Comparison of Cardiovascular Outcomes Between Chlorthalidone and Hydrochlorothiazide in Hypertensive Patients

ABSTRACT

Chlorthalidone (CLTD) and hydrochlorothiazide (HCTZ) are widely used thiazide diuretics for hypertension management. This study aimed to evaluate and compare the cardiovascular outcomes of patients treated with CLTD versus HCTZ. This multicenter, retrospective cohort study utilized data from the Korea University Medical Center, derived from electronic health records. A total of 14 257 hypertensive patients treated with either CLTD (n = 1920) or HCTZ (n = 12 337) were identified. Patients were matched 1:1 using propensity scores, resulting in 1606 patients in each treatment group. Demographic and clinical characteristics, incidence of major adverse cardiovascular events (MACE), and safety profiles were analyzed. Baseline characteristics after propensity score matching were well balanced between the two groups. The average age was 61.8 ± 14.6 years for CLTD users, with 59.3% being male. The 3‐year MACE occurred in 1.2% of the CLTD group compared with 1.4% of the HCTZ group (hazard ratio 0.91, p = 0.77). For secondary outcomes, cardiovascular mortality was 0.2% in both groups (p = 0.92). Myocardial infarction occurred in 0.3% of CLTD users and 0.4% of HCTZ users (p = 0.65). The incidence of hypokalemia was 19.2% in the CLTD group versus 16.7% in the HCTZ group (p = 0.07). In conclusion, in hypertensive patients, CLTD and HCTZ showed comparable cardiovascular outcomes and safety profiles.

Abbreviations

CLTD

chlorthalidone

EHR

electronic health records

HCTZ

hydrochlorothiazide

MACE

major adverse cardiovascular events

MI

myocardial infarction

OMOP‐CDM

Observational Medical Outcomes Partnership‐Common Data Model

PS

propensity score

1. Introduction

Hypertension represents a critical global public health issue and is a leading risk factor for cardiovascular disease. Thiazide diuretics are recommended as first‐line treatments for hypertension in most international guidelines, with hydrochlorothiazide (HCTZ) being the most frequently prescribed [1, 2]. However, recent guidelines have increasingly favored chlorthalidone (CLTD) over HCTZ due to its longer half‐life and superior evidence for cardiovascular benefits [3]. Despite this preference, observational studies have shown that the two drugs reduced cardiovascular events at a similar rate, and CLTD may be associated with an increased risk of adverse events, including hypokalemia [4, 5].

One large pragmatic trial with random assignment has shown that patients who received CLTD did not have a lower occurrence of major cardiovascular outcome events or non‐cancer‐related deaths than patients who received HCTZ [6]. The study's limitations include the fact that all participants were 65 or older, predominantly male and mostly white, with very few Asian participants. Therefore, the results may not be applicable to younger individuals, females, or other racial groups. This study aims to fill this gap by comparing the anti‐hypertensive effects of CLTD and HCTZ in patients with hypertension and assessing their impact on the incidence of cardiovascular events.

2. Methods

2.1. Study Design and Population

This multicenter, retrospective cohort study utilized the Observational Medical Outcomes Partnership‐Common Data Model (OMOP‐CDM) database, based on electronic health records (EHR) from three tertiary hospitals in South Korea: Korea University Anam Hospital, Korea University Guro Hospital, and Korea University Ansan Hospital. The OMOP‐CDM database encompasses extensive healthcare service data, including demographic details, diagnoses, prescriptions, medical equipment, and procedural records. Individual clinical diagnoses were classified using the International Classification of Diseases 10th Revision (ICD‐10). (Figure 1) Within the OMOP‐CDM database, every diagnosis is allocated a distinct concept identifier (ID) that corresponds to a specific ICD‐10 code. The prescribed medications were accurately documented and categorized based on their chemical compositions and dosages. Data were stored on an SQL server and extracted via direct SQL queries. This study was approved by the institutional review board of each hospital, and the requirement for informed consent was waived due to the use of an anonymized retrospective study design that posed minimal risk to study participants.

FIGURE 1.

Study flowchart. From 21 482 patients with CLTD or HCTZ prescriptions between January 2017 and June 2023, a final propensity score‐matched cohort of 3212 patients was selected for analysis. CLTD, chlorthalidone; Cr, creatinine; HCTZ, hydrochlorothiazide; HDL‐C, high‐density lipoprotein cholesterol; MACE, major adverse cardiovascular events; SBP, systolic blood pressure; TC, total cholesterol.

Patients who were prescribed either CLDT or HCTZ for at least 30 days between January 1, 2017 and June 30, 2023, were included in this study. The index date was defined as the date of the first prescription. Patients who had been prescribed both HCTZ and CLDT were assigned to the CLDT group if their CLTD prescription occurred after the HCTZ prescription, with the index date defined as their first CLDT prescription date. Patients were excluded if they received HCTZ after the CLDT prescription or before the occurrence of an outcome event. Exclusion criteria were also as follows: patients under 20 years of age at the index date, no hospital visit records for at least 6 months prior to the index date, occurrence of an outcome event or hospitalization within 30 days from the index date, and missing values for key variables including blood pressure (BP), creatinine, total cholesterol (TC), and high‐density lipoprotein cholesterol (HDL‐C). The study was approved by the institutional review board of each participating center (Korea University Anam Hospital; 2023AN0444). Considering the retrospective nature of the study and the use of anonymized data, the requirement for written informed consent was waived.

2.2. Study Variables and Outcomes

Diabetes mellitus (DM) was identified based on the glycated hemoglobin (HbA1c) level of 6.5% or higher, the use of oral hypoglycemic medications, or the presence of a special concept ID for DM in the OMOP‐CDM database. Dyslipidemia was defined as serum TC ≥ 240 mg/dL, LDL‐C ≥ 160 mg/dL, triglyceride ≥ 200 mg/dL, or HDL‐C < 40 mg/dL; use of lipid‐lowering drugs; or OMOP‐CDM concept ID for dyslipidemia. During the daytime, blood samples were obtained for routine laboratory analysis after an overnight fasting period. Colorimetric enzymatic technique was employed to evaluate serum lipid profiles.

The primary outcome was the occurrence of major adverse cardiovascular events (MACE), which included cardiovascular mortality, myocardial infarction (MI), stroke, and hospitalization for heart failure for 30 days. Patients were followed up for 3 years after the index day, and data describing the dates and causes of mortality were extracted from death certificates contained in the EHR. Cardiovascular mortality was defined as hospitalization due to MI, heart failure, or stroke within 30 days of death. MI was defined as a serum creatine kinase‐myocardial band level that was above the upper limit of normal and exhibited a rising and/or decreasing pattern. Stroke was defined as brain magnetic resonance imaging that revealed acute, subacute, or recent cerebral infarction or a matching OMOP‐CDM concept ID. Hospitalization for heart failure was defined as a case of ≥3 days in the hospital, including the emergency department, and an N‐terminal pro‐brain natriuretic peptide level of ≥300 pg/mL during hospitalization. Secondary outcomes were the individual components of the primary outcome and coronary revascularization. Safety outcomes at 3 years included hyponatremia (serum sodium <135 mEq/L) and hypokalemia (serum potassium <3.5 mEq/L). Clinical events for the 24 negative control outcomes were determined using clinical diagnosis codes, including foot fracture, leg fracture, hip fracture, forearm fracture, hand fracture, traumatic subdural hemorrhage, peptic ulcer, pneumonia, urinary tract infection, cystitis, conjunctivitis, herpes zoster, contact dermatitis, urticaria, psoriasis, cataract, glaucoma, lung cancer, gastric cancer, liver cancer, colorectal cancer, prostate cancer, breast cancer, rheumatoid arthritis, and osteoporosis.

2.3. Statistical Analysis

Categorical variables are represented by numerical values in the form of percentages, while continuous variables are represented by means and standard deviations. The comparison of continuous variables was conducted using both parametric (Student's t‐test) and non‐parametric (Mann–Whitney U) tests. Categorical variables were compared using either the χ ² test or Fisher's exact test, depending on the circumstances. To analyze PS matching, a multivariate logistic regression model was used to assess the probability of receiving CLTD. This model considered several factors, such as demographic data, past medical history, concurrent usage of drugs, and laboratory measurements. We paired each patient in the CLDT user group with a patient in the HCTZ user group at a 1:1 ratio using the nearest‐neighbor method. We set the caliper width to be 0.2 times the standard deviation of the logit PS. Using the standardized mean difference (SMD), the balance of baseline features between CLDT users and HCTZ users was examined; an SMD of <0.15 indicated an acceptable difference.

Cumulative incidences were computed using the Kaplan–Meier method with censoring estimates, and the time‐to‐event outcomes were analyzed over the entire follow‐up period. Patients were either censored at the time of death or at the ultimate follow‐up. The hazard ratios (HR) and 95% confidence intervals (CI) for each outcome were determined through a Cox proportional hazards model analysis. In the PS‐matched cohort, we conducted double adjustment by incorporating additional covariates, including age, diabetes, chronic kidney disease, baseline SBP, and the use of renin‐angiotensin system blockers. Subgroup analyses were stratified by age (≥65 vs. <65 years), sex, the presence of diabetes, chronic renal failure, a history of MI, and a history of stroke. The Schoenfeld residual test was employed to verify the proportional hazards assumption, and no significant violations were detected. The threshold for statistical significance was established at p < 0.05. Analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) and the R Statistical Software version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

3. Results

A total of 21,482 patients received either CLDT or HCTZ for at least 30 days. Of these, 14,257 met the inclusion criteria and were included in the final analysis (Figure 1). Exclusions were made for 130 patients who were under 20 years old at the index date, 699 patients who had a MACE or were hospitalized within 30 days from the index date, 6386 patients who had missing values in key variables, and 10 patients who were prescribed HCTZ after CLDT but before MACE. A 1:1 PS matching was performed, resulting in 1606 matched pairs of patients in each group. The patients were followed up by a median of 693 days (inter‐quartile range; 213–1293 days).

3.1. Demographic and Clinical Characteristics

Before propensity score matching, the CLDT group was significantly younger (61.9 years vs. 65.1 years, p < 0.001), had a higher proportion of males (59.0% vs. 50.4%, p < 0.001), and a higher mean body mass index (26.8 kg/m² vs. 26.1 kg/m², p < 0.001) compared to the HCTZ group (Table 1). Additionally, the CLDT group had higher baseline systolic and diastolic BP and higher rates of alcohol consumption and smoking but a lower prevalence of diabetes and dyslipidemia. After PS matching, no significant differences remained between the two groups across most baseline characteristics, with an SMD of <0.15.

TABLE 1.

Demographic and clinical characteristics before and after propensity score matching.

	Unmatched cohort			Propensity score‐matched cohort
	Chlorthalidone (n = 1920)	Hydrochlorothiazide (n = 12 337)	p value	Chlorthalidone (n = 1606)	Hydrochlorothiazide (n = 1606)	p value	SMD
Age (years)	61.9 ± 14.5	65.1 ± 13.0	<0.001	61.8 ± 14.6	61.8 ± 14.1	0.902	0.0043
Male	1132 (59.0%)	6217 (50.4%)	<0.001	953 (59.3%)	961 (59.8%)	0.801	0.0101
Body mass index, kg/m2	26.8 ± 4.0	26.1 ± 3.8	<0.001	26.8 ± 4.0	26.7 ± 3.8	0.949	NA
Baseline SBP (mm Hg)	147.1 ± 17.0	135.7 ± 18.6	<0.001	147.0 ± 16.9	147.0 ± 18.8	0.930	−0.0033
Baseline DBP (mm Hg)	85.6 ± 14.1	80.0 ± 18.0	<0.001	85.7 ± 14.2	85.6 ± 14.7	0.901	0.0045
Alcohol	347 (18.1%)	1825 (14.8%)	<0.001	300 (18.7%)	287 (17.9%)	0.584	0.0208
Smoking	282 (14.7%)	1465 (11.9%)	0.001	246 (15.3%)	246 (15.3%)	1.000	0.0000
Comorbidities
Diabetes	1126 (58.6%)	7908 (64.1%)	<0.001	956 (59.5%)	930 (57.9%)	0.370	0.0330
Dyslipidemia	1682 (87.6%)	10 471 (84.9%)	0.002	1435 (89.4%)	1442 (89.8%)	0.729	−0.0141
Chronic kidney disease	853 (44.4%)	5647 (45.8%)	0.282	726 (45.2%)	701 (43.6%)	0.394	0.0313
History of heart failure	190 (9.9%)	1025 (8.3%)	0.023	158 (9.8%)	156 (9.7%)	0.953	0.0042
History of myocardial infarction	69 (3.6%)	352 (2.9%)	0.087	59 (3.7%)	51 (3.2%)	0.497	0.0265
History of Stoke	412 (21.5%)	2557 (20.7%)	0.481	339 (21.1%)	339 (21.1%)	1.000	0.0000
Medications
Beta‐blocker	640 (33.3%)	3280 (26.6%)	<0.001	540 (33.6%)	547 (34.1%)	0.823	−0.0092
Calcium channel blockers, DHP	1808 (94.2%)	6256 (50.7%)	<0.001	1504 (93.6%)	1507 (93.8%)	0.884	−0.0077
Calcium channel blockers, non‐DHP	133 (6.9%)	907 (7.4%)	0.536	123 (7.7%)	128 (8.0%)	0.793	−0.0117
RAS blockers	1909 (99.4%)	10 810 (87.6%)	<0.001	1599 (99.6%)	1592 (99.1%)	0.189	0.0662
Number of anti‐hypertensive drugs	2.3 ± 0.6	1.7 ± 0.8	<0.001	2.3 ± 0.6	2.3 ± 0.6	0.817	NA
			<0.001			0.730
Thiazide alone	3 (0.2%)	836 (6.8%)		2 (0.1%)	1 (0.1%)
Thiazide plus one additional drug	76 (4.0%)	3990 (32.3%)		70 (4.4%)	81 (5.0%)
Thiazide plus two additional drugs	1149 (59.8%)	5344 (43.3%)		944 (58.8%)	915 (57.0%)
Thiazide plus three additional drugs	652 (34.0%)	2093 (17.0%)		552 (34.4%)	573 (35.7%)
Thiazide plus four additional drugs	40 (2.1%)	74 (0.6%)		38 (2.4%)	36 (2.2%)
Statin	1078 (56.1%)	7087 (57.4%)	0.296	912 (56.8%)	904 (56.3%)	0.803	0.0101
Antiplatelets	875 (45.6%)	5431 (44.0%)	0.212	726 (45.2%)	722 (45.0%)	0.915	0.0050
Anticoagulants	115 (6.0%)	838 (6.8%)	0.207	104 (6.5%)	100 (6.2%)	0.828	0.0101
Oral hypoglycemic agents	470 (24.5%)	3803 (30.8%)	<0.001	417 (26.0%)	390 (24.3%)	0.290	0.0383
Laboratory measurements
Total cholesterol	164.7 ± 41.0	162.5 ± 41.0	0.027	164.9 ± 40.8	164.4 ± 41.2	0.761	NA
LDL‐cholesterol	95.3 ± 35.2	95.2 ± 33.9	0.836	95.3 ± 35.1	95.7 ± 34.7	0.768	−0.0103
HDL‐cholesterol	49.5 ± 13.5	48.8 ± 13.4	0.041	49.5 ± 13.3	49.3 ± 13.4	0.590	0.0191
Triglyceride	159.3 ± 119.0	146.4 ± 112.7	<0.001	159.8 ± 122.8	157.3 ± 155.0	0.605	0.0208
Glucose	121.7 ± 41.1	125.4 ± 46.2	<0.001	121.9 ± 40.6	121.8 ± 42.3	0.931	NA
Hemoglobin A1c	6.4 ± 1.2	6.6 ± 1.3	<0.001	6.4 ± 1.2	6.4 ± 1.3	0.559	NA
Creatinine	0.91 ± 0.36	0.93 ± 0.53	0.013	0.91 ± 0.34	0.91 ± 0.41	0.607	−0.0203
Sodium	140.2 ± 2.7	139.8 ± 2.9	<0.001	140.2 ± 2.7	140.1 ± 2.7	0.566	0.0201
Potassium	4.20 ± 0.44	4.24 ± 0.45	0.003	4.21 ± 0.43	4.20 ± 0.44	0.598	0.0187

Notes: Values are presented as numbers (percentages) or means ± standard deviation.

Abbreviations: ARB, angiotensin receptor blockers; DBP, diastolic blood pressure; ECG, electrocardiogram; GFR, glomerular filtration rate; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; NA, not applicable due to exclusion from propensity‐score matching; PCI, percutaneous coronary intervention; SBP, systolic blood pressure; SMD, standardized mean difference.

3.2. Cardiovascular Outcomes

The cumulative incidence of MACE at 3 years in the PS‐matched cohort was 1.8% in the CLDT group and 2.2% in the HCTZ group, with no statistically significant difference observed between the groups (unadjusted HR 0.91, 95% CI 0.50–1.66, p = 0.77) (Table 2, Figure 2). Secondary outcomes, including cardiovascular death, MI, stroke, and hospitalization for heart failure, similarly showed no significant differences between the groups. However, the incidence of coronary revascularization was significantly lower in the CLDT group compared to the HCTZ group (HR 0.39, 95% CI 0.16–0.93, p = 0.03) (Table 2).

TABLE 2.

Incidence of the primary and secondary outcomes at 3 years in the propensity‐score matched cohort.

	Propensity score‐matched cohort (n = 3212)
	Chlorthalidone (n = 1606)	Hydrochlorothiazide (n = 1606)	HR [95% CI]	Adjusted HR* [95% CI]	Log‐rank p
The primary outcome
MACE	20 (1.8%)	23 (2.2%)	0.91 [0.50–1.66]	0.88 [0.48–1.61]	0.77
Secondary outcomes
Cardiovascular mortality	4 (0.4%)	4 (0.4%)	1.08 [0.27–4.31]	1.04 [0.26–4.21]	0.92
Myocardial infarction	5 (0.5%)	7 (0.7%)	0.76 [0.24–2.41]	0.76 [0.24–2.41]	0.65
Stroke	6 (0.5%)	6 (0.6%)	1.03 [0.33–3.19]	1.00 [0.32–3.10]	0.96
Hospitalization due to heart failure	6 (0.6%)	8 (0.8%)	0.80 [0.28–2.31]	0.81 [0.28–2.36]	0.68
Coronary revascularization	7 (0.8%)	19 (1.9%)	0.39 [0.16–0.93]	0.38 [0.16–0.91]	0.03

Notes: Values are presented as numbers (an estimate of the cumulative incidence of events over time). *Adjusted for age, diabetes, chronic kidney disease, baseline systolic blood pressure, and the use of renin‐angiotensin system blockers.

Abbreviations: CI, confidence interval; HR, hazard ratio; MACE, major adverse cardiovascular events.

FIGURE 2.

MACE occurrence by diuretic group. The Kaplan–Meier curves present the cumulative incidence of MACE in the CLTD and HCTZ groups. (A) Unmatched cohort and (B) PS‐matched cohort.

CLTD, chlorthalidone; HCTZ, hydrochlorothiazide; MACE, major adverse cardiovascular events; PS, propensity score.

3.3. Blood Pressure Controls

During 3 years of follow‐up, office SBP and DBP were controlled to a similar level in both diuretic groups. No significant differences were observed for either SBP or DBP at 1, 3, 6, 12, 18, 24, 30, or 36 months, except for SBP at 18 months (134.4±15 mmHg in CLTD vs. 131.8±15 mmHg in HCTZ), which was considered an incidental finding. (Figure 3)

FIGURE 3.

Blood pressure control by diuretic group. The systolic BP was relatively constant throughout the follow‐up period. BP; blood pressure.

3.4. Subgroup Analysis

The pre‐specified subgroup analyses showed consistent treatment effects across age, sex, diabetes, renal function, and cardiovascular disease history (Figure 4). Among patients aged ≥65 years, the HR for CLDT versus HCTZ was 0.72 (95% CI 0.35–1.49), while in patients <65 years, the HR was 1.65 (95% CI 0.54–5.03), with no significant interaction observed (p for interaction = 0.22). Similar findings were observed for male and female patients, with no significant interaction between the sexes.

FIGURE 4.

Subgroup analyses for effects by diuretic group. CI, confidence interval; DM, diabetes mellitus; CKD, chronic kidney disease; HF, heart failure; MI, myocardial infarction; SES, socio‐economic status.

3.5. Safety Profile and Negative Control Outcomes

No significant differences were found in the occurrence of hyponatremia and hypokalemia at 3 years between the two groups (Table 3). Negative control outcomes did not show significant differences between the two groups.

TABLE 3.

Safety profiles and negative control outcomes (ICD‐10 codes) at 3 years in the propensity‐score matched cohort.

	Chlorthalidone (n = 1606)	Hydrochlorothiazide (n = 1606)	p value
Hyponatremia (<135 mEq/L)	248 (15.4%)	278 (17.3%)	0.17
Hypokalemia (<3.5 mEq/L)	308 (19.2%)	268 (16.7%)	0.07
Negative control outcomes
Foot fracture (S92)	1 (0.1%)	1 (0.1%)	1.00
Leg fracture (S82)	6 (0.4%)	4 (0.2%)	0.75
Hip fracture (S72)	6 (0.4%)	6 (0.4%)	1.00
Forearm fracture (S52)	4 (0.2%)	2 (0.1%)	0.68
Hand fracture (S62)	1 (0.1%)	4 (0.2%)	0.37
Traumatic subdural hemorrhage (S06.5)	3 (0.2%)	1 (0.1%)	0.62
Peptic ulcer (K25‐27)	16 (1.0%)	19 (1.2%)	0.73
Pneumonia (J09‐J18)	15 (0.9%)	23 (1.4%)	0.25
Urinary tract infection/cystitis (N30, N39.0)	25 (1.6%)	27 (1.7%)	0.89
Conjunctivitis (H10)	7 (0.4%)	4 (0.2%)	0.55
Herpes zoster (B02)	5 (0.3%)	7 (0.4%)	0.77
Contact dermatitis (L25)	3 (0.2%)	3 (0.2%)	1.00
Urticaria (L50)	6 (0.4%)	5 (0.3%)	1.00
Psoriasis (L40)	3 (0.2%)	1 (0.1%)	0.62
Cataract (H25, H26, H28)	49 (3.1%)	56 (3.5%)	0.55
Glaucoma (H40)	34 (2.1%)	38 (2.4%)	0.72
Lung cancer (C34)	1 (0.1%)	5 (0.3%)	0.22
Gastric cancer (C16)	3 (0.2%)	4 (0.2%)	1.00
Liver cancer (C22)	8 (0.5%)	2 (0.1%)	0.11
Colorectal cancer (C18‐20)	4 (0.2%)	1 (0.1%)	0.37
Prostate cancer (C61)	8 (0.5%)	3 (0.2%)	0.23
Breast cancer (C50)	5 (0.3%)	1 (0.1%)	0.22
Rheumatoid arthritis (M05, M06)	4 (0.2%)	6 (0.4%)	0.75
Osteoporosis (M81.0, M81.8, M81.9)	33 (2.1%)	22 (1.4%)	0.17

Notes: Values are presented as numbers (percentages).

Abbreviations: ARB, angiotensin receptor blocker; ICD‐10, International Classification of Diseases, 10th revision.

4. Discussion

This multicenter cohort study encompassing 14,257 patients with hypertension provides a comparative analysis of the anti‐hypertensive effects and cardiovascular outcomes of CLDT and HCTZ. The study results indicate no significant differences in the incidence of MACE between the two treatment groups. Secondary outcomes, including cardiovascular death, MI, stroke, and hospitalization for heart failure, also demonstrated no significant differences. However, the incidence of coronary revascularization was notably lower in the CLDT group. The safety profiles of the two medications revealed no significant differences in the rates of hyponatremia and hypokalemia. These findings suggest that both CLDT and HCTZ are effective in managing hypertension and preventing cardiovascular events, with similar safety profiles.

Thiazide diuretics have been the cornerstone of hypertension treatment for over 50 years and are the earliest class of anti‐hypertension drugs to be documented in literature. HCTZ is the most commonly prescribed anti‐hypertensive drug across the globe, and the most frequently prescribed diuretic for hypertension management in many countries, including the United States and South Korea [2, 7]. Despite its relative lack of demand, CLTD has shown superiority over HCTZ in controlling BPs and also in long‐term clinical outcomes [8, 9, 10]. CLTD lost its popularity in the 1980s due to its tendency to cause hypokalemia, raising concerns about increases in mortality [11]. This led to a sharp decline in CLTD prescriptions, especially in the “more susceptible” patients, such as the elderly or those with renal insufficiencies. Currently, formal comparisons between the two drugs are rare, partially due to insufficient prioritization, which may stem from economic issues associated with conducting large‐scale clinical trials and the prevailing preference for HCTZ in clinical practice.

Recently, a pragmatic trial with random assignment of either CLTD or HCTZ in a Veterans hospital has shown that patients who received CLDT did not have a lower occurrence of major cardiovascular outcome events or non–cancer‐related deaths, compared with patients who received HCTZ [6]. The trial found that HCTZ and CLTD had comparable performance. This experiment exclusively enrolled elderly patients aged 65 or older, limiting the application of the results to younger individuals. Another large retrospective cohort study compared cardiovascular outcomes in patients who were prescribed with either CLTD or HCTZ and found that CLTD was not associated with fewer cardiovascular events, but this study was also conducted in older adults aged 66 years or older [4].

The study by Ishani et al. also does not provide information about whether patients with severely impaired renal function were included, despite 23% of patients having a glomerular filtration rate (GFR) of less than 60 mL/min/1.73 m² at baseline [6]. This information is crucial because thiazide diuretics have historically been believed to become less effective when GFR is significantly reduced, and also because lower GFR is related to increased incidences of electrolyte imbalances [12, 13]. A recent 12‐week research, which included a control group receiving a placebo, demonstrated that CLTD effectively reduced BP in patients with GFRs ranging from 15 to 30 mL/min [14].

From the results of our study, at baseline, patients who were prescribed CLTD tended to be younger, with a higher BMI, SBP, and DBP, and more likely to be female and have diabetes or dyslipidemia. Overall, these represented the group of patients who were more likely to benefit from a stronger BP reduction and less likely to suffer from electrolyte imbalances as a side effect. PS‐matching balanced these discrepancies for a better comparison between the two drugs. After PS matching, there was no significant difference in MACE or the secondary outcomes of cardiovascular mortality, MI, stroke, or hospitalization due to heart failure. In MRFIT (Multiple Risk Factor Intervention Trial), CLTD had been suggested to be so potent and more effective at reducing cardiovascular events that patients on the HCTZ arm were recommended to be switched to the CLTD group. In a retrospective analysis of MRFIT, however, the differences in cardiovascular death, stroke, or heart failure were negligible between the two groups [9].

Another point to consider is the under‐representation of females in the recent studies for comparison of clinical hard outcomes. The DCP (Diuretic Comparison Project) by Ishani et al. was conducted at a Veterans Hospital, allowing for less than 5% of the study population to be females [6]. The retrospective analysis of the MRFIT exclusively contained male patients, due to the study design of the original MRFIT project [9]. In our study, females constituted around half of the total population, and ∼40% of the matched population. In subgroup analysis, sex had no significant interaction with diuretic type in MACE occurrence (Table 4, Figure 4). Not many studies have compared the differential effects of sex on thiazide diuretics and cardiovascular outcomes. Animal studies have revealed that ovariectomy decreased thiazide receptors in female rats, and that the density of the thiazide receptor appears to be twice as high in female rats compared with male rats [15]. Despite the apparent effects of sex on the sites of action of thiazide diuretics, significant inferences in clinical settings have not been distinguished [16, 17]. Reports of adverse reactions and electrolyte imbalances tend to be more common in females, however, this should be taken into consideration when prescribing either diuretic in a female patient [18].

TABLE 4.

Subgroup analysis for 3‐year MACE.

	No. of patients	Chlorthalidone (n = 1606)	Hydrochlorothiazide (n = 1606)	Log‐rank p	Hazard ratio (95% CI)	p value	p for interaction
No. of MACE events (%)
Age							0.22
≥65 years	1457	12 (2.3%)	18 (3.8%)	0.37	0.72 [0.35–1.49]	0.38
<65 years	1755	8 (1.4%)	5 (0.8%)	0.39	1.65 [0.54–5.03]	0.39
Sex							0.98
Male	1914	13 (1.9%)	15 (2.4%)	0.80	0.91 [0.43–1.91]	0.80
Female	1298	7 (1.6%)	8 (1.9%)	0.87	0.92 [0.33–2.55]	0.87
Diabetes							0.28
Yes	1886	17 (2.5%)	16 (2.7%)	0.82	1.08 [0.55–2.14]	0.82
No	1326	3 (0.8%)	7 (1.5%)	0.26	0.47 [0.12–1.82]	0.27
Chronic kidney disease							0.82
Estimated GFR < 60 mL/min/1.73 m2	1427	11 (2.4%)	13 (2.9%)	0.69	0.85 [0.38–1.90]	0.69
Estimated GFR ≥ 60 mL/min/1.73 m2	1785	9 (1.3%)	10 (1.7%)	0.97	0.98 [0.40–2.40]	0.96
History of myocardial infarction							0.85
Yes	110	6 (13.0%)	5 (17.4%)	0.98	0.94 [0.29–3.08]	0.92
No	3102	14 (1.4%)	18 (1.8%)	0.59	0.82 [0.41–1.66]	0.59
History of Stoke							0.66
Yes	678	13 (5.5%)	13 (5.7%)	0.96	0.99 [0.46–2.13]	0.97
No	2534	7 (0.8%)	10 (1.3%)	0.58	0.75 [0.28–1.96]	0.55

Notes: Values are presented as numbers (an estimate of the cumulative incidence of events over time).

Abbreviations: CI, confidence interval; HR, hazard ratio.

Notably, the incidence of coronary revascularizations was less common in the CLTD group, with a 60%‐reduced risk of coronary revascularization in the CLTD group. In the randomized trial by Ishani et al., CLTD showed a differential advantage over HCTZ in the subgroup of patients with a history of MI or stroke. The authors initially explained that it could be an incidental finding, but then showed in a secondary analysis that there was a significant interaction [6, 19]. The retrospective analysis of MRFIT also showed that the risk of MI (by either clinical or ECG‐based definition), angina, and CABG were all lower in the CLTD group compared with the HCTZ group [9]. The reasons behind this observed tendency of CLTD benefits regarding the coronary arteries are not completely clear. Some explanation may be found in the recent analysis by Pareek et al., whereby CLTD and HCTZ both lowered office BP to a similar extent, but only CLTD lowered the nocturnal BP, and thus the 24‐h BP [20]. Previous studies have also suggested that the use of office measurements as sole markers of anti‐hypertensive effects of HCTZ may be inappropriate [21]. Nocturnal hypertension has been associated with myocardial ischemia in hypertensive patients with coronary artery disease [22]. As our study was a retrospective study of office BP measurements, it was impossible to perform a separate analysis of nocturnal BP and outcomes. Further research involving 24‐h BP as part of the analysis may be needed to correctly delineate the association between CLTD or HCTZ and adverse coronary outcomes.

Regarding electrolyte disturbances, no significant differences were found between the groups for hyponatremia or hypokalemia. Hypokalemia was numerically more common in the CLTD group (19.2% vs. 16.7%, p = 0.07), although it was statistically insignificant. Considering the observed tendency of CLTD to induce hypokalemia in prior studies, this finding is not surprising and implies that the potassium‐lowering effects of CLTD may not be as significant in contemporary medical practice as previously believed.

There are some limitations to this study. First, being a retrospective study, it may have been affected by selection biases. To address this issue, we applied propensity score matching to equalize discrepancies in baseline characteristics. Second, despite PS matching, the two groups may have unmeasured differences. We have included a wide variety of factors for matching, however, and the matched variables were overall well‐balanced. Third, owing to the predominance of HCTZ over CLTD in the diuretic market in Korea, numerous patients prescribed with CLTD are on a crossover from HCTZ. Consequently, the study design permitted the enrolment of patients with previous HCTZ prescriptions into the CLTD group, provided no additional HCTZ prescriptions were issued. This may make the study susceptible to selection bias, and the remnant effects of HCTZ on the CLTD population cannot be eliminated. However, as the median follow‐up days were over 1.5 years in both groups, we do not believe that the residual effects would have had a significant influence on the outcomes. Fourth, most of the study population was Korean, which means extrapolation of the results to patients of other ethnicities should be cautious. Fifth, the drug doses of the two diuretics were not controlled, as the study was a retrospective analysis. Prescriptions were at each attending physician's discretion. We could not determine what led to the clinical decision‐making regarding the drug choice or the appropriate doses. The levels of optimal BP control were checked for the possibility of confounding factor bias by BP levels on the hard outcomes, and BP control at each visit was not significantly different across the follow‐up period between the two diuretic groups.

5. Conclusion

After adjusting for clinical baseline characteristics, both CLTD and HCTZ appear to be viable options for the treatment of hypertension, with insignificant differences in cardiovascular outcomes. However, being a retrospective cohort study, the results should be interpreted carefully.

Author Contributions

Conceptualization: Hyung Joon Joo. Data curation: Ju Hyeon Kim and Seungmi Oh. Formal analysis: Ju Hyeon Kim, Seungmi Oh, and Subin Lim. Methodology: Hyung Joon Joo and Seungmi Oh. Software: Hyung Joon Joo. Validation: Hyung Joon Joo and Seungmi Oh. Investigation: Ju Hyeon Kim and Subin Lim. Writing–original draft: Subin Lim and Ju Hyeon Kim. Writing–review & editing: Hyung Joon Joo, Soon Jun Hong, Cheol Woong Yu, Yong Hyun Kim, and Eung Ju Kim.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data used to support the findings of this study are available from the corresponding author upon reasonable request.