Effect of Body Size, Ponderosity, and Blood Pressure on Left Ventricular Growth in Children and Young Adults in the Bogalusa Heart Study
Background The measurement of left ventricular mass (LVM) is important because individuals with increased LVM are at increased risk for cardiovascular diseases, including myocardial infarction and congestive heart failure. There are limited longitudinal data on the acquisition of LVM in children and young adults and the relative importance of sex, growth, excess body weight, and blood pressure (BP) on change in LVM.
Methods and Results The study cohort consisted of a cross section of 160 healthy children and young adults 9 to 22 years of age at first exam in the biracial community of Bogalusa, La. All had stable BP levels recorded over a 2- to 3-year period. Repeated examinations were performed 4 to 5 years apart. At each exam, 6 BPs were obtained with a mercury sphygmomanometer by trained examiners. The mean of the observations was used, with the fourth Korotkoff phase serving as the measure of diastolic BP. Anthropometric data, including height (HT), weight (WT), and triceps skin fold thickness (TSF), were also obtained, and M-mode echocardiograms were performed. Ponderal index (PI=WT/HT3) was used as a measure of weight-for-height. Tracking of HT (r=.68 to .76), WT (r=.73 to .82), PI (r=.77 to .89), TSF (r=.70 to .80), BP (r=.47 to .60), and LVM (r=.40 to .70) was strong in both sexes (P<.0001). LVM indexed for linear growth (LVM/HT2.7) tracked in females (r=.56, P<.0001) but not in males. In univariate cross-sectional analyses, LVM/HT2.7 correlated with WT, PI, and TSF in both sexes (r=.21 to .60, P<.05) and with systolic BP (SBP) in females (r=.23, P<.05). WT was the only independent correlate of LVM/HT2.7 in both sexes in multivariate cross-sectional analysis in a model containing age, SBP, WT, and TSF as independent variables (r2=.08 to .28, P<.02). In longitudinal univariate analyses, initial measurements of WT, PI, and TSF predicted final LVM/HT2.7 in both sexes (r=.28 to .56, P<.01), and SBP was significant for females (r=.27, P<.05). In multivariate analyses, initial WT was associated with final LVM and LVM/HT2.7 in both sexes (r2=.27 to .54, P<.01). Finally, baseline LVM correlated with final SBP in both sexes (r=.21 to .27, P<.05), and initial LVM/HT2.7 correlated with final SBP in females (r=.26, P<.05) with a trend for males (r=.17).
Conclusions These data indicate that linear growth is the major determinant of cardiac growth in children and that excess weight may lead to the acquisition of LVM beyond that expected from normal growth. Increased mass may also precede the development of increased BP. The development of obesity may therefore be a significant, and possibly modifiable, risk factor for developing left ventricular hypertrophy and hypertension, risk factors for cardiovascular morbidity and mortality.
The measurement of left ventricular mass (LVM) has increased in importance in recent years because individuals with left ventricular hypertrophy are at increased risk for cardiovascular diseases, including myocardial infarction and congestive heart failure.1 2 Because mass can be measured noninvasively with echocardiography, it is now possible to study longitudinally the factors predisposing an individual to increased LVM.
Of critical importance is the differentiation of pathological versus appropriate increases in LVM. The left ventricle hypertrophies to normalize left ventricular wall stress and maintain efficient ventricular systolic function.3 LVM thus normally increases with growth4 5 and physical conditioning6 and increases pathologically in certain diseases that create a pressure load on the left ventricle, such as aortic stenosis and hypertension.7 Recent investigations in healthy populations using cross-sectional study designs showed that measures of body size, obesity, and blood pressure are the major determinants of LVM.4 8 9 10 De Simone et al11 reviewed data from cross-sectional studies of adults and children in an effort to adjust LVM for height, body surface area (BSA), and other indexes so that the effects of weight and blood pressure independent of height and growth can be better assessed.
Few longitudinal studies of the acquisition of LVM in children and young adults have been made. To assess the relative importance of sex, growth, excess body weight, and blood pressure on change in LVM, M-mode echocardiograms were performed 4 to 5 years apart in a healthy cross section of children and young adults. Specifically assessed was the relative importance of baseline versus follow-up measures to mass at follow-up and change in mass.
The Bogalusa Heart Study is a long-term epidemiological study of cardiovascular disease risk factors from birth through early adulthood. The study sample was selected from the children residing in ward 4 of Washington Parish, Louisiana, which includes the city of Bogalusa. This biracial community of approximately 22 000 people is two-thirds white and one-third black. Four cross-sectional surveys of school-age children were conducted between 1973 and 1982. Blood pressure, laboratory, family history–health habit questionnaire, and anthropometric data were collected on all participants, as previously defined.12
Children with stable blood pressure levels over a 2- to 3-year period were selected for the echocardiographic study. Specifically, those children whose height-, race-, and sex-specific diastolic blood pressure rank remained within 10% from either the second (1976 to 1977) to the third (1978 to 1979) or from the third (1978 to 1979) to the fourth (1981 to 1982) school survey were eligible to participate in the first Bogalusa Echocardiography Study where baseline measurements were obtained.12 A follow-up echocardiogram was obtained approximately 4 years later during the next school survey on as many of the original participants as possible (42%). Follow-up blood pressure, laboratory, questionnaire, and anthropometric data were also collected. For both studies, approval from the committee on use of human subjects was obtained, and all subjects or their guardians gave informed consent.
The mean age of the 160 participants at the time of the first examination was 13.3 years (range, 9 to 22 years). The children were mostly between the ages of 9 and 16 years (81%). Slightly more than half were male (56%); approximately two thirds of the participants were white (63%), and one third was black (37%). The race and sex distributions reflect the composition of the general community. The follow-up echocardiograms were performed 4.2 to 4.5 years later.
The study cohort was compared with individuals examined during the third Bogalusa school screening (1978 to 1979). There were no significant differences in race-sex composition, blood pressure, or anthropometric measurements (data not shown).
Blood Pressure and Anthropometric Measures
The screening protocol began with measurement of the right upper arm length to the nearest 1/10th cm with an anthropometric caliper. This value was divided in two, and the halfway point was marked. The right upper arm circumference was measured at the mark to the nearest 1/10th cm with a cloth tape measure. Proper blood pressure cuff size was selected from a table listing cuff size as a function of right upper arm length versus midarm circumference as developed in previously published protocols.12 Trained examiners then obtained blood pressures on seated, relaxed subjects. Six blood pressure values were recorded for each participant with mercury sphygmomanometers (W.A. Baum Co, Inc). Blood pressures were recorded as the first, fourth, and fifth Korotkoff phases. The fourth Korotkoff phase was used as the measure of diastolic blood pressure. The mean of the six blood pressures was used in the analyses. Participants were then randomly assigned to each of the three remaining anthropometric measurements. Height was measured to the nearest 0.1 cm and weight to the nearest 0.1 kg. The mean of two measurements for both height and weight was used in all analyses. Ponderal index (PI, weight divided by height cubed) was used as a measure of weight-for-height. Triceps skin fold thickness was measured to the nearest 1.0 mm with Lange skin fold calipers (Cambridge Scientific Industries), and the mean of three measurements was used.
Baseline M-mode echocardiographic examinations of the left ventricle were performed by standard techniques with participants in the supine position by use of a System 2 instrument (Irex Medical Systems Inc) and recorded on strip charts as previously described.12 Follow-up assessments were performed with two-dimensional guided M-mode echocardiograms (Hewlett-Packard or Toshiba echocardiography machines) by trained technical personnel using 2.25- and 3.5- MHz transducers. Data were recorded on standard VHS videocassette tapes. Baseline echocardiograms were measured at Children’s Memorial Hospital in Chicago; follow-up echocardiograms were read at Tulane University Medical Center in New Orleans, La. All echocardiograms were digitized and measured on Freeland Cardiology Workstation digitizing systems (Freeland Systems).
Echocardiogram quality was assessed by use of the criteria of Schieken et al.13 Briefly, these criteria require the presence of a single dominant line on the area of the M-mode tracing to be measured, continuity of the line for 5 mm, and an interface with motion characteristics for the specific cardiac structure being imaged.13 All studies were measured by at least one of three individuals (E.M.U., S.S.G., or K.B.) according to American Society for Echocardiography guidelines.14 The coefficient of variation for interreader and intrareader variability for all measures of cardiac anatomy was less than or equal to 10%, comparable to previously published studies.13 15 16 17 Standard formulas18 19 were used to calculate left ventricular minor axis shortening and LVM:
where SF=left ventricular minor axis shortening fraction, LVEDd=left ventricular end dimension in diastole, LVEDs=left ventricular end dimension in systole, IVSd=interventricular septum dimension in diastole, and LVPWd=left ventricular posterior wall dimension in diastole.
The statistical analysis system was used for the analyses.20 Descriptive statistics were used to describe the anthropometric, blood pressure, and echocardiographic characteristics of the study cohort. Sex, race differences, and race-by-sex interaction were examined by an ANOVA. Significant sex but not race differences were seen; therefore, analyses were done in a sex- but not race-specific manner. All study variables were not normally distributed; therefore, Spearman correlations were calculated to evaluate the associations between baseline and follow-up measurements.
PI is used as the measure of obesity because as a three-dimensional body increases in size with its configuration and density held constant, it will maintain a constant weight divided by height cubed. Voors et al21 demonstrated this concept by showing that when this index of weight divided by height cubed is used, the independent contribution of height to the variability of percent body fat is less than when weight divided by height squared or weight divided by height is used. The correlations, however, were repeated with BSA. These correlations are not presented because the results for PI were consistently equal to or stronger than those for BSA. This is not surprising because as McMahon22 discussed, the human body and measures of its weight are three-dimensional, but BSA is only two-dimensional and height is one-dimensional.
To account for linear growth primarily, all analyses were repeated for LVM adjusted by height2.7 (LVM/HT2.7, the method of De Simone et al11 ). Data on both LVM unadjusted and LVM indexed for growth are presented. Additional analyses sought correlations with the sum of the left ventricular septal wall and posterior wall in diastole (h) and the ratio of h to left ventricular end-diastolic dimension in diastole. The directions of these correlations were similar but less strongly significant than those for LVM; therefore, only results for mass are presented. A stepwise regression analysis was also used to study the predictability of follow-up LVM. The follow-up LVM was the dependent variable. Independent variables included LVM, weight, triceps skin fold thickness, and systolic blood pressure at baseline and change of weight from baseline to follow-up. Colinearity was examined during this analysis.
Comparison of Baseline and Follow-up Values
Baseline and follow-up values for age, anthropometric measures, blood pressure, and echocardiographic data are listed in Table 1⇓ for boys and in Table 2⇓ for girls. Baseline sex comparisons showed that males were 1.9 cm taller and 1.1 kg heavier and had a triceps skin fold thickness 3.3 mm less than females did (P<.02). For all echocardiographic values except left ventricular posterior wall dimension in systole and shortening fraction, males and females were different. Males had larger chamber dimensions, thicker walls (all P<.02 or lower), and higher mass (87.5 versus 71.9 g, all P<.02). At follow-up, males were 11 cm taller and weighed 8 kg more (P<.001). Females had a higher PI (1.1 kg/m3) and a thicker triceps skin fold thickness (7.4 mm, P<.02). Males continued to have larger chamber dimensions, wall thicknesses, and LVM compared with females (129 versus 92 g, P<.001). There were no sex differences with regard to systolic or diastolic blood pressure at either examination. All baseline and follow-up anthropometric and blood pressure values were within normal limits for age, and echocardiographic measurements were normal compared with published parameters.4 5 11
The average time between examinations was 4.2 years for males and 4.5 years for females. Males grew 9.9 cm more in height and 6.7 kg more in weight than females. The PI of males did not differ significantly between exams, while the PI for females increased by 1.23 kg/m3. Males’ triceps skin fold thickness decreased, whereas that for females increased. Systolic and diastolic blood pressures increased similarly in both males and females. For echocardiographic measures, chamber dimensions and wall thicknesses tended to increase; however, these measurements increased in females to a smaller degree. For both sexes, shortening fraction decreased but remained within normal limits. Mass increased by 42.3 g in males and 20.5 g in females.
Tracking coefficients for anthropometric measures, blood pressure, and LVM by sex were derived. Tracking of anthropometric variables was strong in both sexes, with correlation coefficients ranging from 0.68 to 0.89 (all P<.0001). Blood pressure tracking was less strong but as significant (r=.47 to .60, P<.0001). Tracking of LVM was significant in both sexes but slightly stronger in females (0.40 for males and 0.70 for females, P<.0001). An unexpected finding was that after LVM was divided by height2.7 to adjust for growth by the method of De Simone et al,11 this ratio tracked only in females (0.56, P<.0001).
Table 3⇓ shows the results of the univariate cross-sectional correlation analyses of critical variables with LVM and LVM/HT2.7. For unadjusted LVM, age, height, and weight were the most important correlates at each examination and for change in mass regardless of sex. Systolic blood pressure was a significant correlate in females. After LVM/HT2.7, measures of ponderosity (PI, weight, and triceps skin fold thickness) at both exams were still significant determinants of mass in females and were significant at the follow-up assessment in males. Systolic blood pressure remained a significant determinant of mass in females at each examination. In multivariate cross-sectional analyses, weight was the only significant predictor of adjusted LVM for both sexes (females: parameter estimate=0.29, r2=.08, P<.02 at baseline and parameter estimate=0.47, r2=.28, P<.001 at follow-up; males: parameter estimate=0.35, r2=.11, P<.002 at follow-up). No other independent associations were encountered in this model containing age, systolic blood pressure, weight, and triceps skin fold thickness.
Longitudinal Determinants of LVM
Table 4⇓ lists the factors contributing to the acquisition of LVM in univariate longitudinal analyses. Higher age, height, weight, adiposity, and systolic and diastolic blood pressures at the time of the first examination all tended to predict higher mass at follow-up, regardless of sex (P<.05, except for triceps skin fold thickness in males [P<.1]). For LVM/HT2.7, initial weight and measures of obesity continued to be important in determining ultimate heart mass for both males and females (P<.01). Systolic blood pressure retained its influence on final LVM in females (P<.05) but failed to reach statistical significance in males, although the trend was similar.
Stepwise multivariate regression was performed to look at determinants of follow-up LVM. Independent variables included baseline measures of age, weight, triceps skin fold thickness, systolic blood pressure, and LVM and change in weight (Table 5⇓). Only baseline weight was a significant factor in determining final raw and adjusted LVM consistently across sexes. In females, baseline LVM was also important in the prediction. Change in weight had a small influence on raw LVM but was not significant for corrected LVM/HT2.7. In males, baseline triceps skin fold thickness was an independent correlate in the model for LVM/HT2.7. Systolic blood pressure was not significant in any of the models.
Longitudinal Effect of Baseline LVM
The final derived longitudinal correlation coefficients explored the strength of baseline mass as a predictor of final anthropometric and blood pressure values (Table 6⇓). Individuals with higher initial LVM were more likely to become taller and heavier and have higher systolic blood pressure (P<.05). Females were likely to become more obese with higher PI and triceps skin fold thickness (P<.0001). Higher LVM/HT2.7 at baseline was predictive only of higher weight, PI, and triceps skin fold thickness in females (P<.05). Increased initial LVM/HT2.7 did show a trend toward prediction of higher follow-up systolic blood pressure in both sexes and reached significance in females (P<.05).
In multivariate analysis (Table 7⇓), the strongest predictor of follow-up systolic blood pressure was baseline systolic blood pressure in both sexes. In females, change in weight exerted a weak effect. Systolic blood pressure at baseline and change in pressure were then removed from the model because of the strong effect. In the repeated analysis, age at first examination became important in males, and initial LVM and change in triceps skin fold thickness were predictive of follow-up systolic blood pressure in females.
The present study has shown that LVM demonstrates tracking through late childhood and adolescence. The major factor determining tracking was linear growth. Height was an important predictor of mass in cross-sectional analyses; after LVM was indexed by division by height2.7, the correlation between initial and follow-up LVM was close to zero in males and reduced in females. Weight and measures of ponderosity were significant determinants of mass before and after adjustment for linear growth in both univariate and multivariate analyses. These results suggest that weight increases in excess of that predicted for linear growth determines the acquisition of increased ventricular mass in children. Finally, LVM at baseline was correlated with follow-up systolic blood pressure in both sexes, and this effect tended to persist for LVM indexed for growth, particularly in females.
Cross-sectional studies of children, adolescents, and young adults found that measures of body size and adiposity are the strongest correlates of LVM.4 5 9 10 12 23 24 25 26 27 28 Blood pressure was also found to be important, although the strength of the correlation is weaker than for measures of body size. This seems reasonable because the range in body size in a young and otherwise healthy population may present a greater physiological stress than the range in blood pressure. In a study of older children, Goble et al26 suggested that weight and not adiposity accounts for the largest portion of the variance of LVM. Recent investigations in normotensive children and adults proposed models to account for the exponential effect of linear growth on left ventricular size and showed independent effects of both weight and blood pressure on LVM after adjustment for these height effects.9 11 De Simone et al29 extended these analyses to a hypertensive population and found that the effect of obesity on myocardial growth is independent of and additive to that of high blood pressure. This study corroborates these findings in regard to the effect of weight on LVM and provides support from longitudinal analyses for the hypothesis that excess weight is an important, independent cause of the acquisition of increased LVM.
Several cross-sectional studies of healthy children and young adults found weak or minimal correlations of blood pressure with LVM. Hammond et al8 found that systolic blood pressure did not explain a significant portion of the variability in LVM in normotensive adults when analyzed by sex. Johnson et al25 also failed to demonstrate a difference in LVM between young adults with systolic blood pressure remaining greater than the 95th percentile over a 5-year period compared with those with pressures in the middle of the distribution. In younger children and adolescents in the Muscatine Study, analyses also showed no correlation of initial or final LVM adjusted for body size with blood pressure measured at the time of the first echocardiogram.30 In the Coronary Artery Risk Development in Young Adults (CARDIA) study, a cross-sectional study of a biracial cohort of young adults, Gardin et al10 found that the measures of body size were more important than blood pressure in determining LVM. Our finding of a smaller impact of blood pressure than anthropometric measures on LVM, particularly after adjustment for body size, is consistent with and confirmatory of these cross-sectional results. One explanation of this small effect may be that blood pressure in healthy children varies over a small range cross sectionally and does not change substantially over a 4- to 5-year interval. Therefore, longitudinal analyses over the time frame in this study may not allow for sufficient hemodynamic stress to produce significant compensatory hypertrophy. Another explanation may be that casual blood pressure recordings may not accurately characterize an individual’s true blood pressure. Devereux and Pickering31 found that in adults, “left ventricular muscle mass and wall thickness were more closely related to 24-h than casual systolic blood pressure in 13 studies that reported both correlations.”
However, LVM at baseline may be helpful in predicting those who are at increased risk of having a higher blood pressure in the future. This study demonstrated that increased LVM at baseline was correlated with higher follow-up blood pressure in univariate and multivariate analyses in females. De Simone et al32 found initial sex-adjusted LVM in adults was also “directly related to the risk for developing subsequent hypertension.” In fact, LVM and urinary sodium-potassium ratio were the only variables helpful in predicting the development of high blood pressure in an analysis where sex, race, age, initial blood pressure, body mass index, plasma renin activity, cholesterol, triglycerides, and glucose were also entered into the model.32 In children studied in Muscatine, Iowa, baseline LVM was also found to add independently to the value of baseline blood pressure for predicting systolic blood pressure 3 years later.30
Longitudinal assessment of LVM in children and adolescents is confounded by linear growth during the follow-up interval; this growth rate may vary substantially from individual to individual based on sex, age, sexual maturity, and genetic influences. Therefore, a method for indexing LVM for growth is essential. Studies have corrected LVM for BSA, PI, or Quetelet’s index.12 25 33 34 Recent studies suggested that indexing mass to a measure that includes weight in its calculation may not allow for assessment of the independent effects of weight.4 8 9 11 35 36 Indexing to height alone may not be accurate either because the relation between overall growth of an organism and that of a particular body part may be curvilinear or logarithmic.11 Thus, it is not surprising that De Simone et al11 and Malcolm et al9 found that height was related to ventricular mass exponentially. In our study, we used nonlinear regression analysis to relate LVM to different measures of body size using the data from our study cohort. We found an allometric relation between LVM and height similar to that found by De Simone et al11 and used their method in this study (data not shown).
Another potential source of error is the bias introduced when echocardiograms must be discarded because of poor quality. More studies are discarded from obese individuals owing to the difficulty in obtaining adequate echocardiographic windows. However, Savage et al37 found that even in the most obese young men, acceptable echocardiograms could be obtained 90% of the time. Only in individuals more than 60 years of age who had lower forced vital capacity or overt cardiovascular disease could an adequate percentage of echocardiograms not be obtained.37
In the current study, 9 of 160 echocardiograms obtained (5.3%) were discarded owing to inadequacy of the M-mode tracing. However, these individuals did not differ statistically in respect to height, weight, or blood pressure from the study population who had two adequate echocardiograms or the subjects who participated only in the initial echo study (data not shown).
The selection of study participants by stability of blood pressure may be considered a limitation of the study. LVM is known to relate more closely to average blood pressure load over time than to blood pressure obtained at a single point.12 30 38 If a person has recently had a change in blood pressure, his or her LVM may not have had a chance to respond to this stressor, and the blood pressure–mass relation may be obscured. For this reason, individuals whose blood pressure was “tracking” (retaining relative rank over time) were selected for the study. Ambulatory blood pressure recordings may be better able to characterize an individual’s true blood pressure. However, this modality for studying blood pressure was not available when the first echo study was performed (1984). Therefore, the most reliable and tested method for characterizing an individual’s blood pressure at that time was used: replicate measures performed at different points in time by trained observers. It is also important to note that the study cohort did not differ statistically in race-sex composition, blood pressure, or anthropometric measurements from all individuals screened in Bogalusa during either the baseline (1978 to 1979) or follow-up (1981 to 1982) cross-sectional surveys. However, one must still use caution in attempting to generalize the results obtained in this study to children who do not have stable blood pressures. Other physiological mechanisms may be operating that might influence the relation between blood pressure and LVM.
Left ventricular hypertrophy is known to carry an increased risk of cardiovascular morbidity in adults.1 2 The factors leading to the acquisition of excess LVM may have their origin in youth because both measures of body size and blood pressure in youth track into adulthood. This study demonstrates in longitudinal analyses that excess weight may lead to the acquisition of LVM in excess of that expected from normal growth. The effect of obesity on cardiovascular structure and function may evolve over years. The acquisition of obesity may be a significant, and possibly modifiable, risk factor for developing left ventricular hypertrophy. This may be most important for younger adults because it was observed in the Framingham study39 that the effect of obesity on cardiovascular morbidity and mortality is most striking in younger adults.
This research was supported by funds from the NHLBI of the US Public Health Service Early Natural History of Arteriosclerosis (No. 5R01 HL38844). The Bogalusa Heart Study represents the collaborative efforts of many people whose cooperation is gratefully acknowledged. We wish to thank the Bogalusa staff and the children of Bogalusa, without whom this study would not have been possible. We appreciate the efforts of Dr and Mrs Daniel D. Savage in helping to obtain these data.
- Received July 26, 1994.
- Revision received November 2, 1994.
- Accepted November 26, 1994.
- Copyright © 1995 by American Heart Association
Casale PN, Devereux RB, Milner M, Zullo G, Harshfield GA, Pickering TG, Laragh JH. Value of echocardiographic measurement of left ventricular mass in predicting cardiovascular morbid events in hypertensive men. Ann Intern Med. 1986;105:173-178.
Gunther S, Grossman W. Determinants of ventricular function in pressure-overload hypertrophy in man. Circulation. 1979;59:679-688.
Henry WL, Gardin JM, Ware JH. Echocardiographic measurements in normal subjects from infancy to old age. Circulation. 1980;62:1054-1061.
Morales MC, Gleim GW, Marino ND, Stachenfeld NS, Coplan NL. Left ventricular mass in adolescent basketball players. Cardiovasc Rev Reports. 1992;13:60-62.
Friedman WF. Aortic stenosis. In: Adams FH, Emmanouilides GC, Riemenschneider TA, eds. Moss’ Heart Disease in Infants, Children and Adolescents. Baltimore, Md: Williams & Wilkens; 1989:224-255.
Malcolm DD, Burns TL, Mahoney LT, Lauer RM. Factors affecting left ventricular mass in childhood: the Muscatine Study. Pediatrics. 1993;92:703-709.
Gardin JM, Wagenknecht LE, Anton-Culver H, Flack J, Gidding S, Wong ND, Manolio TA. Relationship of race, sex, systolic blood pressure levels, and body mass to left ventricular mass in healthy young adults: the CARDIA Study. Circulation. 1991;84(suppl II):II-34. Abstract.
Burke GL, Arcilla RA, Culpepper WS, Webber LS, Chiang YK, Berenson GS. Blood pressure and echocardiographic measures in children: the Bogalusa Heart Study. Circulation. 1987;75:106-114.
Schieken RM, Clarke WR, Mahoney LT, Lauer RM. Measurement criteria for group echocardiographic studies. Am J Epidemiol. 1979;110:504-514.
Sahn DJ, DeMaria A, Kisslo J, Weyman A. Recommendations regarding quantitation in M mode echocardiography: results of a survey of echocardiographic measurements: the committee on M-mode standardization of the American Society of Echocardiography. Circulation. 1978;58:1072-1083.
Wallerson DC, Devereux RB. Reproducibility of echocardiographic left ventricular measurements. Hypertension. 1987;9(suppl II):II-6-II-18.
Feigenbaum H. Echocardiography. Philadelphia, Pa: Lea & Febiger; 1994.
Devereux RB. Detection of left ventricular hypertrophy by M-mode echocardiography: anatomic validation, standardization, and comparison to other methods. Hypertension. 1987;9(suppl II):II-19- II-26.
SAS/STAT User’s Guide. Cary, NC: SAS Institute Inc; 1989.
Voors AW, Harsha DW, Webber LS. Relation of blood pressure to stature in healthy young adults. Am J Epidemiol. 1982;115:833-840.
McMahon T. Size and shape in biology. Science. 1973;179:1201-1204.
Steinberger J, Bass JL, Berry JM, Gomez-Marin O, Rocchini AP, Sinaiko AR. Obesity and female sex influence left ventricular size in children with high normal blood pressure. Circulation. 1993;88(suppl I):I-98. Abstract.
Schieken RM, Clarke WR, Lauer RM. Left ventricular hypertrophy in children with blood pressures in the upper quintile of the distribution: the Muscatine Study. Hypertension. 1981;3:669-675.
Goble MM, Mosteller M, Moskowitz WB, Schieken RM. Sex differences in the determinants of left ventricular mass in childhood: the Medical College of Virginia Twin Study. Circulation. 1992;85:1661-1665.
De Simone G, Devereux RB, Roman MJ, Alderman MH, Laragh JH. Relation of obesity and gender to left ventricular hypertrophy in normotensive and hypertensive adults. Hypertension. 1994;23:600-606.
Mahoney LT, Schieken RM, Clarke WR, Lauer RM. Left ventricular mass and exercise responses predict future blood pressure: the Muscatine Study. Hypertension. 1988;12:206-213.
Devereux RB, Pickering TG. Relationship between the level, pattern and variability of ambulatory blood pressure and target organ damage in hypertension. J Hypertens. 1991;9(suppl 8):S34-S38.
De Simone G, Devereux RB, Roman MJ, Schlussel Y, Alderman MH, Laragh JH. Echocardiographic left ventricular mass and electrolyte intake predict arterial hypertension. Ann Intern Med. 1991;114:202-209.
Culpepper WS, Sodt PC, Messerli FH, Ruschhaupt DG, Arcilla RA. Cardiac status in juvenile borderline hypertension. Ann Intern Med. 1983;981-987.
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.
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.
Devereux RB, Pickering TG. Relationship between ambulatory or exercise blood pressure and left ventricular structure: prognostic implications. J Hypertens. 1990;8(suppl 6):S125-S134.
Hubert HB, Feinleib M, McNamara PM, Castelli WP. Obesity as an independent risk factor for cardiovascular disease: a 26-year follow-up of participants in the Framingham Heart Study. Circulation. 1983;67:968-977.