Meta-Analysis of Dose Response Characteristics of Hydrochlorothiazide and Chlorthalidone: Effects on Systolic Blood Pressure and Potassium

nature publishing group Meta-Analysis of Dose Response Characteristics of Hydrochlorothiazide and Chlorthalidone: Effects on Systolic Blood Pressure and Potassium Michael E. Ernst 1,2, Barry L. Carter 1,2, Shimin Zheng 2 and Richard H. Grimm Jr 3 5 Background Evidence supporting the benefit of low-dose thiazide-based regimens to reduce cardiovascular events is primarily derived from studies using chlorthalidone, yet low-dose hydrochlorothiazide (HCTZ) (12.5 25 mg) remains more widely prescribed. We sought to describe their comparative dose response relationships for changes in systolic blood pressure (SBP) and potassium. Methods PubMed from 1948 to July 2008 was systematically searched to identify clinical trials using either HCTZ or chlorthalidone monotherapies. A total of 108 clinical trials with HCTZ and 29 with chlorthalidone were analyzed. Data were pooled to evaluate the effects on SBP and potassium of both drugs throughout their respective dose response curves. Equivalence analysis was performed for the clinically recommended low-dose range of 12.5 25 mg, grouped by study duration, using the two one-sided tests procedure described by Schuirmann. Results When evaluated on a milligram-per-milligram basis using pooled data, chlorthalidone generally produces slightly greater reductions in SBP and potassium than HCTZ. In the low-dose range of 12.5 25 mg, equivalence analysis reveals that the reductions in SBP are not equivalent between the two drugs, using upper and lower equivalence bounds of 4 mm Hg. Within the same dosing range, the mean changes in potassium were determined to be equivalent when upper and lower equivalence bounds of 0.29 meq/l are used. Conclusions Equivalence analysis using data from several studies suggests that the SBP reductions achieved with HCTZ and chlorthalidone cannot be considered equivalent within the low-dose range currently recommended. However, within this dosing range, reductions in potassium can be considered equivalent. Keywords: blood pressure; chlorthalidone; diuretics; hydrochlorothiazide; hypertension; potassium Am J Hypertens 2010; 23:440-446 2010 American Journal of Hypertension, Ltd. Thiazides and the thiazide-like diuretic, chlorthalidone, have been important pharmacologic agents in the treatment of hypertension for over 50 years. 1 National treatment guidelines do not specify a thiazide of preference, 2 and although several are available, low-dose (12.5 25 mg/day) hydrochlorothiazide (HCTZ) is the thiazide predominately used in the United States. This is despite the fact that most clinical trials demonstrating reductions in cardiovascular events with thiazide-based regimens were performed using either higher doses of HCTZ (50 mg/day) than currently recommended or with low-dose (12.5 25 mg/day) chlorthalidone. 3 9 More 1 Department of Pharmacy Practice and Science, College of Pharmacy, The University of Iowa, Iowa City, Iowa, USA; 2 Department of Family Medicine, Roy J. and Lucille A. Carver College of Medicine, The University of Iowa, Iowa City, Iowa, USA; 3 Berman Center for Outcomes and Clinical Research, Minneapolis Medical Research Foundation, Minneapolis, Minnesota, USA; 4 Division of Clinical Epidemiology, Hennepin County Medical Center, Minneapolis, Minnesota, USA; 5 Department of Medicine, University of Minnesota Medical School, Minneapolis, Minnesota, USA. Correspondence: Michael E. Ernst (michael-ernst@uiowa.edu) Received 14 August 2009; first decision 21 September 2009; accepted 24 December 2009; advance online publication 28 January 2010. doi:10.1038/ajh.2010.1 2010 American Journal of Hypertension, Ltd. recently, low-dose HCTZ-based regimens were proven inferior to nonthiazide comparators in two large CV outcome-based trials. 10,11 The widespread adoption of low-dose HCTZ in clinical practice in the absence of supportive outcome evidence suggests that clinicians may consider a thiazide class benefit. To date, little direct comparison of HCTZ and chlorthalidone exists to prove otherwise. The two agents have important pharmacokinetic differences; chlorthalidone is a naturally long-acting agent, possessing a half-life of 45 60 h compared to 8 15 h for HCTZ, a characteristic which may produce relevant differences in pharmacodynamic effects. 12 Interest in chlorthalidone has increased recently, 1,13 19 but it remains minimally used. Besides lack of availability in practical fixed-dose combinations, concerns about greater loss of potassium persist, possibly influenced by the historical use of higher doses (50 100 mg/day) than now typically used. 5,6 Our objective was to systematically examine the literature to better characterize the comparative dose response characteristics of HCTZ and chlorthalidone monotherapies on systolic blood pressure (SBP) and potassium. 440 April 2010 VOLUME 23 NUMBER 4 440-446 AMERICAN JOURNAL OF HYPERTENSION

Dose Response of HCTZ and Chlorthalidone original contributions Methods Identification of clinical trials. A systematic literature search using the subject terms hydrochlorothiazide and chlorthalidone was performed in PubMed to identify articles published between 1948 and 31 July 2008. The search was limited to adult, human subject, English-language articles indexed under clinical trial. Study inclusion and exclusion criteria were identified a priori. Each article identified was selected for inclusion if it used either HCTZ or chlorthalidone as monotherapy and reported office SBP and serum or plasma potassium for baseline and at least one other time point during the study, or reported the mean change from baseline in these values. All study exclusions were made in order to achieve the most homogeneous comparison of the change in SBP and potassium occurring with monotherapy for either agent. First, only trials where antihypertensive efficacy was one of the primary outcomes were included. Pure pharmacokinetic trials or those not including analysis of antihypertensive efficacy were excluded. Second, study durations were limited to 1 year. Third, trials in which the thiazide was part of an initial fixed-dose combination, or was combined with a potassium-sparing agent (or supplemental potassium, if directly specified), were excluded. Similarly, stepped-care studies which included an initial thiazide monotherapy portion were excluded if they did not report measurements of SBP and potassium specifically for the thiazide monotherapy, thus avoiding the influence of any background therapy on the change in these parameters. Lastly, only the data from one article was selected when duplicate study data were encountered. Data abstraction. Following the initial literature search, abstracts were reviewed by one author (M.E.E.) for inclusion/ exclusion criteria. If an abstract did not provide sufficient information about whether office SBP or potassium measurements were reported, the full article was obtained for review. Due to the volume of articles yielded by the initial search, citations lacking abstracts were not considered for further review. All articles appearing to meet inclusion criteria were selected for data extraction. Articles found to not meet inclusion criteria during the data extraction process were discarded. The final list of articles were reviewed and data extracted independently by two authors (M.E.E. and B.L.C.); a random sample of ~10% of the articles were reviewed by both authors for consistency, and discrepancies were resolved by consensus. No formal duplicate data extraction process was performed. A data extraction form was created to record information from each eligible study. Data on study drug, trial design, sample size, study duration, specific dose (or end-of-study mean dose), dose-titration regimen, SBP measurements (actual or deltas), and serum or plasma potassium levels (actual or deltas) were collected. When multiple doses were used in a single study, as in stepped-titration studies, each dose and the corresponding data points for SBP and potassium for the specific time point prior to dose adjustment were entered as individual data entries. In this manner, one study could provide several SBP and potassium measurements corresponding to different dose levels and treatment durations for each drug. Some stepped titration studies did not report the specific SBP and potassium measurements, but reported only endof-study values for SBP and potassium. In this circumstance, the study was included if the data on number of patients at each dose level were available to calculate an overall mean dose corresponding to the end-of-study SBP and potassium measurements. Data analysis. The primary end point for our analysis was the change in SBP and potassium occurring throughout the dose response curves for HCTZ and chlorthalidone. These changes were computed as the difference in the potassium or SBP values at the final follow-up (or specific time point if multiple time points were provided) compared to the baseline or initial measurement. To compare treatment effects between the two drugs, we pooled the data to determine the mean change in SBP and potassium for the following tertiles of dose: 12.5 dose 25 mg, 25 < dose 37.5 mg, and 37.5 < dose 50 mg, and grouped by study durations of 4 weeks, 4 < weeks 8, 8 < weeks 12, and 12 < weeks 52. The assumption of a common variance parameter across all of the studies was investigated using Bartlett s test; this test was not statistically significant, indicating lack of hetero geneity. A fixed-effects analysis was employed, with all statistical comparisons weighted by the number of subjects in the particular dose interval and study duration. Equivalence testing was performed on the changes in SBP and potassium for the two drugs using the two one-sided tests method described by Schuirmann 20 for the low-dose range inclusive of 12.5 25 mg, and grouped by study duration. In contrast to superiority testing, equivalence analysis determines whether the mean difference in a variable of interest between two populations is so small, that one population mean is considered practically equivalent to the second population mean. Accordingly, we hypothesized that the mean changes in SBP and potassium were equivalent for HCTZ and chlorthalidone within the 12.5 25 mg dose range. A prespecified equivalence interval of 4, +4 mm Hg was chosen for SBP and 0.29, +0.29 for potassium, to represent clinically relevant upper and lower bounds of the true difference between the treatments. 90% Confidence intervals for the true difference between the two treatments was then constructed based on the data, and two, one-sided statistical tests conducted to establish equivalence. According to the method of Schuirmann, if the calculated confidence interval for the true difference between the two treatments is included in the prespecified equivalence interval, then the null hypothesis of nonequivalence is rejected and equivalence is concluded; otherwise, equivalence cannot be concluded. The null hypothesis of nonequivalence was rejected at the α = 0.05 level. All analyses were performed using SAS (version 9.2.; SAS Institute, Cary, NC). Results Clinical trial characteristics The initial search produced 1,713 citations for HCTZ and 366 citations for chlorthalidone. All available abstracts of these AMERICAN JOURNAL OF HYPERTENSION VOLUME 23 NUMBER 4 april 2010 441

Dose Response of HCTZ and Chlorthalidone citations were reviewed. Of the HCTZ abstracts, 257 appeared to meet inclusion criteria and full text was obtained of these articles for possible data extraction. For chlorthalidone, 59 abstracts appeared to meet inclusion criteria and full text of these articles was obtained. Review of the full text articles further reduced the final number to 108 evaluable studies for HCTZ and 29 for chlorthalidone. The most common reasons for exclusion at this step are shown in Figure 1, and include the following: lack of either SBP or serum potassium measurements linked to a specific or mean dose, thiazide used in combination with another agent rather than monotherapy, and insufficient information to determine mean dose corresponding to end-of-study SBP or serum potassium values (for stepped-titration studies). Descriptive information from the trials is provided in Table 1. Several thousand observations for dose, SBP, and potassium were available for evaluation. A wide range of doses were used in the studies with an overall mean of 42.7 mg for HCTZ (median, 33.0 mg; range, 3 450 mg) and 31.6 mg for chlorthalidone (median, 25 mg; range, 12.5 200 mg). Pooled mean baseline SBPs and serum potassium levels were similar for both agents. The median change in SBP associated with the median dose of HCTZ was 17 mm Hg, compared to 26 mm Hg for chlorthalidone. Pooled data comparisons of changes in SBP and potassium Figure 2 shows the relationship between the dose of the drugs and their mean change in SBP and potassium plotted using the unweighted, pooled study results for doses 100 mg and treatment durations 52 weeks. Throughout the dosing range of 12.5 100 mg, chlorthalidone reduces SBP in the approximate range of 15 mm Hg to 25 mm Hg, compared to 10 mm Hg to 25 mm Hg for HCTZ. Chlorthalidone produces slightly greater reduction in SBP throughout the range of 12.5 50 mg, at which point HCTZ appears to lower SBP to a greater extent within the upper dosing range. An exception occurs at the 25 mg dose, where the mean (±s.d.) decrease in SBP is 15.9 (3.8) for HCTZ and 15.5 (4.9) for chlorthalidone. The correlation between dose and SBP reduction was highly significant Table 1 Pooled summary of HCTZ and chlorthalidone clinical trials included in the analysis Antihypertensive trial characteristics HCTZ (n = 108) Chlorthalidone (n = 29) Total data points (SBP and serum 6,063 4,380 potassium), all subjects a Duration (weeks), mean (±s.d.) 10.6 (8.5) 32.2 (21.3) Dose (mg), mean (±s.d.) 42.7 (37.1) 31.6 (25.2) Dose (mg), median 33.0 25.0 Dose (mg), range 3 450 12.5 200 Baseline SBP (mm Hg), mean (±s.d.) 162.8 (8.6) 165.8 (9.0) Change in SBP (mm Hg), mean (±s.d.) 17 (5.6) 23 (6.7) Change in SBP (mm Hg), median 17 26 Baseline potassium (meq/l), mean (±s.d.) Change in potassium (meq/l), mean (±s.d.) 4.22 (0.12) 4.38 (0.14) 0.36 (0.19) 0.45 (0.16) Change in potassium (meq/l), median 0.31 0.40 a Some studies included dose-titrations and reported SBP and serum potassium data at different time points throughout the study; each subject could therefore provide more than one data point for SBP and serum potassium in these studies. HCTZ, hydrochlorothiazide; SBP, systolic blood pressure. Initial literature search HCTZ: n = 1,713 Chlorthalidone: n = 366 Full text articles obtained HCTZ: n = 257 Chlorthalidone: n = 59 Excluded on citation review Thiazide used in combination Antihypertensive efficacy not a primary study outcome No abstract available for review Duplicate citation SBP 30 25 20 15 10 CLD HCTZ Final articles extracted HCTZ: n = 108 Chlorthalidone: n = 29 Figure 1 Flow diagram for reasons trials were excluded. Excluded on full text review No SBP or serum potassium values provided HCTZ = 113 Chlorthalidone = 15 No thiazide monotherapy arm HCTZ = 15 Chlorthalidone = 2 Antihypertensive efficacy not a primary study outcome HCTZ = 3 Chlorthalidone = 6 Dose or change in SBP or serum potassium indeterminable HCTZ = 15 Chlorthalidone = 3 Duplicate citation HCTZ = 3 Chlorthalidone = 4 5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 K 10 20 30 40 50 60 70 80 90 100 Dose Figure 2 Mean change in SBP (mm Hg) and potassium (meq/l) for chlorthalidone and HCTZ by dose (mg) using pooled data from all studies and time points. (Sample size at each data point corresponds in a relative manner to the size of the circle or triangle point. Doses >100 mg were not plotted due to limited sample size). 442 april 2010 VOLUME 23 NUMBER 4 AMERICAN JOURNAL OF HYPERTENSION

Dose Response of HCTZ and Chlorthalidone original contributions Table 2 Mean change (±s.d.) in SBP and serum potassium for HCTZ and chlorthalidone at different dose levels and study durations Dose range (x, in mg) ΔSBP (mm Hg) ΔK (meq/l) (N: HCTZ, chlorthalidone) HCTZ Chlorthalidone HCTZ Chlorthalidone All study durations, 52 weeks 12.5 x 25 (2,848, 2,995) 14 (4.1) 24 (6.7)* 0.24 (0.09) 0.39 (0.07)* 25 < x 37.5 (212, 324) 16 (4.7) 26 (0.0)* 0.50 (0.35) 0.50 (0.00) 37.5 < x 60 (1,665, 787) 18 (5.8) 20 (6.6)* 0.43 (0.16) 0.60 (0.19)* 4 Weeks 12.5 x 25 (891, 206) 16 (3.0) 14 (8.2)* 0.24 (0.06) 0.30 (0.09)* 25 < x 37.5 NA NA NA NA 37.5 < x 60 (204, 251) 11 (6.5) 15 (1.1)* 0.57 (0.16) 0.59 (0.19) 4 Weeks < duration 8 weeks 12.5 x 25 (1,225, 406) 13 (3.6) 17 (2.8)* 0.25 (0.09) 0.35 (0.11)* 25 < x 37.5 NA NA NA NA 37.5 < x 60 (709, 105) 21 (4.5) 20 (5.2)* 0.38 (0.16) 0.46 (0.26)* 8 Weeks < duration 12 weeks 12.5 x 25 (302, 340) 11 (2.6) 14 (3.8)* 0.25 (0.14) 0.40 (0.12)* 25 < x 37.5 NA NA NA NA 37.5 < x 60 (715, 217) 18 (4.6) 19 (7.1) 0.44 (0.10) 0.67 (0.19)* 12 Weeks < duration 52 weeks 12.5 x 25 (430, 2,043) 19 (3.1) 28 (2.6)* 0.24 (0.07) 0.40 (0.01)* 25 < x 37.5 (89, 324) 21 (4.1) 26 (0.0)* 0.74 (0.44) 0.50 (0.00)* 37.5 < x 60 (28, 214) 13 (3.0) 27 (3.4)* 0.74 (0.23) 0.61 (0.03)* ΔK, change in serum potassium; ΔSBP, change in systolic blood pressure; HCTZ, hydrochlorothiazide; NA, too few data points available for comparison. *P < 0.05 for comparison, HCTZ vs. chlorthalidone; data weighted by sample size for the respective dose interval and study duration. for both drugs (HCTZ r 2 = 0.229, chlorthalidone r 2 = 0.137; P < 0.001 for both drugs separately). The change in potassium is slightly greater for chlorthalidone throughout most of the dose response curve, although absolute differences at any dose point are approximately 0.1 to 0.2 meq/l. Overall reductions range from a low of approximately 0.3 meq/l to high of 0.6 meq/l for chlorthalidone compared to 0.2 to 0.6 meq/l for HCTZ. The correlation of dose and potassium reduction was again highly significant for both drugs (HCTZ r 2 = 0.519, chlorthalidone r 2 = 0.135; P < 0.001 for both drugs separately). Table 2 shows the mean changes in SBP and potassium for tertiles of dose grouped by study duration. Chlorthalidone produces statistically greater SBP reduction than HCTZ in each dose partition when data from all study durations are pooled. When examined by specific study durations, chlorthalidone generally resulted in greater reduction in SBP with the exception of the inclusive 12.5 25 mg dose range for study durations 4 weeks and in the 37.5 < dose 60 mg range for durations of 4 < weeks 8. The reduction in serum potassium is statistically greater for chlorthalidone compared to HCTZ at all dose intervals when all study durations are pooled, with the exception of the 25 < dose 37.5 mg range where they are identical. Examination by study durations reveals that chlorthalidone produces statistically greater reduction in potassium at most dose levels, except at the 37.5 < dose 60 mg range for treatment durations 4 weeks, and the 25 < dose 37.5 mg and 37.5 < dose 60 mg range for study durations of 12 < weeks 52. Schuirmann s equivalence analysis Table 3 shows the results of Schuirmann s two one-sided tests procedure testing equivalence of the change in SBP, performed for the inclusive 12.5 25 mg dose interval and grouped by study duration. With the exception of durations 4 weeks, the greater mean reduction in SBP was observed for chlorthalidone, ranging from a low of 14 mm Hg to a high of 28 mm Hg, compared to 11 to 19 for HCTZ. For each study duration, the 90% confidence intervals for the true difference crossed either an upper or lower bound of the prespecified equivalence interval; thus, we failed to reject the null hypothesis that the treatments are not equivalent, and equivalence could not be concluded for the change in SBP between the two drugs within this low-dose range. The results of Schuirmann s two one-sided tests procedure testing equivalence for change in potassium, performed for the inclusive 12.5 25 mg dose interval and grouped by study duration, are shown in Table 4. Although chlorthalidone produced a slightly greater mean change in potassium for each study duration, ranging from 0.3 meq/l to 0.4 meq/l, the 90% confidence intervals for the true difference did not cross the upper or lower bounds of the prespecified equivalence interval for AMERICAN JOURNAL OF HYPERTENSION VOLUME 23 NUMBER 4 april 2010 443

Dose Response of HCTZ and Chlorthalidone Table 3 Equivalence analysis of weighted change in SBP for dose range 12.5 25 mg for HCTZ compared to chlorthalidone Study duration (weeks) Drug Sample size Mean (s.d.) ΔSBP, mm Hg 90% CI of true difference Schuirmann s equivalence? (P value) 4 CLD 206 14.0 (8.2) 4.4, 7.4 No (0.24) HCTZ 891 15.5 (3.0) 4 < x 8 CLD 406 17.0 (2.8) 6.9, 1.9 No (0.61) HCTZ 1,225 12.5 (3.6) 8 < x 12 CLD 340 14.4 (3.8) 5.5, 0.5 No (0.19) HCTZ 302 11.4 (2.6) 12 < x 52 CLD 2,043 27.6 (2.6) 13.9, 3.3 No (0.92) HCTZ 430 19.0 (3.1) 4, +4 mm Hg set as the prespecified equivalence interval for the difference between treatments. Null hypothesis: the change in SBP for chlorthalidone is not equivalent to HCTZ for the given study duration within the dose range 12.5 25 mg. With these P values, we fail to reject the null hypothesis for each of four specified durations within the dose range 12.5 25 and cannot conclude the SBP changes are equivalent. CI, confidence interval; CLD, chlorthalidone; HCTZ, hydrochlorothiazide; SBP, systolic blood pressure. Table 4 Equivalence analysis of weighted change in potassium for dose range 12.5 25 mg for HCTZ compared to chlorthalidone Study duration (weeks) Drug Sample size Mean (s.d.) Δ potassium, meq/l 90% CI of true difference Schuirmann s equivalence? (P value) 4 CLD 206 0.30 (0.09) 0.15, 0.02 Yes (<0.001) HCTZ 891 0.24 (0.06) 4 < x 8 CLD 406 0.35 (0.11) 0.17, 0.03 Yes (<0.001) HCTZ 1,225 0.25 (0.09) 8 < x 12 CLD 340 0.40 (0.12) 0.29, 0.02 Yes (0.046) HCTZ 302 0.25 (0.14) 12 < x 52 CLD 2,043 0.40 (0.01) 0.23, 0.09 Yes (0.008) HCTZ 430 0.24 (0.07) 0.29, +0.29 meq/l set as the prespecified equivalence interval for the difference between treatments. Null hypothesis: the change in potassium for chlorthalidone is not equivalent to HCTZ for the given duration within the dose range 12.5 25 mg. With these P values, the null hypothesis is rejected for each of four specified durations within the dose range 12.5 25 mg, indicating the changes in potassium are equivalent. CI, confidence interval; CLD, chlorthalidone; HCTZ, hydrochlorothiazide. any study duration. Thus, the null hypothesis was rejected and the alternative hypothesis accepted indicating that potassium reductions are equivalent between the two drugs within this dose range. Discussion The principle findings of our study suggest that chlorthalidone and HCTZ do not result in equivalent reductions in SBP within the dose range (12.5 25 mg) most commonly used in practice. Although chlorthalidone produces slightly greater loss of potassium within this same dose range, the results of formal equivalence testing indicate the reductions in potassium can be considered practically equivalent. These results should reassure clinicians that they should not avoid chlorthalidone out of concern for more clinically relevant potassium loss than HCTZ at commonly used doses. The findings of our study do not preclude the possibility of a subgroup of patients who may experience more dramatic potassium loss with chlorthalidone. Interestingly, the r 2 for the relationship between dose and potassium loss was much higher for HCTZ than chlorthalidone, which suggests that chlorthalidone s effects on potassium are subject to more variability than can be explained primarily by dose. It is unclear what mechanisms may be responsible for this, but a recent study reporting chlorthalidone to be an unexpectedly strong carbonic anhydrase inhibitor suggested this could contribute to its side effect profile. 21 Nevertheless, because both HCTZ and chlorthalidone can cause hypokalemia, appropriate monitoring is warranted whenever either agent is used. An interesting finding in Figure 2 is the nearly identical reduction in SBP observed at the 25 mg dose for both drugs, and further, the finding that the 12.5 mg dose of chlorthalidone seems to produce greater reduction in SBP than the 25 mg dose. Because these are plotted using the unweighted, pooled results combining all study durations, it is possible that this finding may be an inaccurate estimation of the true effect size. The overall reduction plotted for the 25 mg dose is derived from studies which started at 12.5 mg and up-titrated to 25 mg, and others that started at 25 mg. The reductions reported at 25 mg for studies that began with 12.5 mg doses would reflect a longer overall duration of treatment, which 444 april 2010 VOLUME 23 NUMBER 4 AMERICAN JOURNAL OF HYPERTENSION

Dose Response of HCTZ and Chlorthalidone original contributions is generally believed to influence the chronic response to thiazides. It is not entirely clear how clinician preference for HCTZ over chlorthalidone was formed. It is possible that experiences observed with higher doses may have contributed, particularly with regard to perceptions about differences in their effects on potassium. As observed in Figure 2, differences in potassium loss between HCTZ and chlorthalidone appear greatest for doses between 50 and 75 mg. A retrospective study published in 1981, which included doses in this range, concluded that chlorthalidone produced greater incidence and degree of hypokalemia than HCTZ; 22 however, the mean serum potassium levels after therapy were similar for twice the dose of HCTZ compared to chlorthalidone. It has been suggested that chlorthalidone is 1.5 2.0 times as potent in antihypertensive effect as HCTZ, primarily based on a few, small direct comparative trials where HCTZ at twice the dose achieved similar lowering of SBP as chlorthalidone. 23 25 Discussions about potency differences, although strictly based on antihypertensive effects, may influence perceptions of potassium loss as well. The strengths of our present analysis include that it is the most comprehensive examination of the dose response relationship for the effects on SBP and potassium of HCTZ and chlorthalidone, using pooled data obtained from a large sample of patients. Because both agents are effective antihypertensives, clinicians faced with the decision of which agent to prescribe may be more concerned with practical differences between the two drugs rather than statistical superiority. In this case, equivalence analysis is an appropriate method of comparing two effective therapies. Several limitations are noted with our study. First, we only evaluated changes from baseline for SBP and potassium for each drug as a monotherapy. Blood pressure reductions for HCTZ and chlorthalidone may be equivalent when each is added to an existing antihypertensive regimen, or when other antihypertensives are added to them. It is also possible that the true effect sizes or differences may be smaller when adjusted for placebo or active comparator. Our data should not be extrapolated beyond the change expected from baseline for these agents as monotherapies. Second, factors such as ethni city, diet, and comorbidities may influence response to thiazides. We were unable to adjust for these covariates in our analysis as this information was not uniformly available. However, our analysis was performed on a large sample of observations for each drug, which should reduce potential bias assuming randomness in the distribution of these covariates. Third, we were unable to examine rates of incident hypokalemia as this information was also not widely reported. Incident hypokalemia is a dichotomous end point with thresholds usually varying from 3.0 to 3.5 meq/l depending on the study. Although this end point is of interest, we believe the mean change from baseline in serum potassium can also provide instructive clinical information, in that it can be extrapolated to the individual patient baseline level prior to therapy to anticipate the likelihood of developing hypokalemia. In sum, the degree of SBP lowering between chlorthalidone and HCTZ as monotherapies cannot be considered equivalent in the most commonly used low-dose range of 12.5 25 mg. However, within this dosing range, reductions in potassium are equivalent for the two drugs. Avoidance of prescribing chlorthalidone in favor of HCTZ should not be justified on the basis of relevant differences in effects on potassium. Disclosure: M.E.E. has received consultant fees from Takeda Pharmaceuticals. R.H.G. has received consultant fees from Pfizer, Merck, and Forrest Pharmaceuticals. B.L.C. and S.Z. declared no conflict of interest. 1. Ernst ME, Moser M. Use of diuretics in patients with hypertension. N Engl J Med 2009; 361:2153 2164. 2. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr, Roccella EJ; National Heart, Lung, and Blood Institute Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure; National High Blood Pressure Education Program Coordinating Committee. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report. JAMA 2003; 289:2560 2572. 3. Amery A, Birkenhäger W, Brixko P, Bulpitt C, Clement D, Deruyttere M, De Schaepdryver A, Dollery C, Fagard R, Forette F. Mortality and morbidity results from the European Working Party on High Blood Pressure in the Elderly trial. Lancet 1985; 1:1349 1354. 4. Medical Research Council Working Party. Medical Research Council trial of treatment of hypertension in older adults: principal results. BMJ 1992; 304: 405 412. 5. The Multiple Risk Factor Intervention Trial Research Group (MRFIT). Mortality rates after 10.5 years for hypertensive participants in the Multiple Risk Factor Intervention Trial. Circulation 1990;82:1616 1628. 6. Hypertension Detection and Follow-up Program Cooperative Group. Five-year findings of the Hypertension Detection and Follow-up Program I. JAMA 1979; 242:2562 2571. 7. SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension. Final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA 1991; 265:3255 3264. 8. Neaton JD, Grimm RH Jr, Prineas RJ, Stamler J, Grandits GA, Elmer PJ, Cutler JA, Flack JM, Schoenberger JA, McDonald R. Treatment of Mild Hypertension Study. Final results. Treatment of Mild Hypertension Study Research Group. JAMA 1993; 270:713 724. 9. ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. Major outcomes in high-risk hypertensive patients randomized to angiotensinconverting enzyme inhibitor or calcium channel blocker vs diuretic: The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA 2002;288:2981 2997. 10. Wing LM, Reid CM, Ryan P, Beilin LJ, Brown MA, Jennings GL, Johnston CI, McNeil JJ, Macdonald GJ, Marley JE, Morgan TO, West MJ; Second Australian National Blood Pressure Study Group. A comparison of outcomes with angiotensin-converting--enzyme inhibitors and diuretics for hypertension in the elderly. N Engl J Med 2003; 348:583 592. 11. Jamerson K, Weber MA, Bakris GL, Dahlöf B, Pitt B, Shi V, Hester A, Gupte J, Gatlin M, Velazquez EJ; ACCOMPLISH Trial Investigators. Benazepril plus amlodipine or hydrochlorothiazide for hypertension in high-risk patients. N Engl J Med 2008; 359:2417 2428. 12. Carter BL, Ernst ME, Cohen JD. Hydrochlorothiazide versus chlorthalidone: evidence supporting their interchangeability. Hypertension 2004; 43:4 9. 13. Sica DA. Chlorthalidone: has it always been the best thiazide-type diuretic? Hypertension 2006; 47:321 322. 14. Mitka M. Experts argue not all diuretics the same. JAMA 2007; 298:31. 15. Taler SJ. Should chlorthalidone be the diuretic of choice for antihypertensive therapy? Curr Hypertens Rep 2008; 10:293 297. 16. Elliott WJ, Grimm RH Jr. Using diuretics in practice--one opinion. J Clin Hypertens (Greenwich) 2008; 10:856 862. 17. Saklayen MG. Which diuretic should be used for the treatment of hypertension? Am Fam Physician 2008; 78:444, 446. 18. Ernst ME, Carter BL, Basile JN. All thiazide-like diuretics are not chlorthalidone: putting the ACCOMPLISH study into perspective. J Clin Hypertens (Greenwich) 2009; 11:5 10. AMERICAN JOURNAL OF HYPERTENSION VOLUME 23 NUMBER 4 april 2010 445

Dose Response of HCTZ and Chlorthalidone 19. Kaplan NM. The choice of thiazide diuretics: why chlorthalidone may replace hydrochlorothiazide. Hypertension 2009; 54:951 953. 20. Schuirmann DJ. A comparison of the two one-sided tests procedure and the power approach for assessing the equivalence of average bioavailability. J Pharmacokinet Biopharm 1987; 15:657 680. 21. Temperini C, Cecchi A, Scozzafava A, Supuran CT. Carbonic anhydrase inhibitors. Sulfonamide diuretics revisited--old leads for new applications? Org Biomol Chem 2008; 6:2499 2506. 22. Sumiye L, Vivian AS, Frisof KB, Podany EC. Potassium loss associated with hydrochlorothiazide versus chlorthalidone. Clin Ther 1981; 4:308 320. 23. Bowlus WE, Langford HG. A comparison of the antihypertensive effect of chlorthalidone and hydrochlorthiazide. Clin Pharmacol Ther 1964; 5: 708 711. 24. Finnerty FA Jr. A double-blind study of chlorthalidone and hydrochlorothiazide in an outpatient population of moderate hypertensives. Angiology 1976; 27: 738 744. 25. Ernst ME, Carter BL, Goerdt CJ, Steffensmeier JJ, Phillips BB, Zimmerman MB, Bergus GR. Comparative antihypertensive effects of hydrochlorothiazide and chlorthalidone on ambulatory and office blood pressure. Hypertension 2006; 47:352 358. 446 april 2010 VOLUME 23 NUMBER 4 AMERICAN JOURNAL OF HYPERTENSION