Long-Term Effects of Chlorthalidone Versus Hydrochlorothiazide on Electrocardiographic Left Ventricular Hypertrophy in the Multiple Risk Factor Intervention Trial
Chlorthalidone (CTD) reduces 24-hour blood pressure more effectively than hydrochlorothiazide (HCTZ), but whether this influences electrocardiographic left ventricular hypertrophy is uncertain. One source of comparative data is the Multiple Risk Factor Intervention Trial, which randomly assigned 8012 hypertensive men to special intervention (SI) or usual care. SI participants could use CTD or HCTZ initially; previous analyses have grouped clinics by their main diuretic used (C-clinics: CTD; H-clinics: HCTZ). After 48 months, SI participants receiving HCTZ were recommended to switch to CTD, in part because higher mortality was observed for SI compared with usual care participants in H-clinics, whereas the opposite was found in C-clinics. In this analysis, we examined change in continuous measures of electrocardiographic left ventricular hypertrophy using both an ecological analysis by previously reported C- or H-clinic groupings and an individual participant analysis where use of CTD or HCTZ by SI participants was considered and updated annually. Through 48 months, differences between SI and usual care in left ventricular hypertrophy were larger for C-clinics compared with H-clinics (Sokolow-Lyon: −93.9 versus −54.9 μV, P=0.049; Cornell voltage: −68.1 versus −35.9 μV, P=0.019; Cornell voltage product: −4.6 versus −2.2 μV/ms, P=0.071; left ventricular mass: −4.4 versus −2.8 g, P=0.002). At the individual participant level, Sokolow-Lyon and left ventricular mass were significantly lower for SI men receiving CTD compared with HCTZ through 48 months and 84 months of follow-up. Our findings on left ventricular hypertrophy support the idea that greater blood pressure reduction with CTD than HCTZ may have led to differences in mortality observed in the Multiple Risk Factor Intervention Trial.
- left ventricular hypertrophy
- blood pressure
See Editorial Commentary, pp 994–995
Lowering of blood pressure with a diuretic-based regimen has decreased stroke, heart failure, and cardiovascular disease events in several outcome-based clinical trials.1 Most of these studies used the thiazide-like diuretic chlorthalidone (CTD), yet 95% of thiazide prescriptions are for hydrochlorothiazide (HCTZ).2 There has been little direct comparison of the 2 agents on measures other than blood pressure reduction.1
One source of comparative data comes from the Multiple Risk Factor Intervention Trial (MRFIT), which used both CTD and HCTZ in a nonrandomized manner.3 Almost 5 years into the trial, a recommendation was made by the external data and safety monitoring committee (ie, the MRFIT Policy Advisory Board) to switch all of the hypertensive participants in the special intervention (SI) group who were taking 50 or 100 mg/d of HCTZ to 50 mg/d of CTD.4 This protocol change was prompted by an ecological analysis revealing that coronary heart disease mortality rates for the SI group were significantly higher than for usual care (UC) participants in clinics where HCTZ was predominantly used (ie, H-clinics).4,5 Conversely, mortality rates were lower for SI compared with UC participants in clinics with high use of CTD (ie, C-clinics). After the protocol change, this unfavorable mortality pattern for SI compared with UC participants reversed during follow-up, giving relevance of the ecological association identified earlier.4 Although choice of diuretic was believed to have influenced the findings, conclusions about the specific impact of CTD or HCTZ in MRFIT remain speculative.
A small, prospective study suggested that CTD can sustain reductions in blood pressure throughout a 24-hour period more effectively than HCTZ, probably because of its longer half-life.6 These data led us to investigate the effects of CTD and HCTZ on electrocardiographic left ventricular hypertrophy (LVH) in MRFIT. LVH is strongly influenced by blood pressure and is an established risk factor for coronary heart disease.7–9 Differences in LVH may yield further insight into the MRFIT findings.
The design and methods of MRFIT have been reported extensively.3 Briefly, MRFIT was a large randomized primary prevention trial designed to determine effects on coronary heart disease mortality of a multifaceted intervention targeting smoking cessation, reduction in serum cholesterol, and stepped-care treatment of hypertension. Men aged 35 to 57 years were invited to attend 3 screening examinations in 22 clinical centers located in 18 US cities. Participants were selected based on their having an above-average risk of developing coronary heart disease, as determined from a risk score calculated as a function of serum cholesterol, diastolic blood pressure, and cigarette smoking. Participants who were free of coronary heart disease by history and examination, including a resting ECG, were eligible for randomization.
After providing consent, 12 866 men were randomly assigned to 2 groups, SI or UC. The SI program included counseling on smoking cessation, advice on dietary modifications to reduce cholesterol levels, and treatment for hypertension, whereas the UC group received treatment of their risk factors as considered on an individual basis by their usual sources of care within the community. Of those randomized, 8012 (62%) were classified as hypertensive at baseline, defined as average diastolic blood pressure level of ≥90 mm Hg or those who had been receiving antihypertensive drugs on entry. These 8012 men, also the focus of earlier reports, are included in this analysis.
Hypertension treatment for SI participants used a stepped-care approach, which included moderate salt reduction, weight loss, and antihypertensive drugs.10 Initial medication was an oral diuretic, either CTD or HCTZ, at a dose of 50 or 100 mg daily (the standard doses used at the time MRFIT was carried out). This was followed by stepwise addition of an antiadrenergic drug or a β-blocker (step 2), an arteriolar vasodilator (step 3), and guanethidine (step 4), designed to achieve and maintain a diastolic blood pressure goal of 80 to 89 mm Hg. The choice of which diuretic to initiate in SI participants was left to the discretion of the individual physician. Within many clinics there was a tendency to use predominantly one diuretic or the other, which enabled clinics to be categorized in earlier ecological analyses according to the diuretic predominantly used in that clinic (C-clinic: CTD; H-clinic: HCTZ).
Blood Pressure and Electrocardiographic Measurements
A random 0 sphygmomanometer was used to measure clinic blood pressure in the seated position after 5 minutes of rest, as detailed elsewhere.10 Two measurements were recorded at each of the second and third screening visits and annually thereafter throughout follow-up.
Procedures have been reported for recording resting electrocardiograms and for computerized measurements, including different measures of LVH.11–15 Briefly, standard 12-lead resting electrocardiograms were recorded at the second and third screening visits and then annually for the duration of follow-up. Standard limb electrodes were placed on the torso to permit exercise, as well as resting electrocardiograms, at baseline (at the third of 3 screening visits before randomization).12 This lead configuration was maintained for recording resting electrocardiograms at annual visits after randomization. The electrocardiograms were recorded on magnetic tape cassettes with standard 3-channel ECG machines (Marquette series 3500, Marquette Inc, Milwaukee, WI). The analog cassettes were processed by computer algorithm at the Computer ECG Center (Dalhousie University, Halifax, Nova Scotia, Canada) and were also visually coded at the ECG Coding Center at the University of Minnesota.
The following continuous measurements were used to define LVH: Sokolow-Lyon voltage (μV)= SV1 + max RV5/V616; Cornell voltage (μV)=SV3 + RaVL17; and Cornell voltage product (μV/ms)=Cornell voltage * QRS duration.17 The following computer measurements were used to calculate left ventricular mass (in grams) for whites: 0.023 Cornell voltage + 0.010 Sokolow-Lyon + 1.32 weight + 10.6, and for blacks: 0.0018 Cornell voltage + 0.51 JV5 + 1.45 weight + 17.4.18 In MRFIT, ≈3% of hypertensive men reported their race as something other than white or black, and left ventricular mass for these men is determined using the formula for whites.
We assessed the association between CTD and HCTZ use and LVH using 2 types of analyses, an ecological analysis, with the 22 MRFIT clinics classified (as in previous reports) as C-clinics, H-clinics, or switching clinics,4,5 and an individual participant analysis, with use of CTD, HCTZ, or antihypertensive medication other than CTD or HCTZ considered time-dependent covariates and updated annually. Both analyses were restricted to men hypertensive at baseline. The ecological analysis was carried out through 4 years of follow-up, the last measurement before the recommendation to switch all of the participants to 50 mg of CTD; it used information from both SI and UC participants who were hypertensive at entry into C- or H-clinics. The individual participant analysis was performed for the same 4-year period and also for the duration of the trial (84 months) taking into account individual changes in CTD and HCTZ prescriptions; it used information for all of the SI hypertensive men, including those in the switching clinics but not men in the UC group. Both the ecological and individual participant analyses were repeated excluding participants with major conduction defects, as evidenced by QRS ≥120 ms.19
The ecological analysis takes advantage of the randomization to SI and UC, which was carried out within clinic. As a consequence, SI and UC men within clinic are similar, on average, with respect to baseline characteristics, both those measured and not measured. For the ecological analysis, the estimand used to quantify differences between CTD and HCTZ for blood pressure and LVH is the difference in “intention-to-treat” estimands (SI-UC) for C-clinics versus H-clinics (ie, a difference of differences). Two potential limitations of this estimand are that the grouping variable, CTD or HCTZ, determined postrandomization, may be a marker for a factor other than CTD or HCTZ that is related to LVH, and there is the possibility of misclassification within clinic grouping (eg, not all SI participants in the C-clinics were treated for hypertension, and some used HCTZ).20 To assess the likelihood that the clinic label of C- or H- was because of another factor related to LVH, we carried out an analysis for men who were normotensive at entry. Because fewer of these men were initiated on antihypertensive drugs during follow-up, we hypothesized that differences among C- and H-clinics would be minimal and, if so, would provide support for the inference that differences between clinics for the hypertensive men are attributable primarily to differences in use of CTD and HCTZ.
Treatment groups were compared for blood pressure and continuous measures of LVH using longitudinal regression analyses (SAS PROC MIXED, SAS version 9.2). Within clinic groupings, analyses were stratified by clinic, with baseline level of assessed outcome included as a covariate. To determine whether SI-UC differences in LVH varied between C- and H-clinics, an interaction term (treatment group ×clinic grouping) was included in the regression models.
Individual Participant Analysis
For the individual participant analysis, the estimand of C-H differences in blood pressure and LVH is the difference during follow-up among the subset of hypertensive participants in the SI group who were prescribed these treatments. This estimator is subject to the usual type of confounding present in epidemiological analyses.
Longitudinal regression models were used to study the association of time-updated antihypertensive drug use and LVH. Analyses were also carried out for change in the measures of LVH at 48 months (ie, a single time point as in the ecological analysis instead of all of the follow-up time points). SI participants were categorized as being prescribed CTD, HCTZ, an antihypertensive drug other than CTD or HCTZ, or on no antihypertensive medications, and differences between CTD and HCTZ were assessed. Two models were considered. One model included the following baseline covariates, age, race, serum cholesterol, smoking status, body mass index, systolic blood pressure, use of antihypertensive medication, serum potassium, fasting serum glucose, serum creatinine, uric acid, proteinuria (≥1+), number of alcoholic drinks per week, clinic, and baseline level of the electrocardiographic measure considered. A second model included a propensity score. The propensity score, which was the probability of using HCTZ instead of CTD at 48 months as a function of baseline predictors, was determined using logistic regression. The aforementioned baseline covariates were used as predictors. Regression models similar to those described above were also used to study the relationship of change in LVH with change in blood pressure for SI participants.
Of the 8012 hypertensive men enrolled in MRFIT, 2112 (1046 SI and 1066 UC) were randomly assigned by 6 clinics that predominantly used CTD (ie, C-clinics); 3399 (1725 SI and 1674 UC) were randomly assigned by 9 clinics that predominantly used HCTZ (ie, H-clinics); and 2501 (1248 SI and 1253 UC) were randomly assigned by 7 clinics that initially used CTD and then switched to HCTZ during the first 4 years of the study (ie, switching clinics). As reported previously, the characteristics of SI and UC hypertensive men were well balanced at baseline.3 This was also the case for the SI-UC comparison within the 3 groups of clinics (data not shown).
Table 1 summarizes the baseline characteristics for the hypertensive participants overall and by clinic grouping. Baseline characteristics, including continuous measures of LVH, were similar for participants in C- and H-clinics. Participants in the switching clinics were less likely to be black and to smoke. Overall, average systolic and diastolic blood pressure levels were 141.2 and 95.5 mm Hg, respectively. Approximately 31% were taking antihypertensive medications at entry.
Use of CTD, HCTZ, and Other Antihypertensives in MRFIT
Figure 1 gives the percentage of SI participants prescribed CTD, HCTZ, or other treatment (neither CTD nor HCTZ) by month of follow-up, irrespective of clinic grouping. Use of CTD declined during the first 4 years, from ≈33.4% at 12 months to 26.6% at 48 months, whereas use of HCTZ increased from 27.4% at 12 months to a high of 42.5% at 48 months. After the protocol change, which occurred after all of the participants had completed the 48-month visit, use of CTD increased (to 50.8% by 84 months), whereas use of HCTZ declined to 14.7% by 84 months.
At 48 months, of SI hypertensive participants, 80% in C-clinics, 75% in H-clinics, and 76% in switching clinics were prescribed antihypertensive medication. The distribution of CTD and HCTZ use according to clinic grouping is summarized in Figure 2. Among C-clinics, use of CTD ranged from 43% to 58% (average: 48%), whereas among H-clinics, use of HCTZ ranged from 40% to 61% (average: 49.5%). In the switching clinics, 49.7% of SI participants were prescribed HCTZ and 21.2% were prescribed CTD. After the protocol change (ie, by the 72-month follow-up visit), these percentages were approximately reversed, among SI participants in the switching clinics, 51.4% were prescribed CTD and 16.8% were prescribed HCTZ (data not shown). Less than 10% of UC participants were prescribed CTD in each of the clinic groupings, and ≈35% were prescribed HCTZ at 48 months.
Additional antihypertensive medication was added to the diuretic used to achieve blood pressure goal. In C-clinics, 52.9% of SI participants at 48 months were prescribed a step 2 drug, 13.2% were prescribed step 3, and 0.6% were prescribed step 4. In SI participants of H-clinics, the corresponding percentages were 42.4%, 12.0%, and 0.7%.
Blood Pressure and LVH Measures Through 48 Months According to C- or H-Clinic
Changes in systolic and diastolic blood pressures were greater for SI compared with UC participants in both the C- and H-clinics. The SI-UC differences were greater within C-clinics compared with H-clinics for systolic (−10.4 versus −8.6 mm Hg; P=0.001 for clinic difference) and diastolic blood pressure (−6.5 versus −5.1 mm Hg; P<0.001 for clinic difference; Table 2). All of the continuous LVH measures were significantly lower in the SI compared with the UC group, regardless of clinic designation. Mean differences between SI and UC in these measures were all uniformly of greater magnitude in the C- compared with the H-clinics (P values for interaction between C- and H-clinics: Sokolow-Lyon, P=0.049; Cornell voltage, P=0.019; Cornell voltage product, P=0.071; left ventricular mass, P=0.002).
Analyses were also carried out for the men who were normotensive at entry to determine whether differences between C- and H-clinics were evident in this subgroup. Participants randomized to SI with repeated diastolic blood pressure ≥90 mm Hg during follow-up were placed on antihypertensives. Overall, the percentage of SI men who were normotensive at entry receiving antihypertensives at 48 months was much smaller (22.2%) than for men hypertensive at entry (76.5%). Likewise, although differences in use of CTD and HCTZ were evident among clinics (13.4% used CTD and 9.3% used HCTZ among C-clinics and 5.0% used CTD and 12.9% used HCTZ among H-clinics at 48 months), the percentages using these drugs were much smaller than for hypertensive men. Consequently, SI-UC differences in blood pressure change were smaller. Differences in systolic blood pressure for SI-UC averaged −2.6 mm Hg (SE=0.46) for C-clinics and −3.0 mm Hg (SE=0.34) for H-clinics (P=0.56 for difference), whereas SI-UC differences in diastolic blood pressure averaged −1.8 mm Hg (SE=0.30 mm Hg) and −2.1 mm Hg (SE=0.22 mm Hg), respectively, for C- and H-clinics (P=0.48 for difference). Differences in SI-UC in LVH did not vary significantly by C- or H-clinic grouping (Sokolow-Lyon, P=0.35; Cornell voltage, P=0.12; Cornell voltage product, P=0.26; left ventricular mass, P=0.49).
Only a small number of participants had a QRS duration ≥120 mm (11 participants at baseline and ≤12 at each annual visit). Exclusion of these participants did not impact the results.
Individual Participant Analysis With Time-Updated Use of CTD and HCTZ: All SI Hypertensives
Table 3 summarizes differences between CTD and HCTZ for the individual participant analysis of SI men; it is based on all of the annual follow-up measures of LVH through 48 and 84 months. For comparison, the estimand based on the clinic analysis shown in Table 2 is given in the last column. Similar to the clinic-level analysis, differences in blood pressure and in most measures of LVH significantly favored participants taking CTD. However, for Cornell voltage product, the CTD-HCTZ differences were significant in the opposite direction of the ecological analysis, and no significant difference was observed for the Cornell voltage measure between CTD and HCTZ. In contrast to the clinic analyses, we also found that QRS duration (the multiplicand to convert Cornell voltage to Cornell voltage product) was greater for CTD compared with HCTZ. Unadjusted (no covariate adjustment) estimates of the CTD-HCTZ differences were similar to those in Table 3 (data not shown). Likewise, when the covariate adjustment was expanded to include use of step 2, 3, and 4 drugs, the CTD-HCTZ differences in blood pressure and measures of LVH were similar.
A propensity score analysis that compares CTD and HCTZ for each of the LVH measures at 48 months is summarized in the online Data Supplement (please see http://hyper.ahajournals.org). The first step in this analysis was to determine factors associated with the use of HCTZ (versus CTD) among hypertensive men at 48 months (ie, the propensity for using HCTZ). Older age and use of antihypertensive treatment at entry were significantly associated with less use of CTD than HCTZ. No factor other than clinic was significantly associated with the use of HCTZ versus CTD. After adjustment of CTD versus HCTZ, differences in blood pressure, and LVH for propensity score, the results for LVH measures were similar to Table 3.
To explore the differences between clinic and individual participant analyses for Cornell voltage and product, we carried out a subgroup analysis according to the use of antihypertensive treatment at entry, the most important predictor of using CTD in the propensity score analyses. For the individual participant analyses, interaction P values were 0.22 for Cornell voltage and 0.29 for Cornell voltage product; for the clinic analyses, the interaction P values were 0.16 for Cornell voltage and 0.20 for Cornell voltage product.
Blood pressure reductions in the SI group with hypertension at entry were associated with decreases in each of the LVH measures (P<0.001 for each measure). The association of blood pressure change with LVH change was similar for SI participants in C- and H-clinics and for UC participants.
The presence of LVH in MRFIT has been independently associated with an increased risk of cardiovascular death.13–15 Our primary analysis found that SI-UC differences in continuous measures of electrocardiographic LVH were of greater magnitude in C-clinics than in H-clinics, suggesting a differential benefit in favor of treatment with CTD compared with HCTZ. When examined at the individual participant level, those receiving CTD also had less LVH, as assessed by Sokolow-Lyon and left ventricular mass criteria, than those receiving HCTZ. These findings occurred when both CTD and HCTZ were widely used through 48 months and persisted after participants were switched to CTD through long-term (84 months) follow-up. Our analyses offer new insight into the recent report from MRFIT concluding that CTD is more effective at reducing cardiovascular events than HCTZ.21
Our study results are consistent with earlier reports from MRFIT that have shown that overall LVH incidence was lower in the SI compared with the UC group and that reductions in blood pressure were associated with a reduced incidence of LVH.13 It is possible that the differences in LVH between CTD and HCTZ may have occurred as a consequence of improved blood pressure control in participants receiving CTD compared with HCTZ. The change in systolic blood pressure from baseline to 48 months between SI and UC groups was 1.7 mm Hg greater in participants of C-clinics than H-clinics; these differences were also consistent with differences for CTD versus HCTZ observed at the individual participant level. Importantly, office blood pressure measurements were taken in MRFIT during daytime hours and may not have reflected potentially larger blood pressure differences between regimens occurring over an entire 24-hour period.
Several limitations typical of post hoc analyses should be noted. Diuretic assignment in MRFIT was not randomized, leading to heterogeneity of diuretic use within the C- and H-clinics and the potential for misclassification within clinic grouping (ie, not all SI participants in the C-clinics were treated for hypertension, and some received HCTZ). However, unlike the individual participant analysis, confounding within the clinic designation is protected by overall randomization to SI or UC group. The results of the ecological analysis are further strengthened by the absence of differences in LVH between SI and UC groups in the analysis of men normotensive at entry, which supports the notion that hypertensive men were affected differently by use of CTD versus HCTZ. Second, higher doses of both CTD and HCTZ were used in MRFIT than what are currently used. Given established potency differences between the 2 agents, it is not clear whether the lower doses of CTD and HCTZ in common use today would give similar results. Third, LVH was assessed using electrocardiography rather than by echocardiography or MRI. Although these newer methods of cardiac imaging may be more sensitive, electrocardiographically determined LVH remains an established predictor of cardiovascular disease risk in both MRFIT and elsewhere.8,9,12–15
Lastly, the findings for Cornell voltage and product in the individual participant analysis were not consistent with those of the ecological (clinic-level) analysis. We conducted sensitivity analyses of the individual components of the Cornell LVH measures and observed that the discrepancy may result, in part, from a significant increase in QRS duration found in individual participant analyses but not in clinic analyses. We also assessed whether the CTD and HCTZ differences varied by use of antihypertensive treatment at entry, and the evidence for subgroup heterogeneity was weak.
Previous MRFIT reports have documented a lower incidence of LVH for SI compared to UC men, but these analyses did not assess possible differences related to use of CTD versus HCTZ. In ecological analyses, we found that SI-UC differences in electrocardiographically defined continuous measures of LVH were of greater magnitude in clinics with high use of CTD compared with those where HCTZ predominated. At the individual participant level for all of the SI hypertensives, both Sokolow-Lyon and left ventricular mass measures were also lower in users of CTD compared with HCTZ. Given the strong association between LVH and blood pressure, our findings suggest that differences in blood pressure lowering may explain the variation in mortality observed between CTD and HCTZ in MRFIT and also underscore the need for a randomized clinical outcome trial comparing the 2 agents.
Sources of Funding
The Multiple Risk Factor Intervention Trial was supported by a contract and grants from the National Heart, Lung, and Blood Institute. This funding included R01-HL68140.
We express appreciation to the MRFIT study participants and the many investigators who contributed to the design and conduct of the study.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
- Received August 16, 2011.
- Revision received September 1, 2011.
- Accepted October 5, 2011.
- © 2011 American Heart Association, Inc.
Multiple Risk Factor Intervention Trial Research Group. Mortality after 10 ½ years for hypertensive participants in the Multiple Risk Factor Intervention Trial. Circulation. 1990;82:1616–1628.
- Bartsch G,
- Broste S,
- Grandits G,
- Grimm RH,
- Neaton JD,
- Svendsen KH
- Ernst ME,
- Carter BL,
- Goerdt CJ,
- Steffensmeier JJ,
- Phillips BB,
- Zimmerman MB,
- Bergus GR
- Jain A,
- Tandri H,
- Dalal D,
- Chahal H,
- Soliman EZ,
- Prineas RJ,
- Folsom AR,
- Lima JA,
- Bluemke DA
- Zwietz C,
- Schneider B
- Wolf H,
- MacInnis PJ,
- Stock S,
- Helppi RK,
- Rautaharju PM
- MacMahon S,
- Collins G,
- Rautaharju P,
- Cutler J,
- Neaton J,
- Prineas R,
- Crow R,
- Stamler J
- Prineas RJ,
- Rautaharju PM,
- Grandits G,
- Crow R
- Hancock EW,
- Deal BJ,
- Mirvis DM,
- Okin P,
- Kligfield P,
- Gettes LS
- Dorsch MP,
- Gillespie BW,
- Erickson SR,
- Bleske BE,
- Weder AB