Effect of Single-Drug Therapy on Reduction of Left Ventricular Mass in Mild to Moderate Hypertension
Comparison of Six Antihypertensive Agents
Background Antihypertensive drugs may differ in their ability to reduce LV mass. Covariates other than drug selection, such as pretreatment LV mass, body weight, the magnitude of blood pressure reduction, race, and age may modify the response of LV mass to therapy.
Methods and Results Patients with mild to moderate hypertension (diastolic blood pressure, 95 to 109 mm Hg) were randomly allocated to treatment with atenolol, captopril, clonidine, diltiazem, hydrochlorothiazide, or prazosin in a double-masked trial. Patients achieving the goal diastolic blood pressure of <90 mm Hg during drug titration entered a 1-year maintenance period. Longitudinal analysis examined changes from baseline echocardiogram in LV mass at 8 weeks and at 1 year, statistically adjusted for pretreatment LV mass, systolic blood pressure, body weight, sodium excretion, physical activity, race, and age. Significant reductions at 1 year in adjusted LV mass were seen for patients in the highest tertile of pretreatment LV mass treated with hydrochlorothiazide (mean, −42.9; 95% confidence limits, −65.5, −20.2 g), captopril (mean, −38.7; 95% confidence limits, −61.0, −16.4 g), and atenolol (mean, −28.1; 95% confidence limits, −50.9, −5.3 g). These treatment effects differed from those of prazosin, diltiazem, or clonidine.
Conclusions Antihypertensive drugs have disparate effects on LV mass independent of the magnitude of blood pressure reduction. Patients with adequate blood pressure control on captopril, hydrochlorothiazide, and atenolol show a reduction of LV mass after 1 year of treatment, whereas patients on diltiazem, clonidine, or prazosin do not.
Increases in LV mass, even below partition values sufficient to define hypertrophy, are associated with incremental risk of cardiovascular morbidity and mortality.1 2 3 Moreover, evidence suggests that regression of LVH produces benefit over and above that conferred by blood pressure reduction alone.4 5
Although antihypertensive therapy has been shown to reduce LV mass and LVH prevalence,6 some studies6 7 8 indicate that not all drugs are equally effective in reducing LV mass, even with comparable reduction of blood pressure. In particular, it has been suggested that diuretics and vasodilators may be ineffective in decreasing LV mass because of their failure to inhibit neurohumoral mechanisms responsible for LVH.9 However, much of the literature6 on LV mass regression has been based on relatively small, uncontrolled, short-term, and nonrandomized studies with only one or two therapeutic limbs. Moreover, the influence of covariates other than drug selection known to affect LV mass10 11 has often not been evaluated.
The Veterans Affairs Cooperative Study on Single-Drug Therapy in Mild-Moderate Hypertension was a randomized trial that tested the comparative blood pressure–lowering efficacy of six different classes of drugs and placebo.12 One objective of this study was to use echocardiography to assess the response of LV mass and its structural components over the 1-year period of antihypertensive monotherapy. Since the mechanisms responsible for LVH, and presumably its regression, are multifactorial, the data were analyzed with adjustment for potentially contributory factors by appropriate statistical methods.
Criteria for patient selection, randomization, and treatment have been detailed elsewhere.12 Briefly, 1105 men, recruited at 15 Veterans Affairs Medical Centers, with diastolic blood pressure 95 to 109 mm Hg after a 4- to 8-week washout period on placebo were randomly allocated to double-masked treatment with one of six drugs. A seventh group randomized to placebo was not included in analyses for this report. The medications were atenolol 25 to 100 mg daily, captopril 12.5 to 50 mg twice daily, clonidine 0.1 to 0.3 mg twice daily, diltiazem-SR 60 to 180 mg twice daily, hydrochlorothiazide 12.5 to 50 mg daily, and prazosin 2 to 10 mg twice daily. Patients who achieved the diastolic blood pressure goal of <90 mm Hg without adverse drug effects at the end of the 8-week titration period were then entered into a maintenance phase for 1 year.
Echocardiograms were obtained at baseline, at the end of titration, and at 1 year. Two-dimensional targeted M-mode echocardiography was performed with particular care to obtain orthogonal, nonangulated sectors so as to achieve recording of M-mode dimensions through the minor axis of the LV at or just distal to the mitral valve tips. Measurements of ventricular septum, LV cavity, and posterior wall dimensions from paper strip-chart recordings were made by a single reader with an off-line image analysis system according to ASE criteria.13 Intraobserver and interobserver (reading comparison with author J.S.G.) variabilities were determined from duplicate masked readings of 121 echocardiograms. Interobserver errors for septal thickness, posterior wall thickness, and LV cavity size were 9.1%, 8.7%, and 3.1%, respectively. Intraobserver errors were 8.2%, 6.9%, and 2.3%, respectively. LV mass was calculated as described elsewhere.14 For comparison with population normals, LV mass measurements were converted to Cornell-Penn convention values by use of published regression equations.15 For determination of the presence of LVH, LV mass (Cornell-Penn) was indexed to body surface area; 134 g/m2 was selected as the partition value.15 In addition, to avoid the flattening contribution of obesity to the estimation of LVH prevalence, LV mass was indexed to height according to the Framingham convention16 ; 164 g/m was selected as the partition value.
Clinical and Laboratory Assessments
Blood pressure was measured at each visit with a cuff sphygmomanometer after 15 minutes of rest sitting upright with the back and arm supported. All visit blood pressures represented the mean of three seated measurements. The baseline blood pressure was taken as the mean of the blood pressure averages for the randomization visit and the preceding visit. Treatment blood pressures were the mean of the blood pressure averages for the first two consecutive visits at which goal blood pressure was achieved. Physical activity index was obtained at the time of each echocardiogram by a questionnaire that queried work and recreational physical activity for the preceding 6-month time period at baseline and the preceding 1-month period at 8 weeks and 1 year. Plasma renin was determined at baseline by 125I radioimmunoassay (Clinical Assays, Travenol Division, Genentech Diagnostics, Inc). Sodium intake was determined from one 24-hour measurement of urinary sodium excretion (mmol) at baseline, at 8 weeks, and at 1 year.
In a previous publication from this study, baseline analysis of the relationships between LV mass and body weight, systolic blood pressure, sodium excretion, plasma renin, race, age, and physical activity disclosed independent effects of systolic blood pressure and body weight.11 In the present study, these covariates were also examined in analyses of treatment effects on LV mass changes.
Because of the number of randomized patients who were not included in the analysis, primarily because of (1) lack of an adequate blood pressure response, (2) blood pressure rising above safety limits, or (3) inability to obtain a readable echo, patients' baseline characteristics were compared between those with and without an LV mass measurement at 8 weeks and at 1 year. The t test for independent groups was used for continuous variables, and the Χ2 test was used for categorical variables. Baseline comparability of the treatment groups was examined for those with LV mass measurements at 8 weeks and at 1 year. One-way ANOVA was used for continuous variables and the Χ2 test for categorical variables.
Comparisons among treatment groups of changes from baseline to 8 weeks and 1 year in mean LV mass, its components, and patient covariates were performed by ANOVA. Pairwise comparisons of groups by Tukey's method were made if the ANOVA statistic was significant at the .05 level. Within-group changes from baseline were analyzed by the paired t test.
To account for the effect of pretreatment LV mass on changes in LV mass, patients were stratified into three equal-sized groups based on their pretreatment LV mass. At 8 weeks and 1 year, two-way ANOVA was conducted to evaluate the effect of the treatments and pretreatment LV mass on mean changes in LV mass from baseline.
To estimate and compare longitudinal changes in LV mass for the treatments, adjusting for patient covariates, mixed-model ANOVA was performed.17 This is a type of repeated-measures analysis that allows for incomplete data. The computations proceed in two stages. First, an appropriate regression model is selected to characterize the typical LV mass-response curve over time at the patient level of analysis. We assumed a model that was linear in time with a random intercept and a compound symmetry covariance structure among the serial LV mass measurements. Second, the “average” regression curve is estimated for each treatment group to examine whether slopes and intercepts differed across treatment groups after patient covariates were adjusted for.
Mixed-model analysis may be biased if the pattern of missing data is not random. Although inability to obtain a readable echo behaved like a random process, withdrawals from study were primarily systematic, due to blood pressure safety rules. To account for this type of withdrawal, we included a variable in the analysis that identified whether the patient was withdrawn from the study at the end of titration for inadequate blood pressure control.18
Differences in slopes between treatments were assessed by specifying all possible pairwise contrasts. With Bonferroni adjustment, a value of P=.0033 was considered statistically significant for inferences regarding comparisons of treatments. On the basis of the obtained model, we also derived estimates of the within-treatment-group changes in mean LV mass, adjusted for patient covariates, for the three tertiles of baseline LV mass. The values of patient covariates used were the means for those patients in the analysis. At 8 weeks (1 year), these values were systolic blood pressure, 152.5 (150.4) mm Hg; age, 58.1 (57.9) years; weight, 197.4 (195.9) lb; 24-hour sodium excretion, 160.2 (153.5) mmol; and physical activity index, 64.3 (69.7) units.
The statistical computer package SAS was used to generate the statistical analyses.19 Values for quantitative measures are expressed as the mean±SD unless otherwise stated. All statistical tests were two-tailed, and a value of P≤.05 was used to identify statistically significant results.
Eleven hundred five patients were randomized to one of the six treatment groups, 683 of these patients completed the antihypertensive medication titration phase (8 weeks), and 493 completed 1 year of maintenance therapy. Echocardiographic measurements were available for 587 patients (53% of the randomized patients) at baseline, 406 (37%) at 8 weeks, and 230 (21%) at 1 year. Table 1⇓ summarizes the reasons measurements were not obtained at each time point for the original group of randomized patients. At 8 weeks, the acquisition rate was reduced, primarily because 47% of patients either were unable to be scheduled for or did not have a readable echocardiogram. A postrandomization echocardiogram was not obtained if a baseline reading was unavailable. At 1 year, the acquisition rate was reduced further, primarily because 30% of patients did not achieve adequate blood pressure reduction during titration and were ineligible to proceed to maintenance. Of note, after treatment effects were adjusted for, diastolic blood pressure decrement during titration was negatively associated with baseline LV mass (P=.045), ie, patients with higher baseline LV mass had smaller decreases in diastolic blood pressure at 8 weeks. This relationship was not seen for the change in systolic blood pressure at 8 weeks and was not dependent on the assigned treatment. When treatment groups were compared, the acquisition rate at 1 year was highest for diltiazem because superior blood pressure reduction in this group resulted in fewer protocol-dictated withdrawals, whereas the rate was lowest for prazosin because a higher percentage of patients withdrew due to medication adverse effects.
Patient Characteristics at Baseline
At baseline, the average age of the patients was 57.7±10.4 years, blood pressure averaged 152.5±13.5/99.5±3.5 mm Hg, and average LV mass (ASE) was 327.3±92.6 g. Blacks composed 58.3% of the study sample. The prevalence of LVH was 63.7% by Framingham criteria and 45.3% by Cornell criteria.
In comparison with patients who did not have echocardiographic measurements at 8 weeks, those who did were younger and more likely to be black. At 1 year, those with LV mass measurements were more likely to be black and to have lower systolic and diastolic blood pressures and lower 24-hour urine sodium excretion. The lower acquisition rate for whites was primarily due to the limited availability of suitable echocardiographic equipment at a few predominantly white sites and an extraordinarily high acquisition rate at one predominantly black site.
For 8-week measurements, there were small but statistically significant baseline differences between treatment groups for diastolic blood pressure (P=.02), ranging from a mean of 98.6 mm Hg for the atenolol group to 100.4 mm Hg for the prazosin group. For patients with 1-year measurements, baseline comparison across treatment groups revealed no significant differences.
Changes from baseline to 8 weeks and 1 year for systolic blood pressure, weight, sodium excretion, and physical activity are displayed in Table 2⇓. At 8 weeks, significant treatment differences existed for changes in systolic blood pressure and weight, whereas at 1 year, significant differences existed for changes in systolic blood pressure. No statistically significant differences in treatment duration through 1 year were found among those patients who entered maintenance.
Effects of Treatment on LV Mass
Table 3⇓ displays observed mean changes in LV mass from baseline by treatment group. Baselines are presented separately for those with 8-week and 1-year measurements. At 1 year, mean LV mass showed a significant reduction from baseline for patients given captopril (P=.05), and the mean LV mass reduction for the hydrochlorothiazide group approached significance (P=.08).
Patients were then stratified into three equal-sized groups according to their baseline LV mass: ≤275 g, 275 to 350 g, and >350 g. There were several within-group changes of note (Fig 1⇓). For patients in the highest tertile of baseline LV mass, significant decreases from baseline occurred at 8 weeks with diltiazem (−48 g, P=.004) and prazosin (−54 g, P=.005).
Fig 2⇓ shows comparable data summaries for changes from baseline to 1 year. A strong statistical trend (P=.06) for treatment effect differences was evident after adjustment for baseline LV mass. The following within-group changes were found at 1 year. For patients in the highest tertile of LV mass at baseline, significant decreases from baseline occurred with hydrochlorothiazide (−66 g, P=.0005), captopril (−45 g, P=.004), and atenolol (−37 g, P=.02). An increase in mean LV mass was noted among patients in the lowest tertile treated with diltiazem (34 g, P=.008).
Longitudinal Analysis of Treatment Effects on LV Mass
Mixed-model ANOVA was used to assess the independent effect of treatment on serial LV mass measurements after adjustment for patient age and race as well as serial measurements of systolic blood pressure, sodium excretion, weight, and physical activity. Diastolic blood pressure was not considered because of its high correlation with systolic blood pressure. Similarly, baseline plasma renin activity was not included because of its strong relationship with age and race. Finally, we assessed whether protocol-dictated withdrawals for lack of blood pressure response influenced estimates of treatment effects on LV mass.
This model showed significant effects of weight (P=.016) on LV mass over the duration of the study. Age (P=.09), systolic blood pressure (P=.13), race (P=.18), and sodium excretion (P=.21) showed trends, whereas physical activity (P=.98) had no effect. The effects of weight, systolic blood pressure, race, and sodium excretion were constant throughout the duration of follow-up and across treatment groups. Systolic blood pressure effects were probably weakened because of the presence of other covariates in the model, especially whether patients had an adequate blood pressure response to enter maintenance. Patients with lower baseline LV mass tended to have increases over time, whereas those with higher baseline LV mass tended to have decreases (P≤.0001). LV mass decreases over time were greater with increasing age (P=.06). Although patients' dropping from study at 8 weeks because of lack of blood pressure response altered the relationship between baseline LV mass and time (P≤.0001), the comparison of treatment effects on LV mass was not biased by withdrawals. After adjustment for these covariates, treatment group differences in LV mass slopes were highly significant (P=.01).
Comparison of the LV mass slopes adjusted for patient covariates revealed differences between drugs approaching significance (by the Bonferroni-adjusted significance level of .0033) where the negative LV mass slope of hydrochlorothiazide differed from positive slopes for prazosin (P=.009), diltiazem (P=.007), and clonidine (P=.010) and the negative slope for captopril differed from the positive slopes for prazosin (P=.017), diltiazem (P=.015), and clonidine (P=.021).
Fig 3⇓ displays changes in LV mass at 8 weeks and 1 year, adjusted for covariates, for the three tertiles of baseline LV mass. For patients in the highest tertile, those given hydrochlorothiazide (mean, −42.9; 95% confidence limits, −65.5, −20.2 g), captopril (mean, −38.7; 95% confidence limits, −61.0, −16.4), and atenolol (mean, −28.1; 95% confidence limits, −50.9, −5.3 g) are estimated to have a significant reduction in LV mass at 1 year. For patients in the lowest tertile, those given prazosin (mean, 33.2; 95% confidence limits, 9.4, 57.0 g), clonidine (mean, 29.5; 95% confidence limits, 6.5, 52.6 g), and diltiazem (mean, 26.6; 95% confidence limits, 5.2, 48.1) are estimated to have a significant increase in LV mass at 1 year.
Effects of Treatment on Components of LV Mass
Serial changes in the components of LV mass are also displayed in Table 3⇑. The group taking hydrochlorothiazide showed a significant reduction in mean septal thickness at 8 weeks and trends toward decreases in mean posterior wall thickness at 8 weeks and 1 year. The atenolol-treated group showed a significant reduction in mean posterior wall thickness at 8 weeks and 1 year and a significant increase in mean LV cavity size at 8 weeks. Patients taking captopril showed a strong trend toward a reduction in mean septal thickness at 1 year. The clonidine group showed strong trends toward reduction in mean septal and posterior wall thickness at 8 weeks. The diltiazem-treated group showed a significant reduction in mean posterior wall thickness and a significant increase in LV cavity size at 1 year. The prazosin group showed a strong trend toward an increase in mean LV cavity size at 1 year.
The results of this study show that in men with mild to moderate hypertension and a high prevalence of LVH, various classes of antihypertensive drugs have disparate effects on LV mass. Moreover, at least some of the drug-related differences in reduction of LV mass were independent of differences in factors known to affect LV mass, such as the magnitude of systolic blood pressure reduction, body weight, level of physical activity, race, and age. After adjustment for these covariates, baseline LV mass influenced the results of the study such that, at 1 year, decreases in LV mass were noted with hydrochlorothiazide, captopril, and atenolol in the highest tertile, whereas increases were noted with prazosin, diltiazem, and clonidine in the lowest tertile. Reduction of LV mass with hydrochlorothiazide was associated with decreased LV wall thickness rather than decreased LV cavity volume.
The results of the present study are in contrast to the findings of some studies6 8 20 21 22 that have suggested that diuretics are either completely ineffective for reduction of LV mass or produce only small decreases in LV mass disproportionate to their substantial effects on blood pressure. Moreover, studies in animals and in humans21 23 24 25 have suggested a physiological basis for ineffective regression of LVH with diuretics, in contrast to that produced by other drugs. Despite equivalent reduction of blood pressure, drugs that interfere with postulated neurohormonal mechanisms of hypertrophy would be expected to produce greater reduction of LV mass than diuretics, which fail to inhibit, or actually activate, these mechanisms.26
Despite these considerations, some evidence8 20 27 suggests that diuretics can in fact produce decreases in LV mass, although inferior to those produced by calcium channel blockers and ACE inhibitors.8 However, many studies of LV mass reduction have been criticized for methodological limitations,20 28 including short duration, lack of controls, small sample sizes, and uncertain masking.
The present study supports the findings of other, generally smaller trials that have usually not adjusted for covariates in documenting the efficacy of an ACE inhibitor for reduction of LV mass.6 Moreover, with similar classes of antihypertensive drugs, our findings support those of TOMHS,28 29 a double-blind, placebo-controlled trial of 844 patients with mild hypertension, in noting efficacy of a diuretic, an ACE inhibitor, and a β-blocker in decreasing LV mass. However, the results of our study extend those of TOMHS by evaluating patients with greater severity of hypertension, higher LV mass, and greater prevalence of LVH, as well as a higher proportion of black patients. Our findings differ from TOMHS in failing to document efficacy of a calcium channel blocker or α-adrenergic antagonist for reduction of LV mass, possibly as a result of drug-specific effects or differences in study population.
Although we previously found that blood pressure response differences among types of antihypertensive medications were race dependent,12 we found no race differences in LV mass changes for patients whose blood pressure was controlled on each drug
Limitations and Advantages of the Present Study
Analyses were not performed on an “intention-to-treat” basis. The study design required patients to remain on the initially assigned monotherapy and dropped them from further follow-up if they required other or additional medication for blood pressure control or changed therapy because of side effects. This design aspect of the study is consistent with clinical practice, in which patients are not continued on antihypertensive medication in the absence of blood pressure lowering despite the possibility of reduction of LV mass.
Hence, the findings could be biased by differences in dropout rates and redistribution of relevant biological characteristics across treatment groups. However, baseline characteristics of patients with readable echocardiograms remained comparable across treatment groups, and covariate adjustment for changes in biological predictors of LV mass increased the statistical significance of treatment comparisons and showed virtually no impact of missing data on the estimates of treatment effects. Although the requirement for adherence to monotherapy contributed to study dropouts, the study design permitted assessment of patients uncomplicated by drug crossovers. Despite study dropouts, 230 patients (28 to 52 patients per treatment limb) evaluated at 1 year still represent a greater number of patients evaluated for a longer period of time on monotherapy than in most previous studies.
The finding of time effects on LV mass independent of treatment effects, for which patients with lower baseline LV mass tended to have increases on average whereas those with higher LV mass tended to have decreases, most likely represents regression to the mean. Accordingly, regression to the mean at 1 year may account for some of the decreases in the highest tertile of baseline LV mass noted with hydrochlorothiazide, captopril, and atenolol as well as increases in the lowest tertile of LV mass noted with diltiazem, prazosin, and clonidine. However, the converse was not true. Patients in the lowest tertile of LV mass who received atenolol, captopril, or hydrochlorothiazide did not have increases in LV mass, whereas those in the highest tertile who received prazosin, diltiazem, or clonidine did not have decreases. Because for each drug, changes in LV mass consistent with regression to the mean patterns were evident only at one end of the distribution of pretreatment LV mass and not the other, it is unlikely that overall treatment effects are due to regression to the mean.
Additionally, we studied only men who had a high prevalence of LVH. Although there are no data to suggest that women respond differently to antihypertensive therapy, our data cannot be directly extrapolated to women or to hypertensive patients with less severe disease.
Previously, the use of diuretics as monotherapy in hypertension has been questioned, in part because of perceived inferiority in reducing LV mass. Alternatively, it has been suggested that long-term reduction of hemodynamic load with any antihypertensive drug, including diuretics, can produce reduction of LV mass.30 However, the present study shows that patients with adequate blood pressure control on hydrochlorothiazide, captopril, or atenolol show significant reduction of LV mass after 1 year of treatment, whereas patients with adequate blood pressure control on diltiazem, clonidine, or prazosin do not. Moreover, LV mass reduction with hydrochlorothiazide occurs with relatively low doses not associated with important adverse metabolic or clinical effects.12 27 If reduction of LV mass is established as an important goal of antihypertensive therapy as part of overall cardiovascular disease risk reduction, then the efficacy and relative economic advantages of hydrochlorothiazide may have important public health implications.
Selected Abbreviations and Acronyms
|ASE||=||American Society of Echocardiography|
|LVH||=||left ventricular hypertrophy|
|TOMHS||=||Treatment of Mild Hypertension Study|
- Received July 29, 1996.
- Revision received January 6, 1997.
- Accepted January 7, 1997.
- Copyright © 1997 by American Heart Association
Casale PN, Devereux RB, Milner M. Value of echocardiographic measurement of left ventricular mass in predicting cardiovascular morbid events in hypertensive men. Ann Intern Med. 1986;105:173-178.
Koren MJ, Devereux RB, Casale PN. Relation of left ventricular mass and geometry to morbidity and mortality in uncomplicated essential hypertension. Ann Intern Med. 1991;114:345-352.
Levy D, Salomon M, D'Agostino R, Belanger A, Kannel W. Prognostic implications of baseline electrocardiographic features and their serial changes in subjects with left ventricular hypertrophy. Circulation. 1994;90:1786-1793.
Yuranev AP, Dyakonova HG, Novikov ID, Vitols A, Pahl L, Haynemann G, Wallrabe D, Tsifkova R, Romanovska L, Niderle P, Tsiskarishvili D, Davarashvili T, Gelovani K, Kochachidze T, Balash A. Management of essential hypertension in patients with different degrees of left ventricular hypertrophy: multicenter trial. Am J Hypertens.. 1992;5:182S-189S.
Tarazi RC, Sen S, Fouad FM, Wicker P. Regression of myocardial hypertrophy: conditions and sequelae of reversal in hypertensive heart disease. In: Alpert NR, ed. Perspectives in Cardiovascular Research. Vol 7, Myocardial Hypertrophy and Failure. New York, NY: Raven Press; 1983:637-652.
Liebson P. Echocardiographic assessment of the left ventricle in hypertension: perspectives on epidemiologic studies and clinical trials. J Vasc Biol Med. 1993;4:285-310.
Materson BJ, Reda DJ, Cushman WC, Massie BM, Freis ED, Kochar MS, Hamburger RJ, Fye C, Lakshman R, Gottdiener J. Single-drug therapy for hypertension in men: a comparison of six antihypertensive agents with placebo: the Department of Veterans Affairs Cooperative Study Group on Antihypertensive Agents. N Engl J Med. 1993;328:914-921.
Sahn DJ, DeMaria A, Kisslo J, Weyman A, 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 measurements. Circulation. 1978;58:1072-1083.
Troy BL, Pombo J, Rackley CE. Measurement of left ventricular wall thickness and mass by echocardiography. Circulation. 1972;45:602-611.
Diggle P, Liang K, Zeger S. Analysis of Longitudinal Data. Oxford, UK: Clarendon Press; 1994.
SAS Institute Inc. SAS/STAT Users Guide, Version 6. 4th ed. Cary, NC: SAS Institute; 1989.
Massie BM. Effect of diuretic therapy on hypertensive left ventricular hypertrophy. Eur Heart J. 1992;13(suppl G):53-60.
Pfeffer MA, Pfeffer JM. Pharmacologic regression of cardiac hypertrophy in experimental hypertension. J Cardiovasc Pharmacol. 1984;6(suppl 6):S865-S869.
Burnier M, Brunner HR. Neurohormonal consequences of diuretics in different cardiovascular syndromes. Eur Heart J. 1992;13(suppl G):28-33.
Moser M, Setaro JF. Antihypertensive drug therapy and regression of left ventricular hypertrophy: a review with a focus on diuretics. Eur Heart J. 1991;12:1034-1039.
Liebson PR, Grandits GA, Dianzumba S, Prineas RJ, Grimm RH Jr, Neaton JD, Stamler J. Comparison of five antihypertensive monotherapies and placebo for change in left ventricular mass in patients receiving nutritional-hygienic therapy in the Treatment of Mild Hypertension Study (TOMHS). Circulation. 1995;91:698-706.
Frohlich ED. The heart in hypertension: a 1991 overview. Hypertension. 1991;18(suppl 5):62-68.