Diuretics Should Continue to Be One of the Preferred Initial Therapies in the Management of Hypertension

THE ARGUMENT FOR‐ Marvin Moser, MD, Clinical Professor of Medicine, Yale University School of Medicine, New Haven, CT

Thiazide diuretics were introduced in 1957 and became widely accepted as an effective well‐tolerated and inexpensive antihypertensive medication. ¹ Since that time, numerous studies, both placebo‐controlled and comparative trials with other medications, have established that thiazide diuretics reduce blood pressure (BP) and reduce the occurrence of strokes, progression of less‐severe to more‐severe hypertension, heart failure (HF), and coronary heart disease (CHD) events. ² , ³ , ⁴

Physiologic Effects of Diuretics

Although initially the use of thiazide diuretics results in some sodium depletion and a reduction in plasma volume and cardiac output, there is a gradual return of cardiac output to pretreatment levels and a return of plasma volume to just below baseline levels after several weeks of therapy with an ongoing reduction in BP and arterial resistance. ⁵ Thus a diuretic acts as a vasodilator over time to control BP. There are some counter‐regulatory mechanisms that come into play with chronic diuretic use. Because of a continuing, even minimal, reduction in plasma volume, there is an increase in the activity of the renin‐angiotensin system. Despite this reaction, the net result of treatment over time is a reduction in BP and vascular resistance. ⁶

Efficacy

Diuretics have proved to be as, or more, effective in lowering BP, especially systolic BP, than any of the available antihypertensive drugs. For example, in the blinded Antihypertensive and Lipid Lowering to Prevent Heart Attack Trial (ALLHAT), ⁴ BP lowering in patients randomized initially to a diuretic was equal to or slightly exceeded that noted with an angiotensin‐converting enzyme (ACE) inhibitor (lisinopril), a calcium channel blocker (CCB) (amlodipine), or the α blocker (doxazosin). Although other medications were necessary to reach goal levels in more than 40% of cases in each of the different treatment groups, these data are consistent with other studies. ⁷ There are some differences in outcome between different patient groups, i.e., elderly and black patients respond better to diuretics than Caucasian or younger people. These individuals respond better to agents that block the renin‐angiotensin system, i.e., β blockers, ACE inhibitors, or angiotensin receptor blockers. ⁸

Results of Therapy

In trials where diuretics were used as first‐step therapy and compared with control or placebo groups, there was a major reduction in the number of persons with less severe hypertension who progressed to more severe disease (only 95 of 13,389 progressed in the treated compared with 1493 of 13,342 in the control or placebo groups). Regression of left ventricular hypertrophy (LVH) in patients who had pretreatment LVH and prevention of LVH, as well as prevention of congestive heart failure and a decrease in morbidity and mortality not just for cerebrovascular (CV) disease but for CHD as well, has also been noted in diuretic‐based treatment programs compared with control or placebo regimens. ³

Tolerability

Diuretics in low or moderate doses are as well or better tolerated than any of the other available medications. Two large double‐blinded randomized controlled series reported that only 3% of diuretic‐treated subjects withdrew because of side effects. ⁹ , ¹⁰ In addition, in studies such as the Verapamil Atherosclerosis (VHAS) study, ¹¹ the chlorthalidone‐treated patients showed no difference in the percentage of adverse events compared with verapamil. Of interest in this 2‐year trial is that there were also no differences in fatal or nonfatal events and no difference in changes in total cholesterol or glucose levels. In ALLHAT, side effects on diuretics were not significantly greater than with an ACE inhibitor or CCB. In the Systolic Hypertension in the Elderly Program (SHEP) ¹² trial, adverse events on chlorthalidone were also not significantly more common than with placebo.

Outcome Compared With Other Medications

In comparative trials such as ALLHAT, the primary outcome of fatal CHD and nonfatal myocardial infarction (MI) was similar in patients treated with a diuretic, amlodipine, and lisinopril. BP lowering was somewhat greater in the chlorthalidone group compared with lisinopril, primarily because of a difference of 4 mm Hg systolic pressure in black patients (diuretic 4 mm Hg less than lisinopril). Stroke events in the ACE inhibitor group were more than in the diuretic patients, possibly due to the differences in achieved BPs. HF events were less in the chlorthalidone group compared with amlodipine and lisinopril patients.

In the ALLHAT trial, doxazosin‐treated patients showed 10% more CHD events, 18% more strokes, and more than a 100% increase in congestive HF compared with a diuretic‐based treatment group. ⁴ The Australian Study ¹³ did report a better outcome with an ACE inhibitor compared with a diuretic‐based regimen; there was a marginally significant p=0.05 better overall composite outcome in patients treated with an ACE inhibitor. This study, however, was not blinded and only about 40% of patients remained on the study drug. Of great interest is the fact that benefit was only noted in men.

Comparisons of diuretics and β blockers as initial therapy in randomized, controlled hypertension treatment trials, especially in older patients, indicate that risk reduction with a diuretic, especially relating to CHD events, is greater than with a β blocker despite the fact that these agents have proved beneficial in reducing mortality in post‐MI patients. In the SHEP study, outcome benefit was primarily the result of low‐dose chlorthalidone therapy; benefit was not significantly attributable to the addition of atenolol as a step‐2 drug (in 18% of subjects at year 5). ¹²

Both in placebo‐controlled studies and in most comparative trials with other medications, trial outcome data strongly suggest that diuretics are as, or more, effective in reducing CV events than other agents. They also suggest, however, that most hypertensive patients will require more than one drug and that a diuretic should be part of any antihypertensive regimen.

Reasonable Cost

Diuretics are the least expensive of all presently recommended antihypertensive medications. If data had shown that other agents were more effective, then cost would not be a consideration in the choice of medication. ¹⁴ The clinical trial data, however, suggest that quality of care is not compromised when diuretics are used as initial therapy. If additional data indicate a better outcome with more expensive medications, cost should not be a consideration.

Adverse Effects on Insulin Resistance and Glucose Metabolism: Do These Occur and Are They of Clinical Significance?

Thiazide diuretics may exert unfavorable influences on insulin sensitivity and glucose utilization, but evidence of adverse clinical outcome effects in large clinical trials is lacking. ¹⁵ This is an important issue given the cardiovascular risks inherent in insulin resistance and abnormal glucose tolerance, particularly in light of the rising prevalence of obesity and type 2 diabetes in the United States. In trials involving diuretics, fasting glucose changes are minimal, and when data are pooled, new‐onset hyperglycemia or diabetes is increased by only 0.6% above placebo. ¹² , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ Given that older populations, who have been studied in the clinical trials, in all probability had a degree of increased insulin resistance before entering a study, a higher rate of unmasked overt diabetes would have been anticipated even over the 3‐5‐year trial period if diuretics were indeed causative factors in diabetes.

Although hypertensive patients tend to develop diabetes more frequently than normotensive patients, some data suggest that patients on diuretics do not appear to require antidiabetic therapy any more often than those receiving other BP‐lowering agents. ²¹ For example, in SHEP, at 1 year there was a difference of 5 mg/dL in glucose levels between the diuretic and placebo, but no difference in new‐onset diabetes (Table I). ¹⁵ In comparative trials, however, such as ALLHAT, ⁴ Losartan Intervention for End Point Reduction in Hypertension study (LIFE), ²² Heart Outcomes Prevention Evaluation (HOPE), ²³ Study on Cognition and Prognosis in the Elderly (SCOPE), ²⁴ International Verapamil SR and Trandolapril Study (INVEST), ²⁵ International Nifedipine‐GITS Study: Intervention as a Goal in Hypertension Treatment (INSIGHT), ²⁶ and The Candesartan in Heart failure Assessment of Reduction in Mortality and morbidity (CHARM) ²⁷ a difference in new‐onset diabetes was noted in diuretic‐based treatment groups compared with other therapies (Table II). In ALLHAT, 3.5% fewer cases of diabetes were noted in the lisinopril than in the chlorthalidone group (11.6% compared with 8.1%), but fatal and nonfatal MI rates were similar both in diabetics and nondiabetics among diuretic‐treated compared with ACE inhibitor‐ or CCB‐treated patients.

Table I.

Effects of Diuretics Compared to Placebo on Glucose Levels and New‐Onset Diabetes

Study	Yr	Serum Glucose (mg/dL)	Hyperglycemia or Diabetes
Oslo	5	No difference D/Pl	No data
EWPHE	3	Increase of 6.6 D/Pl	Excess of 6 new cases*
MRC	3		Excess of 6 new cases*
HAPPY	4		Excess of 6 new cases
HDFP	5		1.6% (57/3563)
SHEP	1	Difference of 5 D/Pl	No difference—new onset diabetes Rx/C
MRFIT	6		Excess of 7% (SI) diuretics vs. excess of 2% (UC) no diuretics**
VA	2	1.7 D/Pl	No data
TOMHS	1	3.2 Rx/Pl	No data
EWPHE=European Working Party on Hypertension in the Elderly; MRC=Medical Research Council study; HAPPHY=Heart Attack Primary Prevention in Hypertension trial; HDFP=Hypertension Detection and Follow‐up Program; SHEP=Systolic Hypertension in the Elderly Program; MRFIT=Multiple Risk Factor Intervention Trial; VA=Veterans Affairs study; TOMHS=Treatment Of Mild Hypertension Study; D=diuretic; Pl=placebo; C=care; SI=special intervention; UC=usual care; *fasting glucose >126 mg/dL; **fasting glucose >110mg/dL. Reprinted with permission from Am Heart J. 1989;118:1362–1369. 15

Table II.

Effects of Different Medications on New‐Onset Diabetes

Study	No. of Subjects	Ages	Comment
LIFE	919	55–85	Losartan (ARB): Atenolol (β blocker) (HCTZ added)—25% lower incidence (13 vs. 17.5/1000 patient‐yr)
HOPE	5720	55+	ACE inhibitor (ramipril)—1.8% lower incidence: other medications
SCOPE	4964	70–89 (76*)	Candesartan (ARB)‐based therapy; 20% trend to reduction of new‐onset diabetes: other medications
ALLHAT	33,000	67*	3.5% Fewer cases of diabetes with lisinopril than with chlorthalidone
INVEST	>16,000		1.1% Fewer new‐onset diabetes in CCB/ACE compared with β blocker/diuretic
INSIGHT			1.3% higher with diuretics compared to CCB
LIFE=Losartan Intervention for End Point Reduction in Hypertension study; HOPE=Heart Outcomes Prevention Evaluation; SCOPE=Study on Cognition and Prognosis in the Elderly; ALLHAT=Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial; INVEST=International Verapamil SR and Trandolapril Study; INSIGHT=International Nifedipine‐GITS Study: Intervention as a Goal in Hypertension Treatment; ARB=angiotensin receptor blocker; HCTZ=hydrochlorothiazide; ACE=angiotensin‐converting enzyme; CCB=calcium channel blocker; *mean age

Thus a 1%‐4% difference in the absolute risk of occurrence of new‐onset diabetes is reported in diuretic‐based, compared with ACE inhibitor‐or angiotensin receptor blocker‐based comparative studies. CHD or CV disease outcome, however, as in the ALLHAT study, has apparently not been affected by this increase. In the INSIGHT trial, ²⁶ as another example, there was no difference in MI rates, strokes, HE, and CHD events between nifedipine gastrointestinal therapeutic system (GITS) and co‐amelioride, despite the fact that 1.3% more new‐onset diabetes was noted with the diuretic. In this study, as in all of the trials, approximately 40% or more patients required more than one medication.

Effects of Diuretic‐Based Therapy on Serum Cholesterol Levels

In none of the placebo controlled 2‐5 year trials have serum cholesterol levels been elevated by thiazide diuretics compared with placebo (Table III). ²⁸ In a meta‐analysis of trials no increase in cholesterol levels beyond 1 year has been reported. ²⁹

Table III.

Effects of Diuretics on Cholesterol Levels in Controlled Clinical Trials

Berglund & Anderson
6 Yr	−12
MRC Trial
3 + Yr
Men/Women—Rx/Pl	0/‐1/‐5–4
MAPHY
6 Yr	−1
HDFP
4 Yr (SI)
High doses diuretics	−9
Oslo
4 Yr
Active Rx/control	+1/+2
MRFIT
6 Yr
(SI) (UC)	−18/−14
MRC=Medical Research Council study; MAPHY=Metoprolol Atherosclerosis Prevention in Hypertension; HDFP=Hypertension Detection and Follow‐up Program; MRFIT=Multiple Risk Factor Intervention Trial; Pl=placebo; C=care; SI=special intervention; UC=usual care

Within the first year of treatment, thiazide diuretics may increase total and low‐density lipoprotein cholesterol by about 5%‐7% without alteration in HDL cholesterol concentration. Two trials have suggested that diuretics elevate lipids significantly, but these trials either included small numbers of patients, poor follow‐up, or were not carefully controlled. ³⁰ , ³¹ Large long‐term controlled clinical trials, however, report no change, or even a decrease in total cholesterol in the thiazide‐treated groups. ³² Studies that showed no significant effect on lipids include the Medical Research Council Study (MRC), ¹⁶ the Medical Research Council Study in the Elderly (MRC‐Elderly), ³³ the Metoprolol Atherosclerosis Prevention in Hypertension Study (MAPHY), ³⁴ and the Heart Attack Primary Prevention in Hypertension Study (HAPPHY). ¹⁷ In the SHEP in the Elderly study, ¹² a placebo‐controlled, diuretic trial, the reduction in strokes, CHD and CV disease events in the diuretic‐treated patients were similar whether patients had a baseline serum cholesterol below 216, 216‐251, or above 251 mg/dL.

In the Alpine Study, a comparative trial of an angiotensin receptor blocker, candesartan (plus felodipine, a CCB) compared with hydrochlorothiazide (HCTZ) (plus graduating doses of the β blocker, atenolol) there were adverse effects on serum glucose, insulin resistance, triglycerides, HDL and low‐density lipoprotein levels in the diuretic‐based treatment group. ³⁵ However, in this study, more than 50% of patients in the HCTZ group were also on high doses of atenolol (100 mg), a medication that will increase insulin resistance, lower HDL and increase triglyceride levels (the clinical relevance of these changes has not been clearly defined in view of the beneficial effects of these agents in MI patients and when combined with diuretics in the management of type 2 diabetics). The Alpine study cannot be used to define metabolic changes of a diuretic.

On the basis of the available observations, it can be concluded that the long‐term effects of thiazide diuretics on lipid levels are minimal and probably are of limited clinical significance. ²⁸ , ³⁶ Patients can be treated safely with thiazide diuretics whatever their lipid status, particularly in light of the availability of statin‐lipid‐modifying drugs. Although speculations and theories are of interest and the metabolic effects of diuretics, i.e., the increase in insulin resistance, short‐term effects on cholesterol, and a slight increase in new onset diabetes compared with other drugs may occur, these apparently do not appear to have a major clinical effect on outcome.

Effects of Diuretics on LVH

Some investigators continue to state that the use of thiazide diuretics as initial therapy does not cause regression of LVH, but careful evaluation of several controlled studies has shown that diuretic use is as, or more, effective in reducing the left ventricular mass than other medications. ³⁷ , ³⁸

Effects on Potassium and Cardiac Arrhythmias

Hypokalemia as a result of thiazide diuretics is dose related. Up to one‐third of patients will experience a decrease in serum potassium in the range of 0.5‐0.8 mEq/L on high doses of HCTZ (50‐100 mg/d). Less hypokalemia is observed at a dose of 25 mg HCTZ (serum potassium, ‐0.3 to 0.4 mEq/L), and even less is noted at 12.5 mg/d of HCTZ). Potassium‐sparing diuretics can be recommended in patients who may possibly experience a reaction to hypokalemia, especially elderly individuals, those receiving digitalis preparations, or diabetic patients whose insulin utilization may be influenced by hypokalemia. A logical approach to thiazide‐induced hypokalemia would indicate the use of combination agents such as spironolactone/ thiazide or ACE inhibitor or angiotensin receptor blocker/thiazide. In each case, the potassium‐elevating tendency of the former component will serve to moderate the effect on serum potassium induced by the latter.

But are the changes in potassium levels of clinical significance? Hypokalemia as a result of thiazide therapy may attenuate the benefit of reduction of CV events. In the SHEP trial for example, patients with potassium levels below 3.5 mmol/dL did not experience the same reduction in CV events as patients with more normal levels. It should, however, be noted that the overall benefit of diuretic therapy compared with placebo in SHEP was a dramatic reduction of CHD and CV disease events. Hypokalemia is relatively uncommon with a thiazide diuretic given at currently recommended doses, i.e., 12.5‐25 mg/d of HCTZ.

Some reports suggested that hypokalemia induced by thiazide diuretics could lead to increased ventricular ectopy, including ventricular tachycardia and sudden death. ³⁹ In addition, in what was likely a statistical aberration, subjects with abnormal electrocardiograms in the special care higher‐dose diuretic group in the Multiple Risk Factor Intervention Trial (MRFIT) ⁴⁰ were reported to experience a higher CHD mortality compared with usual care patients who received smaller doses of the diuretics. These results have been questioned based on a careful analysis of the MRFIT data. ⁴¹

Nonrandomized prospective case‐control studies have also raised the possibility of increased sudden death in diuretic‐treated patients. Treatment groups were not adequately matched in these studies. ⁴² , ⁴³ Carefully controlled 24‐ and 48‐hour Holter monitoring studies have helped to clarify this debate. Even after high doses (100 mg) of HCTZ, no significant increase in simple or complex ventricular ectopy was reported, despite some degree of hypokalemia (Table IV). ⁴⁴ Ectopy was not significant before or after exercise in patients with or without ventricular hypertrophy.

Table IV.

Ventricular Ectopy in Patients With and Without Left Ventricular Hypertrophy (LVH) Before and After Hydrochlorothiazide (50–100 mg/d for 4 weeks)

	LVH		No LVH
Variable	Baseline	Diuretic	Baseline	Diuretic
LVPWT	1.39	—	1.03	—
PK (mEq/L)	4.06	3.39	4.10	3.33
PVCs	16.6	10.1	2.1	3.0
Total couplets	123	15	6	3
Total VT episodes	5	3	2	0
LVPWT=left ventricular posterior wall thickness; PK=plasma k concentrations; PVCs=premature ventricular contractions; VT=ventricular tachycardia. Reproduced with permission from Arch Intern Med. 1988;148:1272–1276. 44

CONCLUSIONS

Diuretics as monotherapy reduce BP to as great or greater degree than other antihypertensive drugs. These medications used alone or in combination, in placebo‐controlled or comparative studies reduce morbid and mortal events to as great a degree as other agents. The metabolic effects of diuretics are probably of limited clinical significance. Factors of efficacy, tolerability, outcome data, and cost of care should be considered when choosing initial therapy in hypertension management. We agree, therefore, with the recommendations of the 7th Joint National Committee that diuretics should remain as a preferred initial therapy in most hypertensive patients. ⁴⁵ Of course, the argument regarding preferred initial therapy may be moot since more than one medication is often necessary to control BP. If a diuretic is not chosen as initial therapy, it should be the second drug of choice.