Controversy in Hypertension: Are Diuretics Harmful? Edward M. Freis, MD Vasilios Papademetriou, MD Currently the trend is away from thiazide diuretics as a primary treatment for hypertension. Elimination of these agents from the therapeutic arma. mentarium-or a decrease to very low dosage levels-could make control of hypertension much more difficult. In at least 59% of mild hypertension cases, thiazide diuretics in adequate doses are effective in lowering blood pressure. In addition, they are the most efficacious agents now available for use in combination with other antihypertensive drugs. * Should the treatment of hypertension always be ini- tiated with a thiazide diuretic, a beta blocker, or some other antihypertensive drug? Today, there is a trend away from thiazide diuretics as primary or step-one treatment, particularly in Europe. In Scandinavia especially, some physicians use diuretics only as adjuncts to other drugs and in very small doses; for example, hydrochlorothia- zide is prescribed in doses of 6.25 or 12.5 mg per day.' Reluctance to employ standard doses of diuretics in the routine treatment of uncomplicated hypertension is based primarily on two presumed adverse effects. The first, and most important, is the concern that diuretic- induced hypokalemia may predispose to potentially fatal ventricular arrhythmias. The second is that the eleva- tion of serum cholesterol resulting from diuretic treat- ment might increase the long-term risk of developing coronary artery disease. Drs Freti and Papademetriou are with the Hypertension Research Division of the Veterans Administration Medical Center and the Cardiology Department of Georgetown University Medical Center, Washington, DC. Dr Freis Eliminating thiazide diuretics from the therapeutic armamentarium or reducing their dosage to very low levels could make effective control of hypertension con- siderably more difficult than it is at present. The mode of action of the diuretics is unique: they are the only therapeutic agents that reduce blood pressure by lower- ing extracellular and plasma volume. In adequate doses, they are effective in controlling blood pressure in at least 50% of patients with mild hypertension. Also, they are the most effective agents for use in combination with other antihypertensive drugs. Therefore, it is extremely important that the charges against these agents be ex- amined carefully. Misconceptions About Thiazides The case against diuretic-induced hypokalemia as a cause of fatal ventricular arrhythmias remains un- proven. It is, we believe, based on several commonly held misconceptions. I. The hypokalemia is associated with a physiological& significant depletion of intracellular potassium stores. A review of the various studies of changes in total body potassium following thiazide administration indicates a near consensus that the losses of total body potassium are only 5% to 7C70.~ This is far below a level that could be considered biologically important. In contrast, extra- cellular potassium is reduced by about 20% in thiazide- Continued Primary Cardiology January 1985 Are Diuretics Harmful? induced hypokalemia. The intracellular concentration of potassium is unrelated to the extracellular concentra- tion since the former is maintained against a concentra- tion gradient by metabolic pump activity. Therefore, in- tracellular concentration can be maintained despite extracellular hypokalemia. 2. Hypokalemia results in increased irritability of the heart andproduces earlier depolarization. The suscepti- bility of the heart to depolarization is said to depend, at least in part, on the ratio of the concentration of po- tassium inside the cells to that of potassium outside the cells.' The ratio of intracellular to extracellular potas- sium increases with thiazide-induced hypokalemia be- cause the fall in extracellular potassium is greater than the fall in intracellular potassium. Thus, the changes in the ratio should increase the depolarization threshold, which should make the heart less, rather than more, sus- ceptible to the development of arrhythmias. However, a relative increase in intracellular to extracellular potas- sium also is said to raise conduction velocity, which may increase reentry phenomena and thereby possibly influ- ence the induction of arrhythmias. Which of these opposing mechanisms dominates in the presence of thiazide-induced hypokalemia is not known. Indeed, the pathophysiologic consequences of thiazide-induced hypokalemia on the heart are unknown at present and certainly cannot be assumed to be harm- ful as some authors believe. The dangers of thiazide-induced hypokalemia, if any, probably are quite different in patients with overt heart disease compared to asymptomatic patients with mild or moderate hypertension. In congestive heart failure there may be considerable intracellular depletion due to various complicating factors.2 In myocardial infarc- tion (MI), the sodium-potassium pump functions inade- quately in the ischemic zone and intracellular potassium leaks from the cells, causing a local increase in extracel- lular potassium that results in increased arrhythmogenic activity.4 Such hearts are susceptible to ventricular ar- rhythmias, and any further reduction in intracellular po- tassium, however small, may tip the balance. FAST TAKE -.~ Resistance to thiazides: Physicians, particularly in Europe, are moving away from thiazides as primary treatment for hypertension. This trend is based on two presumed adverse effects of these agents: (1) potentially fatal ventricular arrhythmias may be more likely in diuretic-induced hypokalemia and (2) elevated cholesterol caused by diuretic therapy might increase the risk of heart disease. Primary Cardiology January 1985 The situation is quite different in patients with un- complicated hypertension, in whom intracellular potas- sium concentrations are essentially normal. However, it should be emphasized that the evidence for an associa- tion between hypokalemia secondary to diuretics and ventricular fibrillation in acute MI is based on retrospec- tive studies,5 and conclusive evidence of a correlation has not been documented. The only proven relationship between thiazide-induced hypokalemia and increased susceptibility to cardiac arrhythmias is the association with digitalis.6 3. Catecholamines aggravate hypokalemia, which may trigger arrhythmias. Further concern has been gen- erated by the recent finding that catecholamines, by causing a movement of extracellular potassium into cells, further aggravate the hypokalemia secondary to thiazides.' This movement is mostly into skeletal mus- cles with only minimal changes in the myocardial cell potassium concentration. Theoretically, this change in the equilibrium between intracellular and extracellular potassium, if it occurs in the myocardium, should make it less rather than more susceptible to arrhythmias. More importantly, it has not been demonstrated that the reduction in serum potassium from the combined effects of diuretics and catecholamines increases either the frequency or severity of arrhythmias. While high local catecholamine concentrations undoubtedly in- crease the frequency of serious cardiac arrhythmias, the latter are probably caused by other actions of these adre- nergic agents on the myocardium. 4. An increased incidence of cardiac arrhythmias has been objectively documented with diuretic-induced hy- pokalemia. Most of the evidence implicating diuretics has been generated by electrocardiographic (ECG) mon- itoring.8*9 This evidence has been challenged, however, by several recent studies. Papademetriou and co-workers in our laboratory failed to find any evidence of increased cardiac arrhythmias during hypokalemia secondary to thiazides.lO Using 24.~hour continuous ECG monitor- ing, they found that normalization of the hypokalemia with potassium supplements and/or potassium-sparing diuretics failed to change either the frequency or type of arrhythmias significantly. Papademetriou and associates then carried out 4%hour monitoring before and after thiazide-induced hypokalemia. Again, there was no significant difference in the incidence or type of arrhythmia before or after treatment with diuretic." All of the patients had asymptomatic hypertension and no overt heart disease. Using 4%hour monitoring, Lief and colleagues also found no increase in cardiac arrhythmias during thia- zide-induced hypokalemia.12 Continued Are Diuretics Harmful? FAST TAKE Hypokalemia: While some authors have reported that diuretic-induced hypokalemia causes increased cardiac arrhythmias, the findings have been challenged by several investigators. In one study, for example, when potassium or potassium-sparing diuretics normalized thiazide-induced hypokalemia, there were no significant changes in the frequency or type of arrhythmia. A report that hypokalemia secondary to diuretic treat- ment produces increased cardiac arrhythmias is fre- quently cited.9 This study used 24-hour ECG monitor- ing before and after inducing hypokalemia with thiazide diuretics. These results differ from those of Papademe- triou and Lief probably because of their method of case selection. Papademetriou and Lief did not select their patients on the basis of the amount of ectopic activity displayed during the pretreatment period. The other study, however, rejected any patients who had displayed more than five premature ventricular beats per hour. Since the frequency of ectopic beats varies widely from one 24-hour period to another, their study design greatly increased the likelihood of finding increased arrhythmic activity on the posttreatment re- cording. Thus, because monitoring lasted for only 24 hours and patients were selected with recordings taken at the low end of their range of spontaneous variability, their results may have been biased. The Medical Research Council of Great Britain stud- ied a subgroup of their patients. An increased incidence of ventricular ectopy was observed in the long-term thia- zide-treated group compared to the placebo-treated pa- tients using only one posttreatment and no pretreatment ECG monitoring.13 In a larger series, however, they monitored patients both before and after treatment with thiazides and found no difference in arrhythmia. Fur- thermore, there was no correlation between plasma po- tassium levels and ventricular arrhythmia frequency. 5. An increased incidence of sudden death associated with diuretic-induced hypokalemia has been demon- strated in large intervention trials. The evidence usually cited to implicate thiazide diuretics as a cause of fatal arrhythmias is provided by the Multiple Risk Factor In- tervention Trial (MRFIT).r4 A subgroup of patients with ECG abnormalities at baseline showed an increased incidence of sudden death. These patients received hy- drochlorothiazide, 50 mg twice daily. It is not known how the control group was treated, since they were re- ferred to other treatment sources. The MRFIT data may be challenged on several points. First, the evidence is the result of an analysis of many patient subsets chosen by a number of criteria after the study was completed. The correlation was not an objec- tive of the original trial. The more subsets that are ex- amined, the greater the possibility of a positive correla- tion by chance alone. Therefore, such a finding can only be regarded as a lead that requires further investigation to provide substantive corroboration. It is interesting that the sister study to the MRFIT, the Hypertension Detection and Follow-up Program (HDFP), failed to confirm the MRFIT observation. Attempts have been made to explain the difference on the basis of the dosages of diuretics used in the two studies, since HDFP used a dose of 50 mg chlorthali- done once daily. It is questionable, however, whether 50 mg chlorthalidone results in less hypokalemia than the 50 mg hydrochlorothiazide bid dose used in MRFIT. Chlorthalidone is a more potent diuretic than hydro- chlorothiazide., and its duration of action is considerably longer. In fact, in 50 to 100 mg doses, the former agent is known to produce hypokalemia more frequently than most of the other known diuretics. Also, there was no correlation between serum potassium levels and the inci- dence of arrhythmias in the MRFIT. Therefore, it is dif- ficult to ascribe the results in the MRFIT to thiazide- induced hypokalemia. Do Diuretics Increase Plasma Cholesterol? Plasma cholesterol usually rises slightly when thiazide treatment is initiated.15 The concern is that even though the elevation is modest, it might increase the risk of atherosclerosis over a long period. This fear probably is unfounded, however, because the rise in plasma chol- esterol apparently does not persist over the long term. After one or two years of continuous treatment, plasma cholesterol levels are no higher than they were prior to treatment. I6 Three studies demonstrated both the early increase and the later normalization of plasma cholesterol levels during thiazide treatment. The Veterans Administration trial of hydrochlorothiazide versus propranolol treated 343 patients with diuretic alone with no other drug or dietary interventions." Plasma cholesterol rose during the first three months of treatment but at 12 months it had fallen to slightly below baseline. Alcazar and asso- ciates18 also found that the increase in plasma choles- terol did not persist beyond the first few months of treat- ment with thiazides and afterward remained at pretreat- ment values during two years of observation. Williams and colleagues, using data from the HDFP, also found a short-term increase followed by a long-term return to baseline. I9 Conrinued Are Diuretics Harmful? Because the thiazide-induced modest elevation of cholesterol appears to be short-lived, it cannot be con- sidered an aggravating factor in the long-term develop- ment of atherosclerosis. Low-Dose Treatment with Diuretics Because of the concerns about hypokalemia and hyper- cholesterolemia, there is at present a tendency to pre- scribe very small doses of diuretics, such as 6.25, 12.5, or 25 mg of hydrochlorothiazide, once daily. Some re- cent European studies have indicated that such small doses are as effective as the standard therapeutic doses in reducing blood pressure, particularly when the di- uretic is given in combination with a step-two drug.QO This has led to the impression that the dose-response curve is flat between 6.25 and 25 mg per day of hydro- chlorothiazide. FAST TAKE Cholesterol levels: Concern that diuretics cause elevated plasma cholesterol and thus increase the risk of atherosclerosis is largely unfounded. While thiazides do cause an initial slight increase in plasma cholesterol, over the long term, plasma cholesterol levels are normalized. It is difficult to understand how such small doses could exert as great an antihypertensive effect as stan- dard doses. Thiazides lower blood pressure by reducing extracellular and plasma volume.21 Body weight re- flects the reduction in extracellular volume which is manifested in a loss of three to four pounds. In the Euro- pean studies cited above, the small doses had no effect on body weight, and only the highest dose of 50 mg per day of hydrochlorothiazide used by MacGregor and as- sociates20 caused any significant reduction in body weight. Thus, the lowering of blood pressure that oc- curred with the low doses of hydrochlorothiazide was not due to a decrease in volume and, therefore, may not have been caused by the diuretic. MacGregor and associates used a crossover design in which patients received a beta blocker plus a different dose of hydrochlorothiazide every four weeks-12.5,25, and 50 mg once daily. The blood pressure response was the same with all three doses of hydrochlorothiazide. However, it was not possible from this experimental de- sign to determine how much of the blood pressure reduc- tion was due to the beta blocker alone. Furthermore, in crossover trials there is always the possibility of carry- over antihypertensive effects that may persist for weeks or months after the treatment is withdrawn. Also, with Primary Cardiology January 1985 repeated clinic visits, blood pressure usually drifts down even with a placebo.22 Andren's trial, which used once-daily doses of 6.25, 12.5, and 25 mg of hydrochlorothiazide with enalapril did not employ a control of enalapril alone.' Thus, the blood pressure reductions observed could have been due to the converting enzyme blocker alone. If the toxicity of thiazide diuretics is not as great as some authorities believe, the important issue becomes whether the dose-response curve of hydrochlorothiazide is indeed flat between 6.25 and 100 mg per day. If it is not flat, many patients are receiving inadequate doses of the drug. This is unfortunate because the most impor- tant goal of treatment is effective blood pressure control. Recent evidence from the Veterans Administration study on propranolol versus hydrochlorothiazide23 in- dicates that the dose-response curve is not flat even at doses well above 25 mg per day. Hydrochlorothiazide alone was titrated upward from 25 mg twice daily to 50 mg, and then to 100 mg twice daily at monthly intervals until the blood pressure was controlled. Of the patients controlled below 90 mm Hg, 50% re- sponded to the smallest dose of 25 mg bid but 30% re- quired the intermediate dose and the remaining 20% needed 100 mg bid. While it might have been more prac- tical to have added a step-two drug rather than to pro- ceed to the highest dose, this study at least indicates that evidence is lacking for a flat dose response even at high doses and that there is great variation in the dose require- ment among different patients. Other investigators also have found that the dose-response curve is not flat with a daily dose of hydrochlorothiazide above 25 mg.24-26 Because of the present "obsession with potassium,"J physicians tend to administer doses of diuretics that are inadequate to reduce volume and, therefore, probably blood pressure as well. Clinicians may confuse spontan- eous falls in blood pressure that occur even with place- bo with the therapeutic action of the diuretic,22 and thus become convinced of the efficacy of small doses. They may ascribe the spontaneous fall to normotensive levels to the diuretic. Unfortunately, other patients who do not show such a spontaneous drop are considered treatment failures and never receive the benefit of more effective doses of the diuretic. Conclusion In the absence of congestive heart failure and digitalis, there is inadequate evidence to implicate thiazide-in- duced hypokalemia as a cause of cardiac arrhythmias. Also, the small rise in serum cholesterol that occurs with thiazide treatment is short-lived and reverts back to pre- treatment values during long-term treatment. Undue Are Diuretics Harmful? concern over these side effects has led to the use of small doses of diuretics, which may be subtherapeutic in many patients even when given as an adjunct to step-two drugs. If therapy is begun with a small dose of diuretic such as 25 mg of hydrochlorothiazide once daily, either alone or with a step-two drug, the dose should be dou- bled at least or given twice daily before concluding that the diuretic is ineffective. It is not possible to have a reduction in volume with- out some decrease in serum potassium. Indeed, the drop in serum potassium is due primarily to an equilibrium shift secondary to volume loss. Since the blood pressure decrease following diuretic treatment is also volume-de- pendent, a drug-related reduction in blood pressure is probably impossible without at least some reduction in serum potassium. The primary aim of treatment is to control the hyper- tension. Inadequate doses of diuretics or any other drug will not accomplish this aim. It is better to control the blood pressure than to be guided by unjustified fears of adverse drug effects. 0 References 1. Andren L, Weiner L, Svensson A, et al: Enalapril with either "very low" or "low" dose of hydrochlorothiazide is equally effective in essen- tial hypertension. A double-blind trial in 100 hypertensive patients. J Hypertension 1983;1(2):384-386. 2. Kassirer JP, Harrington JT Diuretics and potassium metabolism: A reassessment of the need, effectiveness and safety of potassium ther- apy. Kidney Int 1977;11:505-515. 3. Anderson KE: The heart cell: Electrophysiologic aspect. ActuMed Stand 1981;647:7-13. 4. Nudegger C, Anderson G, Hill J, et al: Extracellular K+ gradient across an ischemic boundary (abstract). JAm CON Curdiol1984;3(2): 587. 5. Harrington JT, Isner JM, Kassirer JP: Our national obsession with potassium. Am J Med 1982;73:155-159. 6. Lawn 9, Weller JM, Wyatt N, et al: Effects of alterations of body potassium on digitalis toxicity. J Clin Invest 1952;31;648-653. 7. Struthers AD, Whitesmith R, Reid JL: Prior thiazide diuretic treat- ment increases adrenaline-induced hypokalemia. Luncet 1983;1:1358- 1361. 8. Hollifield JW, Slaton PE: Thiazide diuretics, hypokalemia and car- diac arrhythmias. Actu h4ed &and 1981;647(suppl):67-73. 9. Holland 09, Nixon JV, Kuhnert I: Diuretic-induced ventricular ectopic activity. Am J Med 1981;770:762-768. 10. Papademetriou V, Fletcher RD, Khatri IM, et al: Diuretic-induced hypokalemia in uncomplicated systemic hypertension: Effect of plas- ma correction on cardiac arrhythmias. Am J Cardiol 1983; 52:1017-1022. 11. Papademetriou V, Price MB, Notorgiacomo A, et al: Effects of thiazide therapy on ventricular arrhythmias in patients with uncompli- cated systemic hypertension. Am Fed Cardiol Res 1984;32(2):337A. 12. Lief PD, Beligon I, Mates J, et al: Diuretic-induced hypokalemia does not cause ventricular ectopy in uncomplicated essential hyperten- sion (abstract). Kidney Int 1984;25:203. 13. Medical Research Council Working Party on Mild to Moderate Hypertension: Ventricular extrasystoles during thiazide treatment: sub- study of MRC mild hypertension trial. ErMed J 1983;287:1249-1253. 14. Kalata G: Heart study produces a surprise result. Science 1982;218: 31-32. Practice Procedures Efficacy Testing of Hydrochlorothiazide If hypertensive therapy is begun with a small dose of diuretic, such as 25 mg of hydrochlorothiazide once daily, either alone or with a step-two drug o The dose should be doubled o The dose should be given twice daily Before concluding that the diuretic is ineffective. 15. Grimm RH Jr, Leon AS, Hernninghake DB, et al: Effects of thia- zide diuretics on plasma lipids and lipoproteins in mildly hypertensive patients. Ann Int Med 1981;94:7-11. 16. Freis ED, Materson BJ: Short-term versus long-term changes in serum cholesterol with thiazide diuretics alone. Letter to the editor. L.uncet 1984;2:1414-1415. 17. Veterans Administration Cooperative Study Group on Antihyper- tensive Agents: Comparison of propranolol and hydrochlorothiazide for the initial treatment of hypertension. II. Results of long-term ther- apy. JAMA 1982;248:2004-2011. 18. Alcazar J, Ruilope L, Ladronde Guevara P, et al: Interrelationship between uric acid, cholesterol and triglycerides in essential hyperten- sion. Proc 9th Meet Int Sot Hypertension (abstract 8), Mexico City, Feb 20-21, 1982. 19. Williams WR, Borhani HD, Schnaper HW, et al: The relationship between diuretics and serum cholesterol in HDFP participants. JAm CON Cardiol 1983;1:623. 20. MacGregor GA, Banks RA, Markandu NM, et al: Lack of effect of beta blocker on flat dose response to thiazide in hypertension: Effi- cacy of low dose thiazide combined with beta blocker. Er Med J 1983; 1535-1538. 21. Shah S, Khatri IM. Freis ED: Mechanism of antihypertensive effect of thiazide diuretics. Am Heart J 1978;95:611-618. 22. Management Committee of the Australian Therapeutic Trial in Mild Hypertension: Untreated mild hypertension. Lnncel 1982;1:185- 191. 23. Veterans Cooperative Study Group on Antihypertensive Agents: Comparison of propranolol and hydrochlorothiazide for the initial treatment of hypertension. I. Results of short-term titration with em- phasis on racial differences in response. JAMA 1982;248:1996-2003. 24. Andersson PO, Anderson HH, Hagman A, et al: Potassium spar- ing by amiloride during thiazide therapy in hypertension. Clin Pharma- cot Ther 1984,36:197-200. 25. Henning R, Karlsberg BE, Odar-Dederlof I, et al: Timolol and hydrochlorothiazide-amiloride in primary hypertension. C/in Pharma- co1 Ther 1980,28:707-714. 26. Lederball-Pederson 0: Comparison of metoprolol and hydrochlor- othiazide as antihypertensive agents. Eur JClin Pharmacol1976;lO: 381-385. 471-5 Primary Cardiology January 1985