.2; SAS Institute, Inc. Cary, NC).
Results
The baseline characteristics of the study cohort are shown in Table 1, demonstrating a low prevalence of prior coronary artery disease and a low prevalence of traditional risk factors for structural heart disease. As expected, higher fitness was associated with a lower BMI, lower blood pressure, and a lower prevalence of diabetes and hypertension (Table 1).
Table 1. Baseline characteristics among men and women by fitness quartiles.
| A. Men | |||||
|---|---|---|---|---|---|
| Q1: low fit N=425 | Quartile 2 N=424 | Quartile 3 N=387 | Q4: high fit N=452 | P-value | |
| Age | 56.0+12.0 | 55.8+12.4 | 56.0+11.8 | 54.9+10.6 | 0.49 |
| Body mass index | 30.1+5.1 | 27.3+3.2 | 26.3+2.6 | 25.0+2.4 | <0.0001 |
| Systolic blood pressure | 129+16 | 127+16 | 128+15 | 125+15 | 0.002 |
| Diastolic blood pressure | 84+11 | 83+10 | 82+10 | 81+9 | <0.001 |
| Hypertension | 280 (69%) | 217 (52%) | 189 (50%) | 170 (38%) | <0.001 |
| Diabetes | 32 (8%) | 15 (4%) | 4 (1%) | 3 (1%) | <0.001 |
| Coronary artery disease | 35 (8%) | 29 (7%) | 32 (8%) | 24 (5%) | 0.14 |
| Current Smoker | 54 (15%) | 36 (10%) | 32 (10%) | 24 (6%) | 0.0001 |
| B. Women | |||||
|---|---|---|---|---|---|
| Q1: low fit N=315 | Quartile 2 N=284 | Quartile 3 N=358 | Q4: high fit N=290 | P-value | |
| Age | 53.4+ 12 | 52.1+12.3 | 52.8+10.7 | 51.3+11.1 | 0.15 |
| Body mass index | 27.7+6 | 24.6+ 4 | 23.2+2.8 | 21.9+3 | <0.0001 |
| Systolic blood pressure | 123+18 | 119+18 | 118+16 | 116+16 | <0.0001 |
| Diastolic blood pressure | 80+10 | 78+10 | 78+9 | 77+9 | <0.0001 |
| Hypertension | 148 (47%) | 88(32%) | 126 (36%) | 61 (22%) | <0.0001 |
| Diabetes | 25 (8%) | 11 (4%) | 6 (2%) | 2 (1%) | <0.0001 |
| Coronary artery disease | 2 (0.6%) | 4 (1%) | 6 (2%) | 4 (1%) | 0.37 |
| Current Smoker | 16 (6%) | 4 (1%) | 11 (4%) | 5 (2%) | 0.051 |
Hypertension: BP>140/90, self-report of hypertension or use of anti-hypertensive medication. Diabetes: fasting blood glucose >126, self report of diabetes or use of anti-diabetic drug. All data reported as means ± SD or N (%).P-value < 0.05 represents statistical significance.
Heart size, concentricity and function across the fitness quartiles
In men, compared to high fitness (quartile 4), low fitness (quartile 1) was associated with smaller heart size (LA Vol/BSA: 25.4 mm vs. 22.7 mm, P<0.0001 and LVEDD: 48.5 mm vs. 49.9 mm, P <0.0001) lower LV mass (LVMI: 84.3 g/m2 vs. 91.9 g/m2, P< 0.0001), and a higher RWT (RWT: 0.42 vs. 0.39; P < 0.0001). Lower fitness was also associated with a higher E/e′ ratio (E/e′: 6.6 vs. 5.3, P < 0.0001). In contrast, there was no association between fitness and LVEF (LVEF: 60.1 vs. 60.9; P=0.32). Overall, a similar pattern of results was observed in women except there was no apparent association between fitness and LVMI in women (68.2 g/m2 vs. 68.9 g/m2, P = 0.688) (Table 2).
Table 2. Echocardiographic characteristics by fitness quartile.
| A. Men | |||||
|---|---|---|---|---|---|
| Q1: low fit N=425 | Quartile 2 N=424 | Quartile 3 N=387 | Q4: high fit N=452 | P-value | |
| MET | 8.5 ± 1.4 | 10.4 ± 1.2 | 11.6 ± 1.2 | 13.8 ± 1.8 | <.0001 |
| LA Vol/BSA | 22.8 ± 7.2 | 22.5 ± 6.5 | 23.9 ± 6.4 | 25.4 ± 7.1 | <0.001 |
| LV Geometry | |||||
| LV mass/BSA (g/m2) | 84.3 ± 22.0 | 84.4 ± 19.9 | 87.3 ± 18.2 | 91.9 ± 43.2 | <.0001 |
| LVEDD (mm) | 48.5 ± 5.2 | 48.6 ± 5.2 | 48.7 ± 4.6 | 49.9 ± 4.8 | <.0001 |
| PWT (mm) | 10.1 ± 1.8 | 9.8 ± 1.4 | 9.8 ± 1.5 | 9.6 ± 1.3 | <.0001 |
| SWT (mm) | 10.5 ± 1.9 | 10.3 ± 1.9 | 10.4 ± 1.7 | 10.3 ± 3.5 | 0.045 |
| RWT | 0.42 ± 0.09 | 0.41 ± 0.08 | 0.41 ± 0.09 | 0.39 ± 0.08 | <.0001 |
| Systolic function | |||||
| LVEF | 60± 6 | 61 ± 5 | 61 ± 4 | 61± 4 | 0.319 |
| Diastolic function | |||||
| Mitral E/A | 1.15 ± 0.42 | 1.19 ± 0.43 | 1.24 ± 0.44 | 1.33 ± 0.43 | <.0001 |
| Lateral e′ (cm/s) | 0.11 ± 0.03 | 0.11 ± 0.03 | 0.11 ± 0.03 | 0.12 ± 0.03 | <.0001 |
| E/e′ | 6.6 ± 2.8 | 5.9 ± 1.7 | 5.9 ± 1.8 | 5.3 ± 1.5 | <.0001 |
| Decel Time (ms) | 204 ± 45 | 208 ± 43 | 209 ± 38 | 209 ± 47 | 0.084 |
| B. Women | |||||
|---|---|---|---|---|---|
| Q1: low fit N=315 | Quartile 2 N=284 | Quartile 3 N=358 | Q4: high fit N=290 | P-value | |
| MET | 7.1 ± 1.1 | 8.7 ± 1.1 | 9.7 ± 1.1 | 11.7 ± 1.5 | <.0001 |
| LA Vol/BSA | 19.7 ± 5.3 | 19.0 ± 5.3 | 19.6 ± 5.5 | 22.0 ± 6.1 | 0.001 |
| LV Geometry | |||||
| LV mass/BSA (g/m2) | 68.2 ± 15.4 | 68.1 ± 15.5 | 66.4 ± 13.4 | 68.9 ± 14.4 | 0.688 |
| LVEDD (mm) | 43.7 ± 4.9 | 44.4 ± 4.1 | 44.2 ± 3.9 | 44.8 ± 4.2 | 0.005 |
| PWT (mm) | 8.6 ± 1.4 | 8.2 ± 1.6 | 8.0 ± 1.2 | 7.9 ± 1.2 | <.0001 |
| SWT (mm) | 9.0 ± 1.7 | 8.4 ± 1.5 | 8.2 ± 1.3 | 8.3 ± 1.3 | <.0001 |
| RWT | 0.4 ± 0.09 | 0.37 ± 0.09 | 0.37 ± 0.07 | 0.35 ± 0.07 | <.0001 |
| Systolic function | |||||
| LVEF | 62 ± 5 | 62 ± 5 | 62 ± 4 | 62 ± 4 | 0.326 |
| Diastolic function | |||||
| Mitral E/A | 1.24 ± 0.42 | 1.36 ± 0.51 | 1.43 ± 0.56 | 1.5 ± 0.54 | <.0001 |
| Lateral e′ (cm/s) | 0.11 ± 0.03 | 0.12 ± 0.03 | 0.12 ± 0.03 | 0.12 ± 0.03 | 0.0001 |
| E/e′ | 7.0 ± 2.5 | 6.7 ± 2.3 | 6.54 ± 2.1 | 6.0 ± 1.8 | 0.001 |
| Decel Time (ms) | 196 ± 42 | 191 ± 41 | 196 ± 37 | 195 ± 37.6 | 0.498 |
MET: maximal workload in METs achieved in a Balke protocol treadmill stress test; LA Vol/BSA: left atrial volume indexed to body surface area: LV mass/BSA: left ventricular mass indexed to body surface area; LVEDD: left ventricular end diastolic diameter; PWT: posterior wall thickness; SWT: septal wall thickness; RWT: relative wall thickness; LVEF: left ventricular ejection fraction; Mitral E/A: mitral peak velocity of early filling (E) to mitral peak velocity of late filling (A); Lateral e′: peak velocity of the lateral mitral annulus; E/e′: mitral peak Doppler E-wave to peak mitral annulus velocity ratio; Decel time: deceleration time. Means ± SD. P-value < 0.05 represents statistical significance.
Prevalence of remodeling patterns by fitness quartile
There was nearly twice the prevalence of concentric remodeling (CR) among the lowest quartile compared to the highest quartile of fitness (38% vs. 21%, Q1 vs. Q4, P<0.0001). In addition, despite a low overall prevalence of either concentric or eccentric hypertrophy in our cohort, lower fitness was associated with a numerically higher prevalence of concentric hypertrophy and a numerically lower prevalence of eccentric hypertrophy (Figure 1).
Figure 1.
Association between fitness and remodeling patterns (N=2,815). Concentric remodeling— LVMI ≤ 95 for women or ≤ 115 for men and a RWT > 0.42. Eccentric hypertrophy— LVMI > 95 for women or >115 for men and a RWT ≤ 0.42. Concentric hypertrophy— LVMI > 95 for women or > 115 for men and a RWT > 0.42. P-trend <0.0001 for concentric remodeling; P-trend=0.293 for eccentric hypertrophy; p-trend=0.071 for concentric hypertrophy
Prevalence of diastolic dysfunction
The overall prevalence of diastolic dysfunction was low (5%, 63/1,235), consistent with the healthy nature of the cohort. Nevertheless, lower fitness was associated with an increased prevalence of diastolic dysfunction (8.9% vs. 2.2%, Q1 vs. Q4, P <0.0001) (Figure 2).
Figure 2.
Association between fitness and diastolic dysfunction, (N=1235) where diastolic dysfunction defined as E/e′≥10. P-trend <0.0001.
The association of fitness with heart size, concentricity and diastolic function
After adjustment for age, BMI, hypertension, and diabetes, cardiorespiratory fitness significantly predicted indexed LAVol , indexed LVEDD, RWT and E/e′ ratio in both men and women. Higher fitness levels (in METs) were associated with larger indexed LA volume (Men: β = 0.769, p <0.0001; Women: β = 0.879, p value = <0.0001) and indexed LVEDD (Men: β = 0.231, p <0.001; Women: β = 0.264, p <0.0001). Similarly, a higher level of fitness was associated with a smaller RWT (Men: β = -0.002; p = 0.04; Women: β = -0.005, p <0.0001) and E/e′ ratio (Men: β = -0.111; p = 0.003; Women: β = -0.13, p value = 0.01) (Table 3). In additional sensitivity analyses performed excluding participants with a low EF or abnormal stress test a similar pattern of results were observed (data not shown).
Table 3. Multivariable analysis of LVEDD, RWT, and E/e′.
| A. Men | ||||||||
|---|---|---|---|---|---|---|---|---|
| LA Vol/BSA | L VEDD/BSA | RWT | E/e′ | |||||
| Beta 95% CI | P-Value | Beta 95% CI | P-Value | Beta 95% CI | P-Value | Beta 95% CI | P-Value | |
| Age (per year) | 0.239 (0.189 - 0.29) | <0.0001 | 0.018 (0.005 - 0.03) | 0.004 | 0.001 (0.001 - 0.001) | <0.0001 | 0.059 (0.045 - 0.073) | <0.001 |
| HTN (y/n) | 0.728 (-0.34 - 1.796) | 0.182 | -0.079 (0.335 - 0.178) | 0.549 | 0.019 (0.011 - 0.028) | <0.0001 | 0.385 (0.094 - 0.675) | 0.009 |
| BMI (per kg/m2) | 0.107 (-0.034 - 0.249) | 0.139 | -0.217 (-0.254 - -0.18) | <0.0001 | 0.002 (0.001 - 0.004) | 0.0001 | 0.04 (0.002 - 0.076) | 0.06 |
| Diabetes (y/n) | 0.385 (-2.282 - 3.052) | 0.777 | 0.439 (-0.255 - 1.133) | 0.215 | -0.005 (-0.028 - 0.017) | 0.646 | 0.66 (-0.088 - 1.407) | 0.08 |
| Fitness (per Met) | 0.769 (0.5 - 1.04) | <0.0001 | 0.231 (0.164 - 0.298) | <0.0001 | -0.002 (-0.004 - 0) | 0.042 | -0.111 (-0.184 - -0.038) | 0.003 |
| B. Women | ||||||||
|---|---|---|---|---|---|---|---|---|
| LA Vol/BSA | L VEDD/BSA | RWT | E/e′ | |||||
| Beta | P-Value | Beta | P-Value | Beta | P-Value | Beta | P-Value | |
| Age (per year) | 0.161 (0.116-0.205) | <0.0001 | -0.009 (-0.024 - 0.006) | 0.246 | 0.001 (0.001 - 0.002) | <0.0001 | 0.069 (0.054 - 0.084) | <0.0001 |
| HTN (y/n) | 0.415 (-0.574- 1.404) | 0.411 | -0.218 (- 0.536 - 0.1) | 0.179 | 0.024 (0.015 - 0.034) | <0.0001 | 0.646 (0.306 - 0.985) | 0.0002 |
| BMI(per kg/m2) | 0.274 (0.167-0.38) | <0.0001 | -0.242 (-0.279 - -0.206) | <0.0001 | 0.002 (0.001 - 0.003) | <0.0001 | 0.013 (-0.023 - 0.049) | 0.486 |
| Diabetes (y/n) | -1.185 (-3.414-1.045) | 0.298 | 0.213 (-0.555 - 0.981) | 0.586 | 0.013 (-0.01 - 0.036) | 0.255 | 1.123 (0.369 - 1.876) | 0.004 |
| Fitness (per met) | 0.879 (0.591-1.167) | <0.0001 | 0.264 (0.17 - 0.358) | <0.0001 | -0.005 (-0.007 - 0.002) | <0.0001 | -0.13 (-0.23 - -0.031) | 0.01 |
Hypertension: BP>140/90, self-report of hypertension or use of anti-hypertensive medication; Diabetes: fasting blood glucose >126, self report of diabetes or use of anti-diabetic drug ; MET: maximal workload in METs achieved in a Balke protocol treadmill stress test; LA Vol/BSA: left atrial volume indexed to body surface area; LVEDD: left ventricular end diastolic diameter; RWT: relative wall thickness; LVEF: left ventricular ejection fraction; E/e′: mitral peak Doppler E-wave to peak mitral annulus velocity ratio. P-value < 0.05 represents statistical significance.
Left atrial size and diastolic function
Among participants with low fitness, LA Vol/BSA positively correlated with increased E/e′ (Rho = 0.29, P < 0.0001), however, we observed no apparent association between LA Vol/BSA and E/e′ among participants with high fitness (Rho = 0.036, P = 0.55) (Figure 3). These data suggest that the association between left atrial volume and filling pressures differ according to fitness status and therefore, the observed association between higher fitness and higher left atrial volume is unrelated to diastolic function.
Figure 3.
Correlation between LA Vol/BSA and E/e′ according to fitness level. (A) Low fitness (quartile 1) (N= 730; r =0.29, P<0.0001); (B) high fitness (quartile 4) (N = 742; r =0.036, P=0.55).
Discussion
We observed two important findings in this study. First, in a low-risk cohort with a low prevalence of cardiovascular disease and traditional risk factors, we observed that low fitness was independently associated with smaller heart size and a pattern of concentric remodeling. Second, although we observed that higher fitness was associated with a lower prevalence of diastolic dysfunction, there was no apparent association between fitness and systolic function. Taken together, these data suggest that diastolic dysfunction might play a significant role in predisposing subjects with low mid-life fitness to a higher risk for HFPEF at a later age.
While self-reported physical activity has previously been associated with higher LV mass and larger LV end diastolic volume (25-28), there are few studies that have comprehensively addressed the association between measured fitness and echocardiographic measures of LV structure and function in middle-aged adults (29). To our knowledge, this is the first paper to examine the association between measured cardiorespiratory fitness and echocardiographic measures of both LV structure and function in a large cohort of healthy, asymptomatic, middle-aged adults not referred for exercise testing. In previous reports from the Mayo Clinic among patients referred for exercise stress echocardiography, higher fitness levels were observed to be inversely proportional to diastolic function. However, because of the higher prevalence of disease in a symptomatic population (49% of this cohort was referred for shortness of breath or chest pain), diastolic dysfunction was potentially a cause rather than an effect of lower fitness levels (30). In a smaller study from this same cohort exercise capacity was inversely associated with RWT and LV mass index. Lower exercise tolerance in these patients was also associated with a higher prevalence of pathologic remodeling phenotypes (31). We extend these findings by examining a subset of patients at the Cooper Clinic who underwent fitness testing for screening purposes and received echocardiographic testing for reasons largely unrelated to symptoms.
Prior studies have observed that physical activity and fitness are inversely associated with the risk of heart failure independent of established heart failure risk factors such as obesity, diabetes, and hypertension(1-4). However, the mechanisms through which exercise might lower heart failure risk remains poorly understood. Data from animal models suggest that exercise has a direct effect on both cardiac structure and function (32,33). For example, exercise training in hypertensive rats not only attenuates the development of heart failure without decreasing blood pressure, but also appears to result in reduced concentric remodeling of the left ventricle(32). Similarly, among athletes and individuals engaged in lifelong training, higher self-reported exercise levels have been associated with a larger heart size and improved early diastolic filling (5,8-10,34-36). The present study significantly extends the available literature, including a large number of healthy subjects with objectively measured fitness and echocardiographic examinations and observing a strong inverse association between fitness and concentric remodeling and diastolic dysfunction.
The benefits of exercise have also been observed among higher risk individuals with concentric hypertrophy and HFPEF. For example, 16 weeks of exercise training in forty-six men with poorly controlled hypertension was associated with a decrease in wall thickness, suggesting that regular physical exercise may be protective against concentric left ventricular hypertrophy (37). However, the effects of exercise training among patients with known HFPEF are not well understood. Exercise training improved heart failure symptoms (38,39) and diastolic function (38) among HFPEF patients in some studies but failed to show any favorable effects in others (40).
Measures of cardiac structure and function represent an important intermediate phenotype in the natural history of structural heart disease and symptomatic heart failure. Pathologic remodeling patterns are poor prognostic indicators for all cause mortality, cardiovascular events and heart failure (11-13), and asymptomatic systolic LV dysfunction represents a strong predictor of systolic heart failure in individuals without prior myocardial infarction or valvular disease (15). Similarly, recent data has shown one in four persons with moderate to severe asymptomatic diastolic dysfunction will progress to HFPEF (16-18). In the present study, we observed a consistent association between low fitness and both concentric remodeling and diastolic dysfunction in a cohort of healthy adults not referred for exercise testing. The effect of fitness on E/e′ is notable as it is a highly reproducible echocardiographic measure strongly associated with adverse cardiac events(41). Taken together, these findings suggest that higher fitness levels are associated with a lower prevalence of concentric remodeling and diastolic dysfunction which could play an important role in lowering risk for heart failure, specifically HFPEF at a later age.
It is well established that the LA enlarges in the setting of increased filling pressures(42) and therefore, LA size is a good marker of diastolic dysfunction (43-45) and is strongly associated with cardiovascular events (46-49). Thus, we observed an apparent conflict between LA size and diastolic function, with higher fitness associated with both a lower prevalence of diastolic dysfunction and a larger LA size. However, when we stratified our cohort according to fitness levels, we observed two distinct patterns of results. Among low fit individuals, diastolic dysfunction was associated with a larger LA size (Figure 3). In contrast, we observed no apparent association between diastolic dysfunction and LA size among high fit individuals. Our findings, as well as those reported by others(50), suggest that the mechanisms of LA dilation may be different among high and low fit individuals, with a larger LA in high fit individuals reflecting physiology unrelated to diastolic function (i.e. greater cardiac output).
Limitations
The present study is cross-sectional in nature and therefore we cannot completely exclude the possibility of reverse causation whereby subclinical diastolic dysfunction promoted exercise intolerance. Although possible, we believe that this is unlikely for several reasons. First, we observe consistent, inverse associations between fitness and subclinical measures of remodeling and diastolic dysfunction within the normal range of fitness levels in healthy adults, suggesting at a minimum that the presence of subclinical heart failure represents an unlikely explanation for the present findings. Second, the Cooper Clinic population represents a healthy cohort of subjects who have lower burden of traditional risk factors for heart failure and normal exercise capacity levels compared to the general population.(19,20) (51). The prevalence of exercise intolerance was very low in the study cohort with only 6% subjects with complaints of subjective shortness of breath. Further studies looking at effects of longitudinal changes in fitness on cardiac structure and function are needed to establish a causal association between low fitness and diastolic dysfunction.
A significant number of participants were referred for either resting or stress echocardiogram for an abnormal exercise ECG. However, only 8 participants out of 871 with stress echocardiograms had an abnormal stress echocardiogram. Furthermore in sensitivity analyses where these participants were excluded, we observed a similar pattern of results between fitness and measures of cardiac structure and function. Finally, in the present study the majority of subjects had a normal EF, and therefore, the lack of association with EF should be interpreted cautiously.
In conclusion, we observed that low fitness was associated with a smaller heart size, increased concentricity and diastolic dysfunction, suggesting that exercise might reduce the risk of heart failure through its favorable effects on cardiac remodeling and diastolic function.
Supplementary Material
Acknowledgments
We thank Dr Kenneth H. Cooper for establishing the Cooper Center Longitudinal Study, the Cooper Center staff for collecting clinical data, and the Cooper Institute for maintaining the database. We also thank Christina Podias (University of Texas Southwestern) for her role in the management of the echocardiographic database.
Sources of Funding: Dr Berry receives funding from the Dedman Family Scholar in Clinical Care endowment at University of Texas–Southwestern Medical Center; grant K23 HL092229 from the National Heart, Lung, and Blood Institute; and grant 13GRNT14560079 from the American Heart Association. Dr. Berry is a member of the Speaker's Bureau for Merck & Co.
Abbreviations
- RWT
relative wall thickness
- LVEDD
left ventricular diastolic diameter
- LVEF
left ventricular ejection fraction
- CVD
cardiovascular disease
- HFPEF
heart failure with preserved ejection fraction
- CCLS
Cooper Center Longitudinal Study
- CAD
coronary artery disease
- LVESD
left ventricular end systolic diameter
- PWT
posterior wall thickness
- SWT
septal wall thickness
- LVMI
left ventricular mass index
- LA Vol/BSA
left ventricular volume indexed to body surface area
- BMI
body mass index
- ECG
electrocardiogram
Footnotes
Disclosures: The other authors report no conflicts and no associations with industry.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.He J, Ogden LG, Bazzano LA, Vupputuri S, Loria C, Whelton PK. Risk factors for congestive heart failure in US men and women: NHANES I epidemiologic follow-up study. Archives of internal medicine. 2001;161:996–1002. doi: 10.1001/archinte.161.7.996. [DOI] [PubMed] [Google Scholar]
- 2.Hu G, Jousilahti P, Antikainen R, Katzmarzyk PT, Tuomilehto J. Joint effects of physical activity, body mass index, waist circumference, and waist-to-hip ratio on the risk of heart failure. Circulation. 2010;121:237–44. doi: 10.1161/CIRCULATIONAHA.109.887893. [DOI] [PubMed] [Google Scholar]
- 3.Kenchaiah S, Evans JC, Levy D, et al. Obesity and the Risk of Heart Failure. New England Journal of Medicine. 2002;347:305–313. doi: 10.1056/NEJMoa020245. [DOI] [PubMed] [Google Scholar]
- 4.Berry JD, Pandey A, Gao A, et al. Physical Fitness and Risk for Heart Failure and Coronary Artery Disease. Circ Heart Fail. 2013 doi: 10.1161/CIRCHEARTFAILURE.112.000054. In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Prasad A, Popovic ZB, Arbab-Zadeh A, et al. The Effects of Aging and Physical Activity on Doppler Measures of Diastolic Function. The American Journal of Cardiology. 2007;99:1629–1636. doi: 10.1016/j.amjcard.2007.01.050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Popovic ZB, Prasad A, Garcia MJ, et al. Relationship among diastolic intraventricular pressure gradients, relaxation, and preload: impact of age and fitness. Am J Physiol Heart Circ Physiol. 2006;290:H1454–9. doi: 10.1152/ajpheart.00902.2005. [DOI] [PubMed] [Google Scholar]
- 7.Borlaug BA, Redfield MM. Diastolic and systolic heart failure are distinct phenotypes within the heart failure spectrum. Circulation. 2011;123:2006–13. doi: 10.1161/CIRCULATIONAHA.110.954388. discussion 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Pluim BM, Zwinderman AH, van der Laarse A, van der Wall EE. The athlete's heart. A meta-analysis of cardiac structure and function Circulation. 2000;101:336–44. doi: 10.1161/01.cir.101.3.336. [DOI] [PubMed] [Google Scholar]
- 9.Naylor LH, George K, O'Driscoll G, Green DJ. The athlete's heart: a contemporary appraisal of the ‘Morganroth hypothesis’. Sports Med. 2008;38:69–90. doi: 10.2165/00007256-200838010-00006. [DOI] [PubMed] [Google Scholar]
- 10.Fagard RH. Athlete's heart: a meta-analysis of the echocardiographic experience. International journal of sports medicine. 1996;17(Suppl 3):S140–4. doi: 10.1055/s-2007-972915. [DOI] [PubMed] [Google Scholar]
- 11.Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med. 1990;322:1561–6. doi: 10.1056/NEJM199005313222203. [DOI] [PubMed] [Google Scholar]
- 12.Jain A, McClelland RL, Polak JF, et al. Cardiovascular imaging for assessing cardiovascular risk in asymptomatic men versus women: the multi-ethnic study of atherosclerosis (MESA) Circulation Cardiovascular imaging. 2011;4:8–15. doi: 10.1161/CIRCIMAGING.110.959403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gardin JM, McClelland R, Kitzman D, et al. M-mode echocardiographic predictors of six-to seven-year incidence of coronary heart disease, stroke, congestive heart failure, and mortality in an elderly cohort (the Cardiovascular Health Study) Am J Cardiol. 2001;87:1051–7. doi: 10.1016/s0002-9149(01)01460-6. [DOI] [PubMed] [Google Scholar]
- 14.Verma A, Meris A, Skali H, et al. Prognostic implications of left ventricular mass and geometry following myocardial infarction: the VALIANT (VALsartan In Acute myocardial iNfarcTion) Echocardiographic Study. JACC Cardiovascular imaging. 2008;1:582–91. doi: 10.1016/j.jcmg.2008.05.012. [DOI] [PubMed] [Google Scholar]
- 15.Wang TJ, Evans JC, Benjamin EJ, Levy D, LeRoy EC, Vasan RS. Natural history of asymptomatic left ventricular systolic dysfunction in the community. Circulation. 2003;108:977–82. doi: 10.1161/01.CIR.0000085166.44904.79. [DOI] [PubMed] [Google Scholar]
- 16.Redfield MM, Jacobsen SJ, Burnett JC, Jr, Mahoney DW, Bailey KR, Rodeheffer RJ. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA : the journal of the American Medical Association. 2003;289:194–202. doi: 10.1001/jama.289.2.194. [DOI] [PubMed] [Google Scholar]
- 17.Lam CS, Lyass A, Kraigher-Krainer E, et al. Cardiac dysfunction and noncardiac dysfunction as precursors of heart failure with reduced and preserved ejection fraction in the community. Circulation. 2011;124:24–30. doi: 10.1161/CIRCULATIONAHA.110.979203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kane GC, Karon BL, Mahoney DW, et al. Progression of left ventricular diastolic dysfunction and risk of heart failure. JAMA : the journal of the American Medical Association. 2011;306:856–63. doi: 10.1001/jama.2011.1201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Blair SN, Kohl HW, 3rd, Paffenbarger RS, Jr, Clark DG, Cooper KH, Gibbons LW. Physical fitness and all-cause mortality. A prospective study of healthy men and women. JAMA : the journal of the American Medical Association. 1989;262:2395–401. doi: 10.1001/jama.262.17.2395. [DOI] [PubMed] [Google Scholar]
- 20.Blair SN, Kohl HW, 3rd, Barlow CE, Paffenbarger RS, Gibbons LW, Macera CA. Changes in physical fitness and all-cause mortality. A prospective study of healthy and unhealthy men. JAMA : the journal of the American Medical Association. 1995;273:1093–8. [PubMed] [Google Scholar]
- 21.Nishimura RA, Tajik AJ. Evaluation of diastolic filling of left ventricle in health and disease: Doppler echocardiography is the clinician's Rosetta Stone. Journal of the American College of Cardiology. 1997;30:8–18. doi: 10.1016/s0735-1097(97)00144-7. [DOI] [PubMed] [Google Scholar]
- 22.Lester SJ, Tajik AJ, Nishimura RA, Oh JK, Khandheria BK, Seward JB. Unlocking the mysteries of diastolic function: deciphering the Rosetta Stone 10 years later. Journal of the American College of Cardiology. 2008;51:679–89. doi: 10.1016/j.jacc.2007.09.061. [DOI] [PubMed] [Google Scholar]
- 23.Ho CY, Solomon SD. A clinician's guide to tissue Doppler imaging. Circulation. 2006;113:e396–8. doi: 10.1161/CIRCULATIONAHA.105.579268. [DOI] [PubMed] [Google Scholar]
- 24.Pollock ML, Bohannon RL, Cooper KH, et al. A comparative analysis of four protocols for maximal treadmill stress testing. Am Heart J. 1976;92:39–46. doi: 10.1016/s0002-8703(76)80401-2. [DOI] [PubMed] [Google Scholar]
- 25.Turkbey EB, Jorgensen NW, Johnson WC, et al. Physical activity and physiological cardiac remodelling in a community setting: the Multi-Ethnic Study of Atherosclerosis (MESA) Heart. 2010;96:42–8. doi: 10.1136/hrt.2009.178426. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Savage DD, Levy D, Dannenberg AL, Garrison RJ, Castelli WP. Association of echocardiographic left ventricular mass with body size, blood pressure and physical activity (the Framingham Study) The American journal of cardiology. 1990;65:371–6. doi: 10.1016/0002-9149(90)90304-j. [DOI] [PubMed] [Google Scholar]
- 27.Gardin JM, Wagenknecht LE, Anton-Culver H, et al. Relationship of cardiovascular risk factors to echocardiographic left ventricular mass in healthy young black and white adult men and women. The CARDIA study. Coronary Artery Risk Development in Young Adults Circulation. 1995;92:380–7. doi: 10.1161/01.cir.92.3.380. [DOI] [PubMed] [Google Scholar]
- 28.Aaron CP, Tandri H, Barr RG, et al. Physical Activity and Right Ventricular Structure and Function: The MESA-Right Ventricle Study. Am J Respir Crit Care Med. 2011;183:396–404. doi: 10.1164/rccm.201003-0469OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Gidding SS, Carnethon MR, Daniels S, et al. Low cardiovascular risk is associated with favorable left ventricular mass, left ventricular relative wall thickness, and left atrial size: the CARDIA study. J Am Soc Echocardiogr. 2010;23:816–22. doi: 10.1016/j.echo.2010.05.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Grewal J, McCully RB, Kane GC, Lam C, Pellikka PA. Left ventricular function and exercise capacity. JAMA : the journal of the American Medical Association. 2009;301:286–94. doi: 10.1001/jama.2008.1022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Lam CS, Grewal J, Borlaug BA, et al. Size, shape, and stamina: the impact of left ventricular geometry on exercise capacity. Hypertension. 2010;55:1143–9. doi: 10.1161/HYPERTENSIONAHA.109.146845. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Miyachi M, Yazawa H, Furukawa M, et al. Exercise training alters left ventricular geometry and attenuates heart failure in dahl salt-sensitive hypertensive rats. Hypertension. 2009;53:701–7. doi: 10.1161/HYPERTENSIONAHA.108.127290. [DOI] [PubMed] [Google Scholar]
- 33.Koch LG, Kemi OJ, Qi N, et al. Intrinsic aerobic capacity sets a divide for aging and longevity. Circulation research. 2011;109:1162–72. doi: 10.1161/CIRCRESAHA.111.253807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Levine BD, Lane LD, Buckey JC, Friedman DB, Blomqvist CG. Left ventricular pressure-volume and Frank-Starling relations in endurance athletes. Implications for orthostatic tolerance and exercise performance. Circulation. 1991;84:1016–23. doi: 10.1161/01.cir.84.3.1016. [DOI] [PubMed] [Google Scholar]
- 35.Fujimoto N, Prasad A, Hastings JL, et al. Cardiovascular effects of 1 year of progressive and vigorous exercise training in previously sedentary individuals older than 65 years of age. Circulation. 2010;122:1797–805. doi: 10.1161/CIRCULATIONAHA.110.973784. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Arbab-Zadeh A, Dijk E, Prasad A, et al. Effect of aging and physical activity on left ventricular compliance. Circulation. 2004;110:1799–805. doi: 10.1161/01.CIR.0000142863.71285.74. [DOI] [PubMed] [Google Scholar]
- 37.Kokkinos PF, Narayan P, Colleran JA, et al. Effects of regular exercise on blood pressure and left ventricular hypertrophy in African-American men with severe hypertension. The New England journal of medicine. 1995;333:1462–7. doi: 10.1056/NEJM199511303332204. [DOI] [PubMed] [Google Scholar]
- 38.Edelmann F, Gelbrich G, Dungen HD, et al. Exercise training improves exercise capacity and diastolic function in patients with heart failure with preserved ejection fraction: results of the Ex-DHF (Exercise training in Diastolic Heart Failure) pilot study. Journal of the American College of Cardiology. 2011;58:1780–91. doi: 10.1016/j.jacc.2011.06.054. [DOI] [PubMed] [Google Scholar]
- 39.Kitzman DW, Brubaker PH, Morgan TM, Stewart KP, Little WC. Exercise training in older patients with heart failure and preserved ejection fraction: a randomized, controlled, single-blind trial. Circ Heart Fail. 2010;3:659–67. doi: 10.1161/CIRCHEARTFAILURE.110.958785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Fujimoto N, Prasad A, Hastings JL, et al. Cardiovascular effects of 1 year of progressive endurance exercise training in patients with heart failure with preserved ejection fraction. Am Heart J. 2012;164:869–77. doi: 10.1016/j.ahj.2012.06.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Nagueh SF, Appleton CP, Gillebert TC, et al. Recommendations for the evaluation of left ventricular diastolic function by echocardiography. J Am Soc Echocardiogr. 2009;22:107–33. doi: 10.1016/j.echo.2008.11.023. [DOI] [PubMed] [Google Scholar]
- 42.Abhayaratna WP, Seward JB, Appleton CP, et al. Left atrial size: physiologic determinants and clinical applications. Journal of the American College of Cardiology. 2006;47:2357–63. doi: 10.1016/j.jacc.2006.02.048. [DOI] [PubMed] [Google Scholar]
- 43.Gottdiener JS, Kitzman DW, Aurigemma GP, Arnold AM, Manolio TA. Left atrial volume, geometry, and function in systolic and diastolic heart failure of persons > or =65 years of age (the cardiovascular health study) Am J Cardiol. 2006;97:83–9. doi: 10.1016/j.amjcard.2005.07.126. [DOI] [PubMed] [Google Scholar]
- 44.Melenovsky V, Borlaug BA, Rosen B, et al. Cardiovascular features of heart failure with preserved ejection fraction versus nonfailing hypertensive left ventricular hypertrophy in the urban Baltimore community: the role of atrial remodeling/dysfunction. Journal of the American College of Cardiology. 2007;49:198–207. doi: 10.1016/j.jacc.2006.08.050. [DOI] [PubMed] [Google Scholar]
- 45.Pritchett AM, Mahoney DW, Jacobsen SJ, Rodeheffer RJ, Karon BL, Redfield MM. Diastolic dysfunction and left atrial volume: a population-based study. Journal of the American College of Cardiology. 2005;45:87–92. doi: 10.1016/j.jacc.2004.09.054. [DOI] [PubMed] [Google Scholar]
- 46.Gerdts E, Oikarinen L, Palmieri V, et al. Correlates of left atrial size in hypertensive patients with left ventricular hypertrophy: the Losartan Intervention For Endpoint Reduction in Hypertension (LIFE) Study. Hypertension. 2002;39:739–43. doi: 10.1161/hy0302.105683. [DOI] [PubMed] [Google Scholar]
- 47.Ristow B, Ali S, Whooley MA, Schiller NB. Usefulness of left atrial volume index to predict heart failure hospitalization and mortality in ambulatory patients with coronary heart disease and comparison to left ventricular ejection fraction (from the Heart and Soul Study) Am J Cardiol. 2008;102:70–6. doi: 10.1016/j.amjcard.2008.02.099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Rossi A, Cicoira M, Zanolla L, et al. Determinants and prognostic value of left atrial volume in patients with dilated cardiomyopathy. Journal of the American College of Cardiology. 2002;40:1425. doi: 10.1016/s0735-1097(02)02305-7. [DOI] [PubMed] [Google Scholar]
- 49.Takemoto Y, Barnes ME, Seward JB, et al. Usefulness of left atrial volume in predicting first congestive heart failure in patients > or = 65 years of age with well-preserved left ventricular systolic function. Am J Cardiol. 2005;96:832–6. doi: 10.1016/j.amjcard.2005.05.031. [DOI] [PubMed] [Google Scholar]
- 50.Nistri S, Galderisi M, Ballo P, et al. Determinants of echocardiographic left atrial volume: implications for normalcy. European journal of echocardiography : the journal of the Working Group on Echocardiography of the European Society of Cardiology. 2011;12:826–33. doi: 10.1093/ejechocard/jer137. [DOI] [PubMed] [Google Scholar]
- 51.Wang CY, Haskell WL, Farrell SW, et al. Cardiorespiratory fitness levels among US adults 20-49 years of age: findings from the 1999-2004 National Health and Nutrition Examination Survey. American journal of epidemiology. 2010;171:426–35. doi: 10.1093/aje/kwp412. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.