Search tips
Search criteria 


Logo of amjepidLink to Publisher's site
Am J Epidemiol. 2013 January 1; 177(1): 20–32.
Published online 2012 December 4. doi:  10.1093/aje/kws224
PMCID: PMC3590041

Prevalence, Prospective Risk Markers, and Prognosis Associated With the Presence of Left Ventricular Diastolic Dysfunction in Young Adults

The Coronary Artery Risk Development in Young Adults Study


The authors sought to determine the prevalence, prospective risk markers, and prognosis associated with diastolic dysfunction in the Coronary Artery Risk Development in Young Adults (CARDIA) Study. The CARDIA Study cohort includes approximately equal proportions of white and black men and women. The authors collected data on risk markers at year 0 (1985–1986), and echocardiography was done at year 5 when the participants were 23–35 years of age. Participants were followed for 20 years (through 2010) for a composite endpoint of all-cause mortality, myocardial infarction, heart failure, and stroke. Diastolic function was defined according to a validated hierarchical classification algorithm. In the 2,952 participants included in the primary analysis, severe diastolic dysfunction was present in 1.1% and abnormal relaxation was present in 9.3%. Systolic blood pressure at year 0 was associated with both severe diastolic dysfunction and abnormal relaxation 5 years later, whereas exercise capacity and pulmonary function abnormalities were associated only with abnormal relaxation 5 years later. After multivariate adjustment, the hazard ratios for the composite endpoint in participants with severe diastolic dysfunction and abnormal relaxation were 4.3 (95% confidence interval: 2.0, 9.3) and 1.6 (95% confidence interval: 1.1, 2.5), respectively. Diastolic dysfunction in young adults is associated with increased morbidity and mortality, and the identification of prospective risk markers associated with diastolic dysfunction could allow for targeted primary prevention efforts.

Keywords: CARDIA study, clinical outcomes, diastolic dysfunction, left ventricle

Left ventricular (LV) diastolic dysfunction detected by echocardiography has been associated with increased mortality in epidemiologic studies (13). Diastolic dysfunction is also an independent predictor of the development of heart failure (4), has been associated with reduced exercise capacity (5), and is an important component in the pathophysiology of heart failure with preserved ejection fraction (6). The prevalence of diastolic dysfunction in the community has ranged from 11% to 35%, depending on the methodology and cohort (1, 2, 7). These studies, however, were overwhelmingly comprised of older white individuals. Data on the prevalence and prognostic implications of diastolic dysfunction in asymptomatic young adults in an ethnically diverse cohort are sparse.

The Coronary Artery Risk Development in Young Adults (CARDIA) Study provides a unique opportunity for longitudinal follow up of healthy young adults. A previous cross-sectional analysis of this cohort showed that diastolic function (based solely on blood flow markers of mitral inflow) was associated with age, sex, blood pressure, and lung function (8). However, the prognostic significance of these findings remains unclear. Given the longitudinal nature of our cohort, we also have the opportunity to determine which factors are antecedents of diastolic dysfunction. We hypothesized that individuals with diastolic dysfunction would have a higher incidence of cardiovascular events than persons with normal diastolic function. We further hypothesized that antecedent traditional risk factors and other physiologic markers, including exercise capacity, are associated with the presence of diastolic dysfunction. Identification of individuals who were at a high risk of diastolic dysfunction and associated complications could allow for improved primary prevention efforts.



The CARDIA Study is an ongoing community-based study funded by the National Heart, Lung, and Blood Institute. Details of the study design and objectives have been published elsewhere (9, 10). Briefly, the initial cohort included 5,115 individuals recruited from 4 urban communities: Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California. The sample consisted of approximately equal proportions of black and white men and women who were 18–30 years of age at the time of enrollment and 23–35 years of age at the time of echocardiography.

For the present study, we used data from examination years 0 (clinical risk markers) and 5 (clinical risk markers and echocardiographic measures) and clinical outcomes. The date of censoring was January 26, 2010. A total of 4,352 participants attended the year 5 examination, and 4,243 participants underwent echocardiographic examination. Women who were pregnant at the time of examination at year 0 or year 5 (n = 59) were excluded from the analysis, leaving 4,184 subjects in the analytic cohort. We also excluded participants who had moderate or severe valvular disease (n = 42) and 2 participants who died but whose time of death was undetermined. Subjects for whom we were missing covariates were excluded from certain analyses; this is described in greater detail below. Thus, the follow-up time for clinical events from the time of echocardiography was a maximum of 20 years and total of 74,939 person-years.

Data collection

The following variables were measured at year 0 as described previously (10): 1) age; 2) sex; 3) race; 4) height; 5) weight; 6) body mass index (calculated as weight in kilograms divided by the square of the height in meters); 7) waist circumference; 8) educational level; 9) systolic and diastolic blood pressure; 10) lipid profile; 11) fasting blood glucose; 12) serum creatinine; 13) graded exercise test, which has been previously described in more detail (11); 14) cigarette smoking history, coded as current, former, or never; 15) alcohol consumption; 16) Cornell voltage-QRS duration product (12) and electrocardiographic LV mass index (13); 17) forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC), and FEV1/FVC ratio; and 18) use of antihypertensive or hypoglycemic medications. Pulmonary function testing, electrocardiographic variables, and exercise testing were not available at the year 5 examination; all other covariates described above were measured at year 5. Blood pressure was measured 3 times by trained and certified technicians using standardized methods after the participants rested for 5 minutes. The mean of the last 2 measurements of blood pressure was used for analyses. At the baseline examination, Hawksley random-zero sphygmomanometers (W.A. Baum Co, Copiague, New York) were used.


The design, collection, analysis, and storage of echocardiographic data obtained using an Acuson ultrasound machine (Siemens Medical Solutions USA, Malvern, Pennsylvania) in the CARDIA Study were similar to what has been previously described (14). In the apical 4-chamber view, the ratio of mitral inflow early to late diastolic velocity (E/A ratio) was determined from pulse-wave Doppler at the tips of the mitral annulus. Isovolumetric relaxation time was also determined from Doppler evaluation in the apical 4-chamber view. LV mass was derived from M-mode measurement of LV diastolic dimension (LVDd), interventricular septum thickness in diastole, and LV posterior wall thickness in diastole. LV mass was calculated using the formula LV mass (in grams) = 0.80{1.04 [(LVDd + IVSTd + LVPWd)3 – LVDd3]} + 0.6, where IVSTd is the interventricular septum thickness in diastole and LVPWd is the LV posterior wall thickness in diastole, and indexed to height2.7, as previously described (15). The relative wall thickness was derived from the formula LVPWd × 2/LVDd. LV mass was also derived using a 2-dimensional echocardiography method by combining images from single beats recorded in the parasternal short-axis view at the papillary muscle level and the apical 4-chamber view (16). LV end diastolic volumes were measured from 2-dimensional images from views with adequate visualization of at least 80% of the endocardium and calculated using truncated ellipsoid and modified Simpson's rule. Left atrial (LA) volume was calculated using the formula LA volume = π/6 × MMLAD × A4LAAD, where MMLAD is the M-mode LA diameter and A4LAAD is the 2-dimensional area of the left atrium in the apical 4-chamber view. Both the LV volume and LA volume were indexed to the body surface area.

Classification of diastolic function

The partition values shown for each of the echocardiographic indices of ventricular morphology, ventricular mass, and Doppler indices of diastolic function are as previously described and are shown in Figure 1 (6, 17). To classify diastolic function, participants were assigned into 1 of 3 groups: severe diastolic dysfunction (SDD), abnormal relaxation, or normal. Briefly, diastolic function is based on the presence or absence of morphologic substrate for abnormal left ventricular relaxation, Doppler markers of abnormal relaxation, reduced left ventricular compliance, or Doppler markers of elevated LV filling pressures (18). Participants with SDD were identified first, based on presence of at least 3 markers of adverse LV or LA morphology or a high mitral inflow E/A ratio. Next, those with abnormal relaxation were identified. To be classified as having abnormal relaxation, the participant must have, in addition to an abnormal mitral filling pattern, at least 1 marker of abnormal cardiac morphology and cannot have an abnormal LV ejection fraction (<50%). Finally, to be classified as normal, participants must have at least 3 markers of normal cardiac morphology and/or mitral inflow Doppler in addition to a normal or only mildly reduced LV ejection fraction (>40%), without belonging to the abnormal diastolic function groups. This classification algorithm avoids the problem of assigning patients into a pseudonormal diastolic function class by using age-appropriate Doppler markers in conjunction with morphologic markers with high specificity for abnormal diastolic function. Participants who failed to meet strict criteria for assignments into 1 of the 3 groups (most commonly because of incomplete or missing echocardiographic measurements) were deemed “indeterminate” and excluded from the primary analysis but were included in a sensitivity analysis after multiple imputations.

Figure 1.
Three-stage classification of diastolic function, Coronary Artery Risk Development in Young Adults Study, 1990–1991. LV, left ventricular.

Statistical analysis

Statistical analyses were conducted using SAS, version 9.2 for Windows (SAS Institute, Inc., Cary, North Carolina). Analysis of variance or chi-square tests were used to compare year 0 characteristics across diastolic function groups. Polytomous logistic regression was then used to estimate age-, race-, and sex-adjusted odds ratios for each variable, comparing persons who had severe diastolic dysfunction and abnormal relaxation with those who had normal diastolic function. For continuous independent variables, odds ratios were estimated per each 1-standard-deviation difference in the variable.

To determine whether variables were independently associated with diastolic group classification, variables that were statistically significant in the age-, race-, and sex-adjusted analysis or that were considered biologically important were included in a multivariable polytomous logistic regression model. Participants who were classified into the indeterminate group (n = 1,188), largely because they were missing data on diastolic function variables, were excluded from these analyses. Because of missing data on the individual predictor variables, the sample size for the age-, race-, and sex-adjusted analyses ranged from 2,815 to 2,952, and that for the full multivariable model was 2,652.

The occurrence of a first event of all-cause mortality, stroke, myocardial infarction, or hospitalization for heart failure comprised the composite endpoint. A separate endpoint with only cardiovascular disease death or nonfatal events (heart failure, stroke, or myocardial infarction) was also analyzed. In this analysis, patients who suffered deaths not attributable to cardiovascular disease were censored at the date of death. Cox proportional hazard models were used to estimate the univariate associations between diastolic function groups and outcomes. The hazard ratios and 95% confidence intervals associated with SDD, abnormal relaxation, or indeterminate class compared with those with normal diastolic function were estimated. Given the small number of events in the group with SDD, we took a stepwise approach to adjustment for potential confounders to examine for and avoid potential over-fitting of the models. In addition to estimating unadjusted associations (model 1), we fitted 2 multivariable models for adjustment of baseline characteristics. In model 2, we adjusted for age, race, and sex. In model 3, we adjusted for the covariates in model 2 plus systolic blood pressure, body mass index, total and high density lipoprotein cholesterol levels, cigarette smoking, and self-reported diabetes mellitus. The sample size for models 1, 2, and 3 was 3,994 because we were missing year 5 covariate data for 146 participants. A likelihood ratio test was done to assess the overall fit of the models. Direct-adjusted survival curves adjusted for the variables in model 3 were obtained using a SAS macro developed by Zhang et al. (19).

We then used multiple imputations on the sample of 4,140 participants to generate 5 complete data sets by replacing missing values with plausible values using the sequential regression imputation method (20) available in the IVEware software package (Institute for Social Research, Ann Arbor, Michigan). After the 5 imputations for missing variables, participants initially classified as “indeterminate” were reclassified per the classification algorithm in each of the 5 data sets. The age-, race-, and sex-adjusted analysis and the multivariable analysis were repeated on the 5 complete data sets using ordinary logistic regression or Cox regression, as appropriate, and the results from the 5 sets of analyses were combined using PROC MIANALYZE in SAS, resulting in 1 overall set of multiple imputation estimates. These estimates were considered a sensitivity analysis and compared with those obtained from the primary analyses described above.


Participant characteristics

The prevalence of diastolic function in the participants who could be classified was as follows: SDD, 1.1% (n = 33), abnormal relaxation, 9.3% (n = 275), and normal, 89.6% (n = 2,644). Of the total number of participants in the sample, 1,188 (28.7%) could not be classified. Participant characteristics at year 0 stratified by diastolic function at year 5 are shown in Table 1. Unadjusted analysis of year 0 characteristics revealed that several covariates, including systolic and diastolic blood pressure, lipid profile, exercise capacity, electrocardiographic LV mass, and percentage of predicted FEV1 and FVC, differed significantly across diastolic function groups. Table 2 shows echocardiographic characteristics stratified by classification of diastolic function, both of which were obtained at year 5. The cross-sectional analysis of year 5 demographic and clinical covariates with diastolic function classification is shown in Web Table 1 (available at, and we observed similar associations of clinical risk markers with diastolic function classification.

Table 1.
Characteristics of Study Sample at Year Zero (1985–1986), Stratified by Diastolic Function at Year 5 (1990–1991), Coronary Artery Risk Development in Young Adults Study
Table 2.
Echocardiographic Characteristics at Year 5 Stratified by Diastolic Function Category, Coronary Artery Risk Development in Young Adults Study, 1990–1991

Age-, race-, and sex-adjusted polytomous logistic regression

The results of age-, race-, and sex-adjusted polytomous logistic regression are shown in Table 3. Systolic blood pressure was the only traditional cardiovascular risk factor associated with SDD 5 years later: For each 1-standard-deviation increase in systolic blood pressure (11 mm Hg), there was a 51% higher odds (odds ratio = 1.51, 95% confidence interval (CI): 1.07, 2.13) of having SDD. Lipid profile, fasting glucose level, and smoking status were not associated with SDD. Higher values of the percentage of predicted FEV1 and symptom-limited graded exercise duration were inversely associated with SDD. Electrocardiographic LV mass index and Cornell voltage-QRS duration product also were significantly associated with SDD.

Table 3.
Age-, Race-, and Sex-adjusted Odds Ratios With Polytomous Logistic Regression, Coronary Artery Risk Development in Young Adults Study, 1985–1991

Systolic and diastolic blood pressure, along with metabolic measures including total cholesterol and triglycerides, were significantly associated with abnormal relaxation 5 years later. The percentage of predicted FEV1, percentage of FVC, and graded exercise duration were inversely associated with abnormal relaxation. Electrocardiographic LV mass index and Cornell voltage-QRS duration product were not associated with abnormal relaxation.

Multivariable analysis

A multivariable model was fitted to simultaneously adjust for covariates that were significantly associated with diastolic function in the age-, race-, and sex-adjusted analysis or that were considered biologically important. These results are presented in Web Table 2. Systolic blood pressure remained significantly associated with both SDD and with abnormal relaxation. The percentage of predicted FEV1 was inversely associated with abnormal relaxation. We also explored including other covariates, such as smoking, triglycerides, insulin levels, and alcohol consumption, in the multivariable model. However, none of these variables contributed importantly to the results, so they were excluded from the final model for the sake of parsimony.

Clinical outcomes

Over 20 years and 74,939 person-years of follow up, there were a total of 223 composite events. The distribution of events and hazard ratios by diastolic function category are shown in Table 4. With normal diastolic function as the referent group, the unadjusted hazard ratios for incidence of the composite event for abnormal relaxation and SDD were 2.40 (95% CI: 1.58, 3.65) and 5.79 (95% CI: 2.70, 12.43), respectively. The results remained significant after adjustment for age, race, and sex in model 2 and after additional adjustment for other cardiovascular risk factors in model 3.

Table 4.
Clinical Outcomes and Hazards Ratios by Diastolic Function Category, Coronary Artery Risk Development in Young Adults Study, 1990–2010

There were a total of 103 cardiovascular disease events, including myocardial infarction, stroke, heart failure, and cardiovascular death. The hazard ratios for the occurrence of any cardiovascular disease event were 3.49 (95% CI: 1.94, 6.30) for abnormal relaxation and 10.96 (95% CI: 4.35, 27.65) for SDD. The hazard ratios remained significant after adjustment for age, race, and sex in model 2 and after additional adjustment for cardiovascular risk factors in model 3. The cumulative survival distribution is shown in Figures 2A and 2B. We further adjusted the models for LV mass and LV end-diastolic diameter, and the hazard ratios were unchanged (results not shown).

Figure 2.
Cumulative probability of event-free survival for composite endpoint and cardiovascular disease endpoint, Coronary Artery Risk Development in Young Adults Study, 1990–2010.

Participants with indeterminate diastolic function comprised 28.7% (n = 1,188) of the overall study sample; the primary reason for the indeterminate classification was missing echocardiographic data. We performed multiple imputations to reclassify these participants and repeated the analyses. The results of multivariable-adjusted associations between year 0 covariates and diastolic function after multiple imputations are shown in Web Table 3. We observed associations that were similar to those that we observed before multiple imputations. Table 5 shows the clinical outcomes and hazard ratios by diastolic function category after multiple imputations. The hazard ratios were similar to those observed in the unimputed analysis.

Table 5.
Clinical Outcomes and Hazards Ratios by Diastolic Function Category After Multiple Imputations (n = 4,128), Coronary Artery Risk Development in Young Adults Study, 1990–2010



LV diastolic dysfunction is associated with increased risk of cardiovascular disease in older adults, but data on diastolic function in young adults are limited. In CARDIA participants who were 23–35 years of age, we found that the prevalence of abnormal relaxation was 9.3% and the prevalence of SDD was 1.1%. After adjustment for clinical and demographic factors and over 20 years of follow up, the presence of SDD was associated with a hazard ratio for our composite outcome of 4.29 (95% CI: 1.97, 9.33), and presence of abnormal relaxation was associated with a hazard ratio of 1.64 (95% CI: 1.06, 2.52). To the best of our knowledge, our study is the first to examine the prognosis associated with diastolic dysfunction in an asymptomatic and ethnically diverse population of younger adults.

We also examined potential antecedents of diastolic dysfunction and found that systolic blood pressure was associated with both SDD and abnormal relaxation. Other traditional cardiovascular risk factors, including lipid profile, smoking status, or fasting blood glucose levels, were not significantly associated with SDD. Although it is possible that some degree of diastolic dysfunction was present at the baseline examination, we did observe that lipid profile and fasting glucose levels were prospectively associated with the presence of abnormal relaxation 5 years later. Increased exercise duration was inversely associated with SDD and abnormal relaxation. We also observed an association between pulmonary function parameters and diastolic dysfunction in asymptomatic young individuals.


In previous studies, investigators have examined the prognostic significance of diastolic dysfunction in community-dwelling adults (1, 3, 4); however, these studies have overwhelmingly included older adults and very few black participants. Redfield et al. studied community-dwelling adults 45 years of age or older in Olmsted County, Minnesota, and found that the prevalence of diastolic abnormalities was 28% (1). More than half of those individuals had no signs or symptoms of heart failure; however, even mild diastolic dysfunction predicted an increase in all-cause mortality. The World Health Organization Multinational Monitoring of Trends and Determinants in Cardiovascular Disease Project (MONICA) Augsburg study included a small number (<300) of younger individuals 25–35 years of age and found that the prevalence of diastolic dysfunction was 2.8% (7). Although the definitions of diastolic dysfunction differed across studies, the prevalence of any diastolic abnormalities in our study was 10.4%, and both abnormal relaxation and SDD were associated with significantly increased risk for cardiovascular events and all-cause mortality.

In the present study, echocardiography was performed in 1990–1991, and we do not have data on more recently developed measures of diastolic function, including tissue Doppler imaging and speckle tracking. However, our classification scheme utilized easily measured markers of blood flow and LA and ventricular morphology. This classification has been validated in patients with coronary artery disease (21) and hypertension (22). Most importantly, the presence of diastolic dysfunction as defined by our classification scheme in the present study was associated with markedly increased incidence of cardiovascular disease and all-cause mortality.

Our findings suggest that SDD and abnormal relaxation may represent biologically distinct phenotypes in young adults. We found that traditional cardiovascular risk factors, including blood pressure, lipid profile, and glucose intolerance, were significantly associated with abnormal relaxation but not with SDD. Given that individuals with SDD had significantly increased LV internal diameter and increased LV mass, our findings regarding this entity should be interpreted with caution; we cannot exclude the possibility that diastolic dysfunction in individuals with SDD was due to an asymptomatic cardiomyopathy. Furthermore, individuals with SDD had abnormalities on electrocardiogram 5 years before echocardiography, suggesting long-standing structural heart disease. Nonetheless, although SDD may be due to a cardiomyopathy in this sample, its presence was not otherwise detected, and it was highly predictive of future morbidity and mortality in these individuals. Given that traditional cardiovascular risk factors were significantly associated with abnormal relaxation but not with SDD, individuals with abnormal relaxation may benefit the most from intensive primary and primordial prevention efforts, although further study is needed.

A limitation of our study was the lack of availability of tissue Doppler imaging and speckle tracking to define diastolic function, as we have discussed above. Another limitation is that 28% of the total study population had indeterminate diastolic function because of incomplete or missing data. After multiple imputations, members of this group were reclassified into normal diastolic function (≈75%), abnormal relaxation (23%), and severe diastolic dysfunction (2%), confirming that this is a heterogeneous group. The secondary analysis confirmed the results that were obtained before imputation and with exclusion of the indeterminate individuals. As expected in this young population, there was a relatively small proportion of individuals with severe diastolic dysfunction. After multiple imputations, we reclassified an additional 24 participants into this group, and the imputed results were concordant with those from the primary analysis. Another limitation is that CARDIA participants who did not return for echocardiography at year 5 were more likely to be men, black, and smokers.

Because echocardiography was not performed at the year 0 examination, we were unable to estimate the incidence of diastolic dysfunction. Thus, we cannot exclude the possibility that diastolic abnormalities were present at the baseline examination. In fact, given the association of electrocardiographic LV mass index with SDD, it is likely that structural abnormalities predate the echocardiographic examination.


We conducted an analysis to determine the prevalence, prospective risk markers, and prognosis associated with diastolic dysfunction in healthy young adults. We found that the prevalence of any diastolic abnormalities in adults 23–35 years of age was 10.4%, and these findings were associated with markedly increased risk for all-cause mortality and cardiovascular events. We also found that elevated systolic blood pressure, abnormal pulmonary function, and electrocardiographic evidence for cardiac structural abnormalities were associated with the presence of severe diastolic dysfunction 5 years later. Traditional cardiovascular risk factors, including blood pressure, lipid profile, and elevated fasting glucose levels, were associated with abnormal relaxation. Increased exercise duration and percentage of predicted FEV1 were inversely associated with the presence of diastolic dysfunction. Further study is needed to determine whether more intensive prevention efforts can reduce complications of diastolic dysfunction in young adults.

Supplementary Material

Web Tables:


Author affiliations: Department of Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois (Chintan S. Desai, Laura A. Colangelo, Kiang Liu, Donald M. Lloyd-Jones); School of Public Health, University of Minnesota, Minneapolis, Minnesota (David Jacobs); and National Heart, Lung, and Blood Institute, Bethesda, Maryland (Nakela Cook); Division of Cardiology, Northwestern University, Chicago, Illinois (Donald M. Lloyd-Jones.); and First Cardiology Consultants, Ikoyi, Lagos, Nigeria (Kofo O. Ogunyankin).

The Coronary Artery Risk Development in Young Adults (CARDIA) Study was conducted and supported by the National Heart, Lung, and Blood Institute in collaboration with the University of Alabama at Birmingham (grants N01-HC95095 and N01-HC48047), the University of Minnesota (grant N01-HC48048), Northwestern University (grant N01-HC48049), and the Kaiser Foundation Research Institute (grant N01-HC48050). Dr. Desai was supported by National Institutes of Health grant T32 HL 69771-8. This manuscript has been reviewed by CARDIA for scientific content and consistency of data interpretation with previous CARDIA publications.

The authors thank the staff of the Coronary Artery Risk Development in Young Adults Study.

Conflict of interest: none declared.


1. Redfield MM, Jacobsen SJ, Burnett JC, Jr, et al. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA. 2003;289(2):194–202. [PubMed]
2. Abhayaratna WP, Marwick TH, Smith WT, et al. Characteristics of left ventricular diastolic dysfunction in the community: an echocardiographic survey. Heart. 2006;92(9):1259–1264. [PMC free article] [PubMed]
3. Achong N, Wahi S, Marwick TH. Evolution and outcome of diastolic dysfunction. Heart. 2009;95(10):813–818. [PubMed]
4. Aurigemma GP, Gottdiener JS, Shemanski L, et al. Predictive value of systolic and diastolic function for incident congestive heart failure in the elderly: the cardiovascular health study. J Am Coll Cardiol. 2001;37(4):1042–1048. [PubMed]
5. Grewal J, McCully RB, Kane GC, et al. Left ventricular function and exercise capacity. JAMA. 2009;301(3):286–294. [PMC free article] [PubMed]
6. Paulus WJ, Tschope C, Sanderson JE, et al. How to diagnose diastolic heart failure: a consensus statement on the diagnosis of heart failure with normal left ventricular ejection fraction by the Heart Failure and Echocardiography Associations of the European Society of Cardiology. Eur Heart J. 2007;28(20):2539–2550. [PubMed]
7. Fischer M, Baessler A, Hense HW, et al. Prevalence of left ventricular diastolic dysfunction in the community. Results from a Doppler echocardiographic-based survey of a population sample. Eur Heart J. 2003;24(4):320–328. [PubMed]
8. Xie X, Gidding SS, Gardin JM, et al. Left ventricular diastolic function in young adults: the Coronary Artery Risk Development in Young Adults Study. J Am Soc Echocardiogr. 1995;8(6):771–779. [PubMed]
9. Cutter GR, Burke GL, Dyer AR, et al. Cardiovascular risk factors in young adults. The CARDIA baseline monograph. Control Clin Trials. 1991;12(1 suppl):1S–77S. [PubMed]
10. Friedman GD, Cutter GR, Donahue RP, et al. CARDIA: study design, recruitment, and some characteristics of the examined subjects. J Clin Epidemiol. 1988;41(11):1105–1116. [PubMed]
11. Carnethon MR, Sternfeld B, Schreiner PJ, et al. Association of 20-year changes in cardiorespiratory fitness with incident type 2 diabetes: the coronary artery risk development in young adults (CARDIA) fitness study. Diabetes Care. 2009;32(7):1284–1288. [PMC free article] [PubMed]
12. Okin PM, Roman MJ, Devereux RB, et al. Electrocardiographic identification of increased left ventricular mass by simple voltage-duration products. J Am Coll Cardiol. 1995;25(2):417–423. [PubMed]
13. Rautaharju PM, Park LP, Gottdiener JS, et al. Race- and sex-specific ECG models for left ventricular mass in older populations. Factors influencing overestimation of left ventricular hypertrophy prevalence by ECG criteria in African-Americans. J Electrocardiol. 2000;33(3):205–218. [PubMed]
14. Gardin JM, Wong ND, Bommer W, et al. Echocardiographic design of a multicenter investigation of free-living elderly subjects: the Cardiovascular Health Study. J Am Soc Echocardiogr. 1992;5(1):63–72. [PubMed]
15. de Simone G, Daniels SR, Devereux RB, et al. Left ventricular mass and body size in normotensive children and adults: assessment of allometric relations and impact of overweight. J Am Coll Cardiol. 1992;20(5):1251–1260. [PubMed]
16. Helak JW, Reichek N. Quantitation of human left ventricular mass and volume by two-dimensional echocardiography: in vitro anatomic validation. Circulation. 1981;63(6):1398–1407. [PubMed]
17. Lang RM, Bierig M, Devereux RB, et al. Recommendations for chamber quantification: a report from the American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr. 2005;18(2):1440–1463. [PubMed]
18. Ogunyankin KO. Assessment of left ventricular diastolic function: the power, possibilities, and pitfalls of echocardiographic imaging techniques. Can J Cardiol. 2011;27(3):311–318. [PubMed]
19. Zhang X, Loberiza FR, Klein JP, et al. A SAS macro for estimation of direct adjusted survival curves based on a stratified Cox regression model. Comput Methods Programs Biomed. 2007;88(2):95–101. [PubMed]
20. He Y. Missing data analysis using multiple imputation: getting to the heart of the matter. Circ Cardiovasc Qual Outcomes. 2010;3(1):98–105. [PMC free article] [PubMed]
21. Ogunyankin KO, Burggraf GW, Abiose AK, et al. Validity of revised Doppler echocardiographic algorithms and composite clinical and angiographic data in diagnosis of diastolic dysfunction. Echocardiography. 2006;23(10):817–828. [PubMed]
22. Ogunyankin KO, Day AG, Lonn E. Cardiac function stratification based on echocardiographic or clinical markers of left ventricular filling pressures predicts death and hospitalization better than stratification by ventricular systolic function alone. Echocardiography. 2008;25(2):169–181. [PubMed]

Articles from American Journal of Epidemiology are provided here courtesy of Oxford University Press