Sex, Age, and Disease Affect Echocardiographic Left Ventricular Mass and Systolic Function in the Free-Living Elderly
The Cardiovascular Health Study
Background Left ventricular (LV) hypertrophy, as measured by M-mode echocardiography, is an independent predictor of mortality and/or morbidity from coronary heart disease (CHD). LV global and segmental systolic dysfunction also have been associated with myocardial ischemia and cardiovascular morbidity and mortality. Echocardiographic data, especially two-dimensional, have not been available previously from multicenter-based studies of the elderly. This report describes the distribution and relation at baseline of echocardiographic LV mass and global and segmental LV wall motion to age, sex, and clinical disease category in the Cardiovascular Health Study (CHS), a cohort of 5201 men and women (4850 white) 65 years of age and older.
Methods and Results M-mode LV mass adjusted for body weight increased modestly with age (P<.0001), increasing less than one gram per year increase in age for both men and women. After adjustment for weight, LV mass was significantly greater in men than in women and in participants with clinical CHD compared with participants with neither clinical heart disease nor hypertension (both P<.001). Across all CHS age subgroups, the difference in weight-adjusted LV mass by sex was greater in magnitude than the difference related to clinical CHD. M-mode measurements of LV mass could not be made in 34% of CHS participants, and this was highly related to age (29% in the 65 to 69 year versus 50% in the 85+ year age group, P<.001) and other risk factors. In participants with clinical CHD and with neither clinical heart disease nor hypertension, LV ejection fraction and segmental wall motion abnormalities were more prevalent in men than women (all P<.001). Of interest, 0.5% of men and 0.4% of women with neither clinical heart disease nor hypertension had LV segmental wall motion abnormalities, suggesting silent disease, compared with 26% of men and 10% of women in the clinical CHD group (P<.0001). Multivariate analyses revealed male sex and presence of clinical CHD (both P<.001) to be independent predictors of LV akinesis or dyskinesis.
Conclusions Significant baseline relations were detected between differences in sex, prevalent disease status, and echocardiographic measurements of LV mass and systolic function in the CHS cohort. Age was weakly associated with LV mass measurements and LV ejection fraction abnormalities. These relations should be considered in evaluating the preclinical and clinical effects of CHD risk factors in the elderly.
Recent data from the Framingham Study have indicated that M-mode echocardiographically determined left ventricular (LV) hypertrophy is an independent predictor of mortality and morbidity from coronary heart disease (CHD).1 2 In studies of patients with clinical CHD, various parameters of LV global and segmental systolic function that can be evaluated by two-dimensional (2-D) echocardiography (echo), such as decreased ejection fraction and abnormal segmental wall motion, have been associated with greater cardiovascular morbidity and mortality.3 4 However, echo data, especially measurements derived from 2-D echo, have not been available from population-based studies of the free-living elderly. Therefore, a comprehensive echo protocol, including M-mode, 2-D, and Doppler recordings, was incorporated into the design of the Cardiovascular Health Study (CHS), a prospective, population-based study of 5201 men and women aged 65 years and older.
The main objectives of incorporating echo into CHS were (1) to examine the relation of M-mode and 2-D echo indexes of cardiac structure and function to both established and newly defined risk factors for CHD and stroke and (2) to determine whether these echo indexes represent independent predictors of morbidity and mortality for CHD and stroke. The purpose of this article is to describe the relation of age, sex, and disease status to baseline measurements of LV mass (from M-mode echo) and global and segmental LV wall motion (from 2-D echo). We examine these relations among three clinically defined subgroups of CHS participants of particular interest: people with clinical CHD, people with hypertension in the absence of CHD, and people with neither CHD nor hypertension. In addition, we present reference equations with lower and upper limits for expected values for LV mass among apparently healthy older people.
The CHS baseline cohort was composed of 5201 participants—2246 men and 2955 women. Of this cohort, 4850 participants were white, 307 were black, and 44 were classified as “other nonwhite.” Ages ranged from 65 to 100 years in both men and women (mean±SD, 73.3±5.8 years in men and 72.4±5.4 years in women). The CHS participants were recruited and examined at four Field Centers: Forsyth County, NC; Sacramento County, Calif; Allegheny County, Pa; and Washington County, Md. The Echocardiography Reading Center was located at the University of California, Irvine, and the Coordinating Center at the University of Washington, Seattle. The overall design, objectives, and recruitment strategy of this study, initiated and supported by the National Heart, Lung, and Blood Institute, have been presented in detail elsewhere.5 6
Echo Performance and Reading Protocol
Echo was performed during an extensive CHS clinical examination that included interviews related to medical history, physical activity, personal habits, cognitive function, and dietary intake; anthropometric data; recumbent, sitting, and standing blood pressures (BP); resting and ambulatory ECG; spirometry; carotid ultrasound; and laboratory studies.
The design of the echo protocol used in CHS has been described in detail elsewhere.7 Briefly, for each subject, a baseline echo was recorded onto super-VHS tape with a Toshiba SSH-160A cardiac ultrasound machine using a standardized protocol. Thirty minutes were allotted for obtaining M-mode, 2-D, spectral, and color Doppler studies in each subject. Videotapes were sent to the Echocardiography Reading Center where images were displayed and digitized. Measurements were made from digitized images using an off-line image-analysis system equipped with customized computer algorithms. Quality-control measures included standardized training of echo technicians and readers, periodic technician observation by a trained echocardiographer, blind duplicate readings to establish interreader and intrareader measurement variabilities, periodic reader review sessions, phantom studies on the ultrasound equipment, and quality-control audits.
This report focuses on M-mode LV mass and 2-D qualitative LV ejection fraction and semiquantitative LV segmental wall motion measurements.
where VSTd is ventricular septal thickness at end diastole, LVIDd is LV internal dimension at end diastole, and PWTd is LV posterior wall thickness at end diastole.
Global LV function was qualitatively assessed from 2-D echo images. Specifically, readers subjectively scored (without quantitative measurement) LV ejection fraction as normal, abnormal, or borderline. In addition, readers scored LV segmental wall motion semiquantitatively in the parasternal short-axis, apical four-chamber, and apical two-chamber views, using the 20-segment model proposed by the ASE.10 However, since in the parasternal short-axis view only the papillary muscle level, but not the mitral leaflet level, was scored, data were obtained for only 16 of the 20 segments. Each segment was scored as normal, hypokinetic (decreased wall motion), akinetic (absence of wall motion), or dyskinetic (paradoxical wall motion). Abnormal wall motion in the ventricular septum was scored without regard to possible nonischemic etiologies, such as postsurgical, pacemaker, left bundle branch block, or right ventricular volume overload.
For the purposes of examining the relation of prevalent disease status to echo measurements, participants were divided into three mutually exclusive groups: (1) participants with clinical CHD, (2) participants with hypertension in the absence of clinical CHD, and (3) participants with neither clinical heart disease (including clinical CHD) nor hypertension. Participants with other (non-CHD) heart disease in the absence of hypertension were excluded from all analyses. Various clinical characteristics, ie, age, weight, height, and prevalence of obesity and diabetes, are shown by age and disease status groups in Table 1⇓. Because of the small number of black and other nonwhite participants in the baseline CHS cohort, detailed racial subgroup analyses were felt to be unjustified.
Participants With Clinical CHD
This group was composed of participants (689 men and 641 women) with a previous history of any of the following: (1) reported or silent myocardial infarction, (2) reported angina pectoris, (3) previous coronary artery bypass surgery or coronary artery angioplasty, or (4) use of nitrates.11 Silent myocardial infarction was defined as evidence of an old myocardial infarction on ECG in the absence of self-reported infarction by the participant. The Minnesota codes used to define old myocardial infarction were (1) 1.1 or 1.2 (except 1.2.8) or (2) 1.2.8 or 1.3 and 4.1 through 4.3 or 5.2 and 5.3.
Participants With Hypertension
This group was composed of participants (951 men and 1471 women) with any of the following: (1) self-reported history of hypertension, (2) use of antihypertensive medication, (3) sitting systolic BP ≥140 mm Hg, or (4) sitting diastolic BP ≥90 mm Hg. BP was measured in the right arm after a 5-minute rest using an appropriately sized cuff and a Hawksley random-zero sphygmomanometer (model 7076, Hawksley & Sons Ltd). The average of two BP measurements was used. These participants did not have evidence of clinical CHD.
Participants With Neither Clinical Heart Disease (Including CHD) Nor Hypertension
This group was composed of participants (531 men and 786 women) with neither clinical CHD nor hypertension as previously defined and without other clinical heart disease.
An additional 132 participants (68 men and 64 women; mean age, 72.8 years) with other heart disease in the absence of CHD or hypertension (as previously defined) were excluded from all analyses because of an insufficient sample size to examine the conditions individually and because the conditions were too diverse to analyze in combination. Reasons for exclusion included self-report of physician-diagnosed rheumatic heart disease or heart valve pathology (45%), use of digitalis or class I antiarrhythmic agents (27%), atrial fibrillation by ECG or self-report of physician diagnosis (30%), pacemaker implant or other heart surgery (8%), congestive failure (4%), and moderate or severe mitral (3%) or aortic (2%) valvular regurgitation on color Doppler echo. Twenty-nine percent of the participants had multiple reasons for exclusion.
Diabetes was defined as the presence of any of the following: (1) self-report of physician-diagnosed diabetes, (2) use of insulin or oral hypoglycemic agents, (3) fasting glucose ≥140 mg/dL, or (4) 2-hour post load glucose ≥200 mg/dL. Obesity was defined using sex-specific body mass index cut points (ie, ≥27.3 kg/m2 for women and ≥27.8 kg/m2 for men) reported by de Simone and colleagues.12
Personal characteristics and prevalent disease were examined as potential correlates of missing M-mode measurements using χ2 tests for association and logistic regression for the categorical and continuous variables, respectively. Linear regression was used to examine the relation between LV mass and measures of body size—specifically, weight, height, and body surface area. Since preliminary analyses indicated that adjustment for body weight was necessary, ANCOVA with weight as the covariate was used to test for differences in mean LV mass measurements across age, sex, and disease status groups. ANCOVA was also used to examine the effects of age, sex, diabetes, and obesity on LV mass in participants with neither clinical heart disease (including CHD) nor hypertension.
Reader Variability for LV Mass Measurements
To assess interreader and intrareader measurement variabilities, percent differences between LV mass measurements made by two technician-readers or between two measurements made by a single reader on the same study in two different sessions were calculated using the following formula:
Overall median and mean percent interreader and intrareader measurement differences were calculated from 270 and 84 paired studies, respectively.
Development of Reference Equations for LV Mass in a Healthy Subgroup
By excluding participants with either LV ejection fraction or wall motion abnormalities from our previously defined group of participants with neither clinical heart disease (including CHD) nor hypertension, we defined a subgroup that was apparently free of clinical heart disease and hypertension. This subgroup, which was used to develop reference equations for LV mass, did not include any participants with a condition that was found in preliminary analyses to be a statistically significant, independent predictor of LV mass. The “healthy” subgroup consisted of 516 men and 773 women, with 339 and 569 having available echo measurements of LV mass, respectively. For determination of reference equations for LV mass among healthy elderly subjects who were apparently free of clinical heart disease and hypertension, log-transformed LV mass was modeled as a linear function of log-transformed body weight and sex in this subgroup. Both LV mass and body weight were log-transformed because a multiplicative model best described the relation of body weight to LV mass. Preliminary analysis indicated that after adjustment for weight, no adjustment for height was necessary. Likewise, after adjustment for weight, age had only a modest effect on LV mass (the mean weight-adjusted LV mass increased less than 10 g between ages 65 and 69 and 80 and 84 years for both men and women in this healthy subgroup). Therefore, for simplicity, the reference equations were not expressed as a function of age. There was no evidence of an interaction between sex and weight. Since in the CHS cohort race was not a significantly independent predictor of LV mass, the reference equations are not race specific.
Based on these reference equations, expected log-transformed LV mass values and the difference between observed and expected values were computed. The 5th and 95th percentiles of the distribution of observed minus expected values were used to define LV mass measurements that were lower or higher, respectively, than expected after adjustment for body weight. Because of the large sample size, the variability in the estimated regression coefficients will be small relative to the variability in LV mass, so it may be ignored in this case.
For ease of application, the reference equations and percentiles have been converted to the linear scale. The reference equations are presented to allow the calculation of the ratio observed to expected LV mass; the percentiles can then be used to determine whether a patient’s LV mass is either below the lower value or above the upper value expected among older people free of clinical heart disease or hypertension. The specificity of our prediction procedure is fixed (at 90%), and the performance of the procedure was examined in participants with clinical CHD to provide insight into its sensitivity.
LV Ejection Fraction and Segmental Wall Motion
“Normal” and “borderline” LV ejection fraction categories were considered “normal” for the purposes of these analyses. LV segmental wall motion was reported as (1) normal if all visualized segments showed normal wall motion, (2) hypokinetic if at least one segment was hypokinetic (but none was akinetic or dyskinetic), and (3) akinetic/dyskinetic if absent or paradoxical systolic motion was noted in one or more segments. Participants with fewer than 12 successfully scored segments (n=388) were excluded from this analysis. χ2 tests for association (for age, sex, and disease status) and t tests (for body weight and height) were used to examine bivariate relations with LV ejection fraction and wall motion abnormalities. Logistic regression was used to determine independent predictors of abnormalities. Because of the low prevalence of abnormalities (n=7 for ejection fraction and n=23 for wall motion [19 with hypokinesis and 4 with akinesis or dyskinesis]) in participants with neither clinical heart disease nor hypertension, the effects of age, sex, diabetes, and obesity were not examined in this group.
Reader Agreement for LV Ejection Fraction and Segmental Wall Motion Scoring
To assess interreader and intrareader agreement for LV ejection fraction and segmental wall motion measurements, κ statistics and the percentage of agreement were calculated. For LV ejection fraction, 339 paired studies were read for assessment of interreader agreement and 310 paired studies for intrareader agreement. In addition, 477 and 239 paired studies were read for assessment of interreader and intrareader agreement, respectively, in scoring LV segmental wall motion.
M-Mode LV Mass
Missing M-Mode Echo LV Mass Measurements and Their Correlates
M-Mode measurements of LV mass could not be made in 34% of CHS participants. There was a progressive increase in the percentage of missing data for LV mass with increasing age, ranging from 29% in the 65 to 69 year age group to 50% in the 85+ year age group (P<.001). Body weight and height were significantly related to the ability to obtain an LV mass measurement, with smaller participants more likely to have a measurement. The percentage of participants with missing data for LV mass was also significantly higher among whites than nonwhites (35% versus 27%, P<.01); men than women (40% versus 31%, P<.001); and those with compared with those without a history of CHD (37% versus 34%, P<.04), a self-reported history of hypertension (37% versus 33%, P<.01), or diabetes (40% versus 34%, P<.01).
Reader Agreement for LV Mass Measurements
Overall mean and median interreader variabilities (percent measurement differences) for LV mass measurements were 17% and 14%, respectively. Measurement variabilities were similar for ventricular septal and LV posterior wall thicknesses. In contrast, overall interreader and intrareader mean percent differences for LV internal dimension (diastole) were much lower (4% to 6%) (see Table 2⇓).
Relations to Weight, Height, Body Surface Area, and Obesity
After adjustment for age, each of the anthropometric variables—weight, height, and body surface area—examined separately, was significantly related to LV mass in both men and women. However, after adjustment for age and weight, neither height nor body surface area was significantly related to LV mass. Because of the high correlation (.95 in men and .94 in women) between weight and body surface area and because weight explained slightly more of the variability in LV mass (R2 for weight, .17 for women and .08 for men; R2 for body surface area, .16 for women and .07 for men), weight was used as the measure of body size in the remaining analyses.
As shown in Table 1⇑, there was a significantly higher prevalence of obesity in participants with clinical CHD (41.7% in women and 31.1% in men) or hypertension (40.1% in women and 36.6% in men) than in participants with neither clinical heart disease nor hypertension (27.7% in women and 23.4% in men). Diabetes was also more prevalent in the clinical CHD and hypertension groups than in participants with neither clinical heart disease nor hypertension. Furthermore, it is interesting to note (Table 3⇓) that there was little difference in weight-adjusted LV mass between obese and nonobese participants with neither CHD nor hypertension, suggesting that these obese participants without CHD or hypertension do not have “undersized” hearts for their body size. In contrast, among participants with clinical CHD, obese participants demonstrated much smaller weight-adjusted LV mass (140.8 g in women and 179.3 g in men) than did the nonobese participants (148.3 g in women and 195.4 g in men), suggesting that their hearts were relatively “undersized” for their body size. The relation is more complicated among the hypertension group, with the findings differing by sex.
Relation to Age, Sex, and Prevalent Clinical Disease Status
Means and standard deviations for unadjusted LV mass by age, sex, race, and disease status group are presented in Table 4⇓. Weight-adjusted mean LV mass data are presented in the Figure⇓ by age, sex, and disease status group. Weight-adjusted LV mass was significantly higher in men (P<.0001) and in participants with clinical CHD (P<.0001). Disease status was also significantly related to weight-adjusted LV mass (P<.0001), with nonobese participants with CHD having significantly higher LV mass than participants with hypertension or participants with neither clinical heart disease nor hypertension (Figure⇓). Likewise, participants with hypertension had significantly higher weight-adjusted LV mass than participants with neither clinical heart disease nor hypertension. Of interest, within each of the four age groups, the magnitude of the sex effect exceeded that of the effect of clinical CHD. LV mass increased with age (P<.0001) but, as seen in the Figure⇓, after adjustment for body weight the age effect was small, with weight-adjusted LV mass increasing less than one gram per year increase in age for each sex-disease status group. There was no evidence that age, sex, and disease status had a nonadditive relation with LV mass.
Reference Equations for LV Mass in the Healthy Subgroup
The expected LV mass (in grams) can be calculated from the following equations:
If the ratio of observed to expected LV mass is between 0.69 and 1.47, the patient’s LV mass should not be considered larger than expected given his or her weight. This prediction procedure classified 28% of the men and 18% of the women with clinical CHD as outside the range of expected LV mass measurements. Exclusion of obese (n=220) participants from this healthy group had a negligible effect on the reference equations and their performance on participants with clinical CHD (classifying 16% of women and 26% of men as abnormal), so they were not excluded in order to maximize the size of this subgroup.
Examination of Participants With Neither Clinical Heart Disease Nor Hypertension (Including Relation of Diabetes and Obesity)
Mean weight-adjusted LV mass did not differ between diabetics (135.3 g) and nondiabetics (135.8 g). After adjustment for weight, sex, and age, diabetes remained unrelated to LV mass (adjusted means of 137 g in diabetics and 139 g in nondiabetics, P=.72). Furthermore, diabetes did not modify the effects of weight, sex, or age on LV mass. There was little difference in mean weight-adjusted LV mass between obese and nonobese participants with neither clinical heart disease nor hypertension (Table 3⇑). Obesity did not modify the effects of weight, sex, or age on LV mass.
2-D LV Ejection Fraction and Segmental Wall Motion
LV Ejection Fraction
LV ejection fraction could be scored qualitatively in 99% of the CHS cohort. The estimated prevalence of LV ejection fraction abnormalities by age, sex, and disease status group is presented in Table 5⇓. The relation of the prevalence of LV ejection fraction abnormalities to sex, age, and disease status is shown in Table 6⇓. The prevalence of abnormalities was significantly higher in men (P<.001) and increased with age (P<.001). The prevalence of abnormalities was significantly higher in participants with clinical CHD (10.5%) than in participants with hypertension (1.7%) or in participants with neither clinical heart disease nor hypertension (0.5%, both P<.001). The prevalence of abnormalities was also significantly higher in participants with hypertension than in participants with neither clinical heart disease nor hypertension (P<.01). Both body weight and height were significantly related to the prevalence of LV ejection fraction abnormalities, with larger participants having more abnormalities (both P<.001). Based on multivariate analyses, sex (P<.0001), clinical CHD (P<.0001), hypertension (P<.01), and age (P<.02) were the only independent predictors of LV ejection fraction abnormalities. There was no evidence of a nonadditive association between LV ejection fraction abnormalities and age, sex, and disease status.
LV Segmental Wall Motion
LV segmental wall motion could be scored in 93% of the CHS cohort. The prevalence of hypokinesis and akinesis/dyskinesis was examined separately and its distribution is presented in Table 7⇓. The relation of sex, age, and disease status to the prevalence of wall motion abnormalities is examined in Table 8⇓. LV wall motion abnormalities were more prevalent among men than women (P<.001 for both hypokinesis and akinesis/dyskinesis). The prevalence of hypokinesis increased with age (P<.001), whereas the prevalence of akinesis/dyskinesis was only weakly associated with age (P=.04). The prevalence of wall motion abnormalities was significantly higher among participants with CHD than among participants without CHD. Likewise, participants with hypertension (in the absence of CHD) had a significantly higher prevalence of wall motion abnormalities than participants with neither clinical heart disease nor hypertension. Height was significantly related (P<.001 for both hypokinesis and akinesis/dyskinesis), and body weight was only modestly related (P=.02 for hypokinesis and P=.04 for akinesis/dyskinesis) to the prevalence of LV wall motion abnormalities. For both height and weight, participants with LV wall motion abnormalities were larger than participants without abnormalities. Age, sex, CHD, and hypertension were the only independent predictors of hypokinetic abnormalities (all P<.01), whereas sex and CHD were the only independent predictors of akinetic or dyskinetic abnormalities (both P<.001). There was no evidence of a nonadditive relation between the independent predictors and LV wall motion abnormalities.
Reader Agreement for LV Ejection Fraction and Segmental Wall Motion
There was agreement on 94% (κ=0.32) of the paired interreader studies for LV ejection fraction, and the agreement for LV segmental wall motion was 92% (κ=0.45). The agreement was better for the paired intrareader studies (98% for LV ejection fraction and 95% for segmental wall motion, with κ=0.82 and 0.59, respectively).
This study demonstrated strong cross-sectional relations of male sex, hypertension, and prevalent CHD to LV mass (by M-mode echo) and to global and segmental LV wall motion (by 2-D echo) in the elderly CHS participants. Age was modestly related to these echo findings among people older than 65 years of age. CHS is one of the largest population-based epidemiological studies of the elderly, having recruited and examined from multiple communities a population-based sample of more than 5000 men and women older than 65 years. This population included predominantly white, ambulatory participants.
M-Mode LV Mass
Missing M-Mode Echo LV Mass Data
The proportion of CHS participants (66% in the overall cohort and 50% older than age 85) in whom LV mass was measurable was comparable to that reported in comparably aged subjects in the Framingham Study. Specifically, in Framingham, M-mode measurements necessary to calculate LV mass could be made in 67% of men and 72% of women in the 60 to 69 year age group but in only 37% of men and 48% of women older than 80 years.13 In contrast, LV mass could be measured in 93% of men and 96% of women in the Framingham 40 to 49 year group. The inability to measure LV mass from the M-mode echo in CHS was higher in (1) older subjects and in men, as in Framingham, (2) heavier and taller subjects, and (3) those with a self-reported history of CHD, hypertension, and diabetes. Because LV mass is known to be higher in men with CHD and hypertension, LV mass measurements reported in this article probably underestimate the true mean LV mass in the entire CHS cohort.
Reader Measurement Agreement
Interreader and intrareader measurement variabilities for LV mass measurements in CHS reflect the fundamental difficulties involved in identifying the echo interfaces of the ventricular septum and LV posterior wall and measuring the thickness of these structures. In contrast, interreader and intrareader mean percent measurement differences for LV internal dimension (diastole) were in a much lower range (4% to 6%). In comparison, previous reports in small non–population-based studies with younger subjects have reported interreader measurement variability to be 16% for M-mode posterior LV wall thickness and 27% for 2-D LV mass.14 15
Relation to Body Size and Obesity
We found body weight to best account for the dependence of LV mass on body size. Other studies1 2 have used height to adjust LV mass for body size, but height was not as good a predictor of LV mass as were weight and body surface area in the CHS age group. Weight and body surface area were highly correlated (r=.94 to .95) and yielded similar results; therefore, we chose the simpler measure, weight, without any loss of precision. After adjustment for age and weight, neither height nor body surface area was significantly related to LV mass.
With respect to the prevalence of obesity, there was a gradient related to disease status group, with participants with clinical CHD having the highest prevalence (41.7% in women and 31.1% in men) and participants with neither clinical heart disease nor hypertension having the lowest prevalence (27.7% in women and 23.4% in men). Furthermore, it was of interest that among participants with clinical CHD, women and men who were categorized as obese had significantly smaller mean weight-adjusted LV mass versus women and men who were not obese. This suggests that the participants with clinical CHD had hearts that were “undersized” for their body size.
This phenomenon has been previously described by Devereux et al,16 who have suggested that lean body mass rather than weight is a more appropriate measure with which to adjust LV mass. In our cohort in nonobese subjects, weight appeared to explain more of the variability in LV mass than did height. However, it is beyond the scope of this article to address in detail the complex relations between body size measures and LV mass in obese subjects. A more complete analysis of this topic is the subject of another CHS manuscript currently in preparation. Of interest, in participants in this study with neither CHD nor hypertension, there was little difference in weight-adjusted LV mass between obese and nonobese participants.
The reference equations presented here easily allow a determination to be made as to whether the LV mass of an elderly subject is beyond that expected for his or her weight. Similar equations based on large numbers of elderly subjects sampled from defined populations are not currently available in the literature. Reference equations are available using body surface area for individuals older than 70 years.17 18 19 However, the prior reference equations were developed on a very small sample of subjects older than 70 years (n=12), which required several unverifiable assumptions. The reference equations given here are an improvement over what is currently available in the literature and should provide useful standards for clinical use.
Relation to Age
Among CHS participants, weight-adjusted LV mass increased minimally with increasing age. The minimal effect of age on LV mass may be due to the highly truncated age range, ie, no subjects were younger than 65 years. However, in a healthy subgroup of 862 participants aged 18 to 79 years from the Framingham cohort (n=177 subjects) and offspring (n=685 subjects), only minimal changes in LV mass were noted with advancing age. Multivariate analyses in this normotensive, nonobese subgroup with no clinical evidence of cardiovascular disease revealed that age was not significantly associated with LV mass in men and only weakly associated with LV mass in women.20 In Framingham, ASE LV measurements were used in conjunction with the Troy LV mass formula, whereas in the current study ASE LV measurements were used with the necropsy-validated LV mass formula reported by Devereux et al9 for use with ASE measurements. The presence of modest differences is not surprising in view of the differences in age between the two cohorts (mean age±SD in Framingham, 42±12 years) and the exclusion in Framingham of all subjects who were greater than 20% above ideal weight of the Metropolitan Life Insurance Co tables.21
Gardin et al17 and Henry et al18 have reported in 136 adults without clinically apparent heart disease that LV mass adjusted for body surface area increases approximately 15% from the 21 to 30 year age group to the older than 70 age group. However, Devereux et al16 reported in 225 subjects (age range, 18 to 72 years) that when LV mass was indexed for body surface area or lean body mass, there was no significant relation of age to LV mass.
Relation to Sex
The CHS cohort demonstrated significantly smaller measurements in women than in men for LV mass across all three disease groups studied. These sex differences in mean weight-adjusted LV mass were greater in magnitude than the differences detected between the group with neither clinical heart disease nor hypertension and the CHD group. The smaller LV mass differences dictated between disease status than sex groups are probably related to the high prevalence of subclinical disease in elderly individuals (documented by necropsy) with neither clinical heart disease nor hypertension. In smaller studies of selected populations, sex differences have not been consistently observed. In one study of adult younger and older subjects without clinically apparent heart disease, women exhibited a slightly smaller LV mass than men for any given age and body surface area (range of age group–related mean differences, 3.6% to 7.2%).17 19 In a second report, although indexing for body surface area did not eliminate sex differences for LV mass, indexing for lean body mass eliminated these differences.16
Relation to Prevalent Disease
As expected, LV mass was greatest in participants with CHD, followed by those with hypertension, and was least in those with neither clinical heart disease nor hypertension. However, within each of the four age groups, the magnitude of the sex effect exceeded that of the effect of clinical CHD.
In the Framingham substudy of healthy participants, there was no significant relation between LV mass and systolic BP in either sex.20 However, Lauer and associates22 demonstrated that hypertension and obesity each had significant independent associations with LV mass and wall thickness (all P<.001 in men and women) and that the strengths of the association were additive.
The Framingham Study has reported that M-mode echo–determined LV hypertrophy is an independent predictor of mortality and morbidity from CHD.1 2 Longitudinal follow-up of the CHS cohort should provide data regarding the relation of LV mass (from M-mode echo) to subsequent mortality and morbidity related to CHD and stroke in men and women aged 65 years and older.
2-D Echo Variables
Reader Measurement Agreement
Intrareader agreement was good to excellent for qualitative assessments of LV ejection fraction (κ=0.83) and semiquantitative assessments of LV segmental wall motion (κ=0.59). However, interreader agreement was only fair for LV ejection fraction (κ=0.32) and LV segmental wall motion (κ=0.45).
Relation to Age and Sex
There was a modest but significant relation between age and the baseline prevalence of abnormalities of LV global or segmental wall motion. In multivariate analyses, the prevalence of LV ejection fraction abnormalities and hypokinetic but not akinetic and dyskinetic LV segmental wall motion increased modestly and independently with age. This is not surprising given the known increase in the prevalence of CHD with increasing age.23
In all disease status subgroups, men had a significantly higher prevalence at baseline of both LV ejection fraction and segmental wall motion abnormalities (primarily hypokinesis). Multivariate analyses revealed male sex to be strongly associated (all P<.0001) with abnormal LV ejection fraction, hypokinesis, and akinesis or dyskinesis. The higher prevalence of abnormalities of LV ejection fraction and segmental wall motion in men is consistent with the known increased risk for CHD conferred by male sex. A previous report from CHS has emphasized the greater prevalence and burden related to clinical CHD in men compared with women.11
Relation to Body Size and Disease Status
Height and weight were both related directly to the prevalence of LV ejection fraction and wall motion abnormalities; the reason for these relations is not readily apparent.
As expected, the clinical CHD group demonstrated the highest prevalence of abnormalities of both LV ejection fraction and LV segmental wall motion, followed by the hypertensive group, then by the group with neither clinical heart disease nor hypertension. Interestingly, in the group with neither clinical heart disease nor hypertension, 0.5% of participants demonstrated abnormalities of LV ejection fraction (global systolic function), and 1.9% of participants demonstrated abnormalities of LV segmental wall motion (1.6% hypokinesis and 0.3% akinesis or dyskinesis). These data are among the most interesting baseline CHS echo findings. On the other hand, it should not be so surprising that elderly subjects with neither clinical heart disease nor hypertension have 2-D echo evidence suggestive of CHD in the elderly, because previous studies have suggested that the autopsy prevalence of CHD approaches 60% to 70% compared with a clinical prevalence of only 20% to 30%.23
The presence of LV segmental wall motion abnormalities in the group with neither clinical heart disease nor hypertension may reflect previously undetected myocardial infarction, “stunned” or “hibernating” (but noninfarcted) myocardium, severe ischemia due to CHD, or a myopathic process (related or unrelated to CHD).24 Lewis and associates24 have noted that LV segmental wall motion abnormalities may be detected by 2-D echo in up to one third of patients without documented myocardial infarction who are evaluated for suspected CHD. In this predominantly symptomatic group, LV segmental wall motion abnormalities were associated with a high likelihood of multivessel disease at cardiac catheterization, as well as with significant narrowing of the artery supplying the region demonstrating abnormal wall motion. When LV segmental wall motion abnormalities improve after revascularization, the presumption is that these areas represented regions of hibernating myocardium. In their series, Lewis et al report that the finding of an LV segmental wall motion abnormality was specific for the presence of significant coronary artery disease in 86% of cases and was highly specific (94%) for the presence of any cardiac disease.
In conclusion, this study has demonstrated potentially important cross-sectional relations of age, sex, and baseline disease status to LV mass (from M-mode echo) and/or global and segmental LV wall motion (from 2-D echo) in the population-based CHS cohort of elderly adults. LV mass adjusted for body weight increased minimally with increasing age but was substantially greater in men and in participants with hypertension or clinical CHD than in those with neither clinical heart disease nor hypertension. Across all CHS age subgroups, the difference in mean weight-adjusted LV mass associated with sex was greater in magnitude than the difference associated with disease status. LV segmental wall motion abnormalities were more prevalent in men and in participants with clinical CHD or hypertension and were also detectable in a small proportion of elderly individuals in the absence of these disease entities.
Because of the relatively small number of nonwhites recruited into the CHS baseline cohort (approximately 7%), the data presented in this article are applicable primarily to white men and women. Efforts are currently under way to study, during the second echo examination (CHS year 7: 1994-1995), an additional 600 black participants recruited into the CHS cohort. Longitudinal follow-up should define the predictive value of these LV measurements for morbidity and mortality from CHD and stroke in this elderly population.
This work was supported by contract Nos. N01-HC85079 through HC-85086 from the National Heart, Lung, and Blood Institute, Bethesda, Md.
Participating institutions and principal staff of the Cardiovascular Health Study:
Forsyth County, NC—Bowman Gray School of Medicine of Wake Forest University: Gregory L. Burke, Marie E. Cody, R. Gale Cruise, Walter H. Ettinger, Curt D. Furberg, Gerardo Heiss, H. Sidney Klopfenstein, David S. Lefkowitz, Mary F. Lyles, Maurice B. Mittelmark, Grethe S. Tell, James F. Tool.
Sacramento County, Calif—University of California, Davis: William Bommer, Marshall Lee, John Robbins, Marc Schenker.
Washington County, Md—The Johns Hopkins University: R. Nick Bryan, Trudy L. Bush, Joyce Chabot, George W. Comstock, Linda P. Fried, Pearl S. German, Joel Hill, Steven J. Kittner, Shiriki Kumanyika, Neil R. Powe, Thomas R. Price, Robert Rock, Moyses Szklo.
Allegheny County, Pa—University of Pittsburgh: Janet Bonk, Julie Thompson-Dobkin, Diane G. Ives, Charles A. Jungreis, Lewis H. Kuller, Robert H. McDonald, Jr, Elaine Meilahn, Peg Meyer, Anne Newman, Gale H. Rutan, Richard Schulz, Vivienne E. Smith, Sidney K. Wolfson.
Echocardiography Reading Center—University of California, Irvine: Hoda Anton-Culver, Julius M. Gardin, Margaret Knoll, Tom Kurosaki, Nathan Wong.
Ultrasound Reading Center—New England Deaconess Hospital: Daniel H. O’Leary, Joseph F. Polak, Jeffrey Potter.
Central Blood Analysis Laboratory—University of Vermont: Edwin Bovill, Elaine Cornell, Paula Howard, Russell P. Tracy.
Pulmonary Function Reading Center—Mayo Clinic and Foundation: Paul Enright, Sheila Toogood.
Electrocardiography Reading Center—University of Alberta, Edmonton: Kris Calhoun, Harry Calhoun, Patty Montague, Farida Rautaharju, Pentti Rautaharju.
Coordinating Center—University of Washington, Seattle: Nemat O. Borhani, Annette L. Fitzpatrick, Bonnie K. Hermanson, Richard A. Kronmal, Bruce M. Psaty, David S. Siscovick, Lynn Shemanski, Patricia W. Wahl.
NHLBI Project Office: Diane E. Bild, Teri A. Manolio, Peter J. Savage, Patricia Smith.
Reprint requests to CHS Coordinating Center, Century Square, 1501 Fourth Ave, Suite 2025, Seattle, WA 98101.
- Received October 5, 1994.
- Accepted October 30, 1994.
- Copyright © 1995 by American Heart Association
Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP. Left ventricular mass and incidence of coronary heart disease in an elderly cohort: The Framingham Heart Study. Ann Intern Med. 1989;110:101-107.
Taylor GJ, Humphries JO, Mellits ED, Pitt B, Schulze RA, Griffith LS, Achuff SC. Predictors of clinical course, coronary anatomy and left ventricular function after recovery from acute myocardial infarction. Circulation. 1980;62:960-970.
Fried L, Borhani N, Enright P, Furberg C, Gardin J, Kronmal R, Kuller L, Manolio T, Mittelmark M, Newman A, O’Leary D, Psaty B, Rautaharju P, Tracy R, Weiler P, for the CHS Collaborative Research Group. Cardiovascular Health Study: design and rationale. Ann Epidemiol. 1991;1:263-276.
Tell G, Fried L, Hermanson B, Manolio T, Newman A, Borhani N, for the Cardiovascular Health Study Collaborative Research Group. Recruitment of adults 65 years and older as participants in The Cardiovascular Health Study. Ann Epidemiol. In Press.
Gardin JM, Wong ND, Bommer W, Klopfenstein HS, Smith V-E, Tabatznik B, Siscovick D, Lobodzinski S, Anton-Culver H, Manolio TA. Echocardiographic design of a multi-center investigation of free-living elderly subjects: the Cardiovascular Health Study. J Am Soc Echocardiogr. 1992;5:63-72.
Sahn DJ, DeMaria A, Kisslo J, Weyman A, for the Committee on M-mode Standardization of the American Society of Echocardiography. Recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic methods. Circulation. 1978;58:1072-1083.
Henry WL, DeMaria A, Feigenbaum H, Kerber R, Kisslo J, Weyman AE, Nanda N, Popp RL, Sahn D, Schiller NB, Tajik AJ. Report of the American Society of Echocardiography Committee on Nomenclature and Standards: Identification of Myocardial Wall Segments. Raleigh, NC: American Society of Echocardiography; 1982.
Mittelmark MD, Psaty BM, Rautaharju P, Fried LP, Tracy RP, Gardin JM, Borhani NO, O’Leary DH, for the CHS Collaborative Research Group. Prevalence of cardiovascular diseases among older adults: The Cardiovascular Health Study. Am J Epidemiol. 1993;137:311-317.
de Simone G, Daniels SR, Devereux R, Meyer R, Roman MJ, de Divitiis O, Alderman M. 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: 1251-1260.
Savage DD, Garrison RJ, Kannel WB, Anderson SJ, Feinleib M, Castelli WP. Considerations in the use of echocardiography in epidemiology: The Framingham Study. Hypertension. 1987;9(suppl II):II-40-II-44.
Kuecherer HF, Kee LL, Modin G, Cheitlin MO, Schiller NB. Echocardiography in serial evaluation of left ventricular systolic and diastolic function: importance of image acquisition, quantitation, and physiologic variability in clinical and investigational applications. J Am Soc Echocardiogr. 1991;4:203-214.
Henry WL, Gardin JM, Ware JH. Echocardiographic measurements in normal subjects from infancy to old age. Circulation. 1980;62:1054-1061.
Gardin JM, Savage DD, Ware JH, Henry WL. Effect of age, sex and body surface area on echocardiographic left ventricular wall mass in normal subjects. Hypertension. 1987;9(suppl II):II-36-II-39.
New weight standards for men and women. Stat Bull Metrop Insur Co. 1959;40:1-4.
Elveback L, Lie JT. Continued high incidence of coronary artery disease at autopsy in Olmsted County, Minnesota, 1950 to 1979. Circulation. 1984;70:345-349.