Diuretic‐Related Side Effects: Development and Treatment

Abstract

Diuretics are important therapeutic tools. First, they effectively reduce blood pressure and have been shown in numerous hypertension clinical trials to reduce both cardiovascular and cerebrovascular morbidity and mortality. In addition, their use has been equally effective in controlling cardiovascular events as angiotensin‐converting enzyme inhibitors or calcium channel blockers. Diuretics are currently recommended by the Seventh Report of the Joint National Commission on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure report as first‐line therapy for the treatment of hypertension. In addition, they remain an important aspect of congestive heart failure treatment in that they improve the congestive symptomatology, which typifies the more advanced stages of congestive heart failure. This article reviews the commonly encountered side effects with the various diuretic classes. Where indicated, the mechanistic basis and treatment of such side effects is further discussed.

The dose‐response relationship for the antihypertensive effect of diuretics has been more fully characterized over the past 20 years. In the process, many of the supposed negative attributes of diuretics are less common than was first thought. In the early days of diuretic use, doses were unnecessarily high, with dosing driven by the belief that “If a little is good, more is better”; however, it was quickly recognized that the blood pressure (BP) lowering effect for a thiazide‐type diuretic, such as hydrochlorothiazide (HCTZ), was relatively flat beyond a daily dose of 25 mg, and that at the higher dosages (100–200 mg/d), more negative metabolic experiences would occur. ² , ³ At lower doses (HCTZ, 12.5–25.0 mg), the metabolic mischief seen with high‐dose thiazide‐type diuretic therapy was much less concerning, with the possible exception of new‐onset diabetes. ³ Recent observations suggest that becoming a new‐onset diabetic as a consequence of diuretic therapy carries a similar negative cardiovascular (CVR) risk as exists for the diabetic population in general. ⁴ This observation varies from that of the Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial, ⁵ where the CVR outcomes in either new‐onset or already present diabetics were not worsened by diuretic therapy.

ADVERSE EFFECTS OF DIURETICS

Diuretic‐related side effects can be separated into several categories, including those with well—worked‐out mechanisms such as electrolyte defects and/or metabolic abnormalities and occurrences, such as impotence, which are mechanistically less well understood. In addition, various drug‐drug interactions are recognized to occur with diuretics. Diuretic‐related side effects are more common and of a greater intensity with loop diuretics. Thiazide‐related side effects are somewhat more common with longer‐acting compounds, such as chlorthalidone and metolazone. Among the thiazide‐type diuretics, indapamide has been touted by some as distinctive in not causing significant metabolic derangements; however, when it is given in equivalent doses to HCTZ, there is little that separates these two drugs relative to side effects.

HYPONATREMIA

Hyponatremia is an uncommon, but serious, complication of diuretic therapy. ⁶ , ⁷ Thiazide diuretics are more likely than loop diuretics to cause hyponatremia. Loop diuretics inhibit sodium (Na⁺) transport in the renal medulla and prevent the generation of a maximal osmotic gradient. Thus, urinary concentrating ability is impaired with loop diuretics. Alternatively, thiazide‐type diuretics increase Na⁺ excretion and preclude maximal urine dilution, while preserving the kidney's innate concentrating capacity. When diuretic‐related hyponatremia occurs, it is typically in elderly females and is usual seen shortly after therapy begins (within the first 2 weeks). ⁸ However, diuretic‐related hyponatremia can occur on a delayed basis even after several years of therapy. ⁷ Multiple factors contribute to the penchant of females to diuretic‐related hyponatremia, including age, reduced body mass, exaggerated natriuretic response to a thiazide diuretic, diminished capacity to excrete free water, and self‐imposed low‐solute intake. Independent of this constellation of risk factors, it has been suggested that the apparent female preponderance of thiazide‐induced hyponatremic adverse events is related to overrepresentation of females in thiazide‐treated cohorts, rather than intrinsic susceptibility to the electrolyte disturbance. ⁷

Mild asymptomatic diuretic‐related hyponatremia (typically between 125–135 mmol/L) can be managed in a number of ways (which are not necessarily mutually exclusive), including: restricting free‐water intake, replacing potassium (K⁺) losses, withholding diuretics, or switching to loop diuretic therapy if diuretic therapy remains necessary. ⁹ , ¹⁰ Severe, symptomatic hyponatremia (generally <125 mmol/L), complicated by seizures or other active neurologic sequelae, represents a true medical emergency. A fall in serum Na⁺ to this degree calls for intensive therapy; however, this level of symptomatic hyponatremia should not be corrected too rapidly because the osmotic demyelinating syndrome has occurred under these circumstances. The risks of ongoing hyponatremia must be weighed against those of too hasty a correction, and current recommendations are that plasma Na⁺ should be corrected by no more than 0.5 mmol/h during the first 24 hours of treatment. ¹¹ , ¹² The pace at which hyponatremia is corrected should be slowed once a mildly hyponatremic serum Na⁺ range has been reached (approximately 125–130 mmol/L). The acuity (≤48 hours) of the hyponatremia also influences the speed with which hyponatremia is corrected. Controversy still surrounds a number of aspects of the therapy of hyponatremia.

HYPOKALEMIA AND HYPERKALEMIA

A serum K⁺ value of ≤3.5 mmol/L, which is the most common definitional criterion for a diagnosis of hypokalemia, is a common finding in patients treated with loop and/or high‐dose thiazide diuretics. ¹² During the first several days of thiazide‐diuretic therapy, plasma K⁺ falls an average of 0.6 mmol/L (in a dose‐dependent manner) in subjects not taking K⁺ supplements, as compared with a 0.3 mmol/L drop in those taking furosemide. ¹³ However, it is unusual for serum K⁺ values to settle <3.0 mmol/L in diuretic‐treated outpatients, apart from a high dietary Na⁺ intake and/or when a long‐acting diuretic is being given (as is the case with chlorthalidone). Mechanisms that contribute to the onset of hypokalemia during diuretic use include: increased flow‐dependent distal nephron K⁺ secretion (more commonly observed with a high Na⁺ intake), a fall in distal tubule luminal chloride (Cl⁻) metabolic alkalosis, and/or secondary hyper‐aldosteronism. ¹⁴ , ¹⁵

The cardiac implications of diuretic‐induced hypokalemia remain controversial. It would seem logical to infer that arrhythmia‐related event rates are connected to the degree of hypokalemia, but this is in no way an unambiguous relationship (at least in an outpatient setting). This theme is confused by several factors including: the inconstant relationship between serum K⁺ concentrations and total body K⁺ deficits in the face of diuretic therapy; the fact that in most clinical trials evaluating arrhythmia risk (and/or sudden cardiac death [SCD]), serum K⁺ values have not been measured frequently enough or under sufficiently standardized conditions to allow for anything more than an educated guess as to the “average” K⁺ value at the time of an event; that the range of serum K⁺ values most commonly associated with increased ventricular ectopy is very small typically between 3.0–3.5 mmol/L and finally the issue of whether hypokalemia produced by transcellular shifts of K⁺ manufactures a similar risk as that generated by a reduced serum K⁺ on the basis of total body losses.

It has been observed that even mild degrees of diuretic‐induced hypokalemia can be coupled with ventricular ectopy. ¹⁶ , ¹⁷ For example, the Multiple Risk Factor Intervention Trial ¹⁶ observed a significant inverse relationship between the serum K⁺ concentration and the frequency of premature ventricular contractions (PVCs); however, in this trial patients on chlorthalidone with the greatest decrease in serum K⁺ levels had the best outcomes. However, this relationship has not been detected in all studies, possibly because of the brief duration of many of these trials. ¹⁸ , ¹⁹ For example, in the Medical Research Council (MRC) study, 287/324 patients with mild hypertension underwent ambulatory electrocardiographic (ECG) monitoring. In the short‐term (8 weeks), there was no increase in the frequency of PVCs; however, after 24 months of therapy, a significant difference, which correlated with serum K⁺ concentrations, emerged in the PVC rate (20% [diuretic treated] vs. 9% [placebo]). ¹⁹

The hazards central to diuretic‐related hypokalemia are most apparent in patients with left ventricular hypertrophy, congestive heart failure (CHF), and/or myocardial ischemia; particularly when they become acutely ill and have need of hospitalization. ²⁰ , ²¹ , ²² , ²³ As mentioned previously, outpatient forms of diuretic‐related hypokalemia are seldom of a severe enough nature to demand urgent attention; however, these mildly lowered serum K⁺ values create a basis for more significant degrees of hypokalemia when transcellular shifts of K⁺ are interposed, as occurs during stressful circumstances marked by high endogenous epinephrine levels ²⁴ ; therein lies one of the major at‐risk scenarios of diuretic‐related hypokalemia.

Despite a sometimes monotonous level of concern about CVR risk (rather than benefit) with diuretic therapy, in part, due to associated electrolyte abnormalities, several clinical trials, including the Systolic Hypertension in the Elderly Program, Swedish Trial in Old Patients with Hypertension, and MRC have shown that low‐dose diuretic therapy reduces CVR event rates by 20%–25%. ²⁵ , ²⁶ , ²⁷ , ²⁸ Perhaps the use of lower doses of thiazides or their combination with a K⁺‐sparing diuretic explains these favorable results as compared with earlier trials, such as the Multiple Risk Factor Intervention Trial, in which higher doses of diuretics were employed (and PVCs were more frequent). In the Systolic Hypertension in the Elderly Program (SHEP), ²⁸ 7.2% of those actively treated developed hypokalemia (serum K⁺ <3.5 mmol/L at year 1).

Those subjects who developed hypokalemia did not secure the treatment benefits on CVR and coronary events as well as stroke recognized in similarly‐treated, but normokalemic patients, with isolated systolic hypertension.

Two additional issues are considerations in the milieu of diuretic‐related hypokalemia: first, the hemodynamic benefit of normalizing serum K^{+ 29} and second, the consequences of different doses, combinations of diuretics and/or K⁺‐sparing diuretics on SCD. ³⁰ , ³¹ To the former, K⁺ supplementation (average increase in serum K⁺ of 0.56 mmol/L) in hypokalemic (serum K⁺ values <3.5 mmol/L), diuretic‐treated patients have been followed by a 5.5 mm Hg average fall in mean arterial pressure. ²⁹ As to the latter, the risk of SCD among patients receiving combined thiazide and K⁺‐sparing diuretic therapy has been shown to be lower than that found in patients treated with thiazides alone, with odds ratios for an event increasing significantly as the monotherapy dose of HCTZ increased from 25–100 mg/d. ³⁰ Of note in these studies, the addition of K⁺ supplements to thiazide therapy had little effect on the risk of SCD, suggesting that other properties of K⁺‐sparing diuretics (such as decreasing urinary losses of magnesium [Mg²⁺]) may have been at play. ³⁰

K⁺‐sparing diuretics (such as triamterene and amiloride) and aldosterone‐receptor antagonists (such as spironolactone and eplerenone) are often used for their ability to conserve K⁺ when it might otherwise be lost with thiazide and loop diuretic therapy. In certain instances, significant enough K⁺ retention occurs so as to result in hyperkalemia. Hyperkalemia with K⁺‐sparing diuretics is usually encountered in patients with an existing reduction in their glomerular filtration rate (when also given K⁺ supplements or salt substitutes), individuals who develop acute‐on‐chronic renal failure, those on an angiotensin‐converting enzyme (ACE) inhibitor/angiotensin receptor blocker (ARB) and/or a nonsteroidal anti‐inflammatory drug, or in other situations that predispose to hyperkalemia, such as metabolic acidosis, hyporeninemic hypoaldosteronism, or heparin therapy (including subcutaneous heparin regimens). ³²

HYPOMAGNESEMIA

Both thiazide and loop diuretics increase urinary Mg²⁺ excretion. All K⁺‐sparing diuretics diminish the magnesuria that accompanies thiazide or loop diuretic use. ³³ Prolonged therapy with thiazide and loop diuretics, on average, reduces plasma Mg²⁺ concentration by 5%–10%, although patients can develop more severe hypomagnesemia in association with similarly‐sized total body deficits. ³⁴ Cellular Mg²⁺ depletion occurs in up to 50% of patients receiving thiazide diuretics and can be present despite normal serum Mg²⁺ concentrations. Hypomagnesemia occurs more frequently in the elderly, and in those receiving high‐dose loop diuretic therapy for extended periods of time (such as heart failure patients). ³⁵ Hypomagnesemia often coexists with hyponatremia and hypokalemia, with one study finding 41% of patients with hypokalemia to also have low serum Mg²⁺ concentrations. ³⁶ Hypocalcemia and/or hypokalemia found in association with low serum Mg²⁺ concentrations can prove refractory to all treatment measures until the underlying Mg²⁺ deficit is corrected. ³⁷

The measurement of serum Mg²⁺ concentration continues as the everyday test for uncovering hypomagnesemia ³⁸ , ³⁹ ; however, the presence of hypomagnesemia can be suspected from characteristic ECG, neurologic and/or neuromuscular findings. On ECG, hypomagnesemia can present as prolongation of the Q‐T and P‐R intervals, widening of the QRS complex, ST segment depression, and low T waves, as well as supraventricular and ventricular tachyarrhythmias. ⁴⁰ The neurological changes with hypomagnesemia are nonspecific and include mental status changes and/or neuromuscular irritability. Tetany, one of the most striking and better known manifestations of Mg²⁺ deficiency, is only rarely seen; instead, less specific neurologic signs such as tremor, muscle twitching, bizarre movements, focal and/or generalized seizures, and delirium/coma are more common findings. ⁴¹

While a low serum Mg²⁺ level is helpful and is typically indicative of low intracellular stores, normal serum Mg²⁺ values can still be observed in the face of a significant body deficiency of Mg²⁺; thus, serum Mg²⁺ determinations are an unreliable measure of total body Mg²⁺ balance. ⁴² Intracellular Mg²⁺ measurements, as well as other technologies, are available to assess Mg²⁺ balance, but are not readily available. A more practical measure of Mg²⁺ balance is the "magnesium loading test," which at the same time is both therapeutic and diagnostic. This test consists of the parenteral administration of magnesium sulfate and a time‐wise assessment of urinary Mg²⁺ retention. This can be accomplished on an outpatient basis with the Mg²⁺ load being given in as short a time interval as 1 hour with subsequent urine collection over 24 hours. Individuals in a state of normal Mg²⁺ balance eliminate at least 75% of an administered load. ⁴³

Several issues arise concerning the treatment of diuretic‐related hypomagnesemia beyond empirically normalizing a laboratory value. These include possible favorable effects on BP control, arrhythmia development, and/or coexisting electrolyte or neuromuscular symptoms. In the instance of BP control, there appears to be little additional reduction in BP when Mg²⁺ deficiency is corrected; this circumstance differs from the BP reduction occasionally seen when Mg²⁺ supplementation takes place in a nondeficient state. In the presence of refractory tachyarrhythmias or torsade de pointe, hypomagnesemia should be rapidly treated. Finally, when quantifiable measures, such as the electrolyte abnormalities of hypokalemia and/or hypocalcemia are present, the value of treating diuretic‐related hypomagnesemia is readily apparent.

It is worthwhile for Mg²⁺ deficiency to be identified in all patients in whom a high index of suspicion for hypomagnesemia exists, but particularly in those with ischemic heart disease or known cardiac arrhythmias. In mild deficiency states, Mg²⁺ balance can often be reestablished by attention to the original causes (e.g., limiting diuretic and Na⁺ intake) and allowing dietary Mg²⁺ to correct the deficit. Parenteral Mg²⁺ administration, however, is the most effective way to correct a hypomagnesemic state, and should be the route used when replacement is of a more emergent nature. In the depleted patient, total body deficits of Mg²⁺ are typically in the order of 1–2 mEq/kg/BW. One commonly employed regimen, albeit empiric, gives 2 g of magnesium sulfate (16.3 mEq) IV over 30 minutes, followed by a constant infusion providing between 32–64 mEq/d until the deficit is presumed corrected.

A variety of oral Mg²⁺ salts are available for the treatment of hypomagnesemia. Mg²⁺ oxide is one commonly employed, but this salt is poorly soluble and acts as a cathartic, which can limit its effect. Mg²⁺ gluconate is the preferred salt for oral therapy, as this agent is very soluble and is minimally cathartic. Mg²⁺ carbonate is poorly soluble and does not appear to be as effective in reversing hypomagnesemia as is the gluconate salt. Oral Mg²⁺ is not recommended for therapy during acute situations, since the high doses needed almost always cause significant diarrhea. The IM route for Mg²⁺ administration is another route of delivery, albeit a potentially painful one, and should be avoided as long as IV access is readily available. ³⁷

ACID‐BASE CHANGES

Mild metabolic alkalosis is a common feature of thiazide diuretic therapy, particularly at higher doses. Severe metabolic alkalosis is much less frequent and, when it occurs, it is in association with loop diuretic use. The generation of a metabolic alkalosis with diuretic therapy is primarily due to contraction of the extracellular fluid space caused by urinary losses of a relatively HCO₃‐free fluid. ⁴⁴ Diuretic‐induced metabolic alkalosis is best managed by administration of K⁺ and/or Na⁺ chloride, although Na⁺ chloride administration may be impractical in already volume‐expanded patients (such as those with CHF). In such cases, a K⁺‐sparing diuretic or a carbonic anhydrase inhibitor, such as acetazolamide, may be considered. Metabolic alkalosis also impairs the natriuretic response to loop diuretics and may play a role in the diuretic resistance occasionally found in the CHF patient. ⁴⁵ All K⁺‐sparing diuretics can cause hyperkalemic metabolic acidosis, which in elderly patients, or in those with renal impairment or CHF, can reach a life‐threatening level. ⁴⁶

HYPERURICEMIA

Thiazide diuretic therapy increases serum urate concentrations by as much as 35%; an effect related to decreased renal clearance of urate, and one that is most prominent in those with the highest pretherapy urate clearance values. ⁴⁷ Decreased urate clearance may be linked to increased reabsorption secondary to diuretic‐related extracellular fluid volume depletion and/or competition for tubular secretion, since both thiazide diuretics and urate undergo tubular secretion by the same organic anion transporter pathway. ⁴⁸ , ⁴⁹ Diuretic‐related hyperuricemia is dose‐dependent and is pertinent for two reasons: first, as a precipitant of gout and second, relative to its effect on CVR event rate.

First, diuretic‐related hyperuricemia does not typically precipitate a gouty attack unless the patient has an underlying gouty tendency or serum urate concentrations routinely exceed 12 mg/dL. ⁴⁸ To this end, in the MRC trial, patients receiving high‐dose thiazide diuretics had significantly more withdrawals for gout than did placebo‐treated patients (4.4 vs. 0.1/1000 patient years). ⁵⁰ Second, in the SHEP, ⁵¹ those with a serum uric acid increase ≥0.06 mmol/L (median change) in the active treatment group had a similar risk of coronary events as the placebo group, suggesting that diuretic‐related hyperuricemia offsets the positive CVR benefits otherwise seen with diuretic therapy. This difference was not explained by BP effects.

Allopurinol should not be routinely started (as often is the case) for asymptomatic diuretic‐related hyperuricemia. If a gouty attack occurs in a diuretic‐treated patient, the diuretic in use should be temporarily discontinued. Oftentimes, a diuretic can be restarted at a lower and sometimes still effective dose. In the process, careful attention should be paid to avoidance of excessive volume contraction. In the patient with preexisting gout and with a need for diuretic therapy, the xanthine oxidase inhibitor, allopurinol, can be considered. However, allopurinol (a renally‐cleared compound) should be used cautiously (dose‐adjusted according to level of renal function) in patients receiving a thiazide‐type diuretic, since allopurinol hypersensitivity reactions are more common with this combination. ⁵² A final consideration is what steps to take in a patient with diuretic‐related hyperuricemia who is intolerant of allopurinol. In such subjects, the ARB losartan, which is a uricosuric compound, can be safely given with a reduction in serum uric acid and no risk of acute urate nephropathy and/or uric acid stones. ⁵³

Hyperglycemia

Prolonged thiazide diuretic therapy can lead to glucose intolerance and may occasionally precipitate diabetes mellitus. ⁴ , ⁵ , ⁵⁴ , ⁵⁵ Short‐term metabolic studies, epidemiologic studies, and a variety of clinical trials suggest a connection between ongoing thiazide diuretic use and the development of type 2 diabetes. However, it should be noted that interpretation of these studies is confounded by multiple factors including: differing definitions of new‐onset diabetes, small numbers of patients, inadequate comparison groups, relatively limited periods of follow‐up, selection criteria that limited the generalizability of the findings, and study designs that prohibited valid comparisons among antihypertensive drug classes. ⁵⁶ Moreover, in a review of all the placebo‐controlled hypertension trials with diuretics, there was only an approximate 1% increase in new‐onset diabetes compared with placebo. ⁵⁷

Hyperglycemia and carbohydrate intolerance have been linked to diuretic‐induced hypokalemia. K⁺ deficiency is known to inhibit insulin secretion by β cells; however, diuretic‐induced changes in glucose metabolism are not conclusively related to altered K⁺ homeostasis, and impaired glucose tolerance occurs even when thiazide‐type diuretics in relatively low doses are combined with K⁺‐sparing agents. The glucose intolerance seen with diuretic therapy can deteriorate further with an increase in sympathetic nervous system activity, which also decreases peripheral glucose utilization. Diuretic‐associated glucose intolerance appears to be dose‐related, less common with loop diuretics, not present with spironolactone, and reversible on withdrawal of the agent, although the data on reversibility in HCTZ‐treated patients is somewhat conflicting. ⁵⁵ Of note, an overview of this issue found that glucose homeostasis was unpredictably affected by low‐dose HCTZ (12.5–50 mg/d). ⁵⁸

Recently, a large, prospective, cohort study (12,550 nondiabetic adults [45‐ to 64‐ years old] who did not have diabetes) concluded (after appropriate adjustment for confounders) that hypertensive patients taking thiazide diuretics were not at greater risk for subsequent diabetes development than patients who were not receiving antihypertensive therapy. The diuretic doses were not reported in this cohort study; thus, because of the perceived variability of this effect, blood glucose should be monitored during thiazide therapy, particularly in those with either the metabolic syndrome or existing diabetes. ⁵⁹ This is particularly so since the CVR risk with new‐onset diuretic‐related diabetes parallels that which accompanies existing diabetes. ⁴ Other drug classes such as ACE inhibitors and ARBs are associated with a lesser incidence of new‐onset diabetes. It remains to be determined the extent to which either of these drug classes reduces the diabetogenic potential of thiazide‐type diuretics.

HYPERLIPIDEMIA

Short‐term thiazide diuretic therapy can dose‐dependently elevate serum total cholesterol levels, modestly increase low‐density lipoprotein cholesterol levels and raise triglyceride levels, while minimally changing high‐density lipoprotein cholesterol concentrations. ⁶⁰ , ⁶¹ , ⁶² , ⁶³ These lipid effects have been reported to be more apparent in blacks, males, diabetics, and nonresponders to diuretic therapy. ⁶² , ⁶³ In nonresponders to diuretic therapy, the observed increase in lipid values likely relates to the higher diuretic doses used (or required) in such patients. ⁶⁴ All diuretics, including loop diuretics, cause these lipid changes, with the possible exception of indapamide. ⁶¹ The mechanisms of diuretic‐induced dyslipidemia remain uncertain, but have been related to worsened insulin sensitivity and/or reflex activation of the renin‐angiotensin‐aldosterone system (RAAS) and sympathetic nervous system in response to volume depletion. Supporting this latter notion is the fact that doses of diuretics, which are low enough so as not to activate the sympathetic nervous system, do not increase lipid values; in contrast, higher diuretic doses are more apt to be associated with reflex sympathetic nervous system activation.

Long‐term clinical trials differ from short‐term studies in that cholesterol levels are little changed from baseline after 1 year of diuretic therapy. ⁶¹ , ⁶² Moreover, data from the Hypertension Detection and Follow‐up Program indicate that diuretic‐treated hypertensive subjects with baseline cholesterol values of >250 mg/dL experience a decline in cholesterol levels from the second to the fifth year of treatment. ⁶⁵ Finally, in the diuretic‐based SHEP, CVR outcome was similar in patients with cholesterol levels <200 mg/dL and >280 mg/dL. Thus, whatever lipid changes that do occur with diuretics are not only short‐term, but also are probably of limited clinical importance.

Impotence

Adverse effects of thiazide and thiazide‐like diuretics on male sexual function, including decreased libido, erectile dysfunction, and difficult ejaculation have been reported in several studies with an incidence that varies from 3%–32%. ⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ As an example, in the MRC trial, in which 15,000 hypertensive subjects received either placebo, thiazide (bendrofluazide), or a β blocker (propranolol) for 5 years, impotence was 22‐fold and four‐fold higher in those receiving bendrofluazide, compared with placebo or a β blocker, respectively. ⁵ In this trial, impotence was the most frequent principal reason for withdrawal from antihypertensive therapy. Another smaller trial reported on by Chang et al. ⁷⁰ also found a higher frequency of decreased libido, difficulty in gaining and sustaining an erection, and trouble in ejaculating in thiazide‐treated patients. Multivariate analysis suggested that these findings were not mediated by either low‐serum K⁺ or by the observed fall in BP.

In a study by Wassertheil‐Smoller et al. ⁶⁹ problems with sexual interest, erection, and orgasm were greater among men receiving chlorthalidone compared with those given placebo or atenolol. Of note, in this trial, weight loss corrected the problem of chlorthalidone‐induced sexual dysfunction. Also, use of a diuretic in combination with other antihypertensive agents has been associated with a higher incidence of sexual dysfunction symptoms than with the use of a diuretic alone. The mechanism by which thiazides effect erectile function or libido is unclear, although it has been suggested that these drugs wield a direct effect on vascular smooth muscle cells and/or decrease the response to catecholamines; however, patients with diuretic‐related impotence can respond to sildenafil without any additional drop in BP. ⁷¹

Impotence and decreased libido are the more frequent sexual side effects with spironolactone. ⁶⁶ Gynecomastia, another fairly frequent complication of spironolactone therapy, may be associated with mastodynia and is typically bilateral. One study reported that 91 (13%) of 699 men prescribed spironolactone, alone or in association with another antihypertensive treatment, developed dose‐related gynecomastia that was reversible. At daily doses of ≤50 mg, the incidence of gynecomastia was 6.9%; at daily doses of ≥150 mg, the incidence was 52.2%. ⁷² The sexual side effects of spironolactone have been attributed to endocrine dysfunction; spironolactone is structurally similar to the sex hormones and inhibits the binding of dihydrotestosterone to androgen receptors, thus producing an increased clearance of testosterone. ⁷³ Eplerenone is another aldosterone‐receptor antagonist which is more selective than spironolactone and is devoid of the sexual side effects seen with spironolactone. ⁷⁴

DRUG ALLERGY

Photosensitivity dermatitis rarely occurs secondary to thiazide or furosemide therapy. ⁷⁵ HCTZ more commonly causes photosensitivity than do the other thiazides. ⁷⁶ Diuretics may rarely cause a more serious generalized dermatitis and, at times, even a necrotizing vasculitis. Cross‐sensitivity with sulfonamide drugs may occur with all diuretics, with the exception of ethacrynic acid; however, the frequency with which cross‐sensitivity occurs is much less common than was first thought and appears to be due to a predisposition to allergic reactions, rather than to specific cross‐reactivity with sulfonamide‐based drugs; thus, patients with a sulfonamide allergy that was not “extreme” (such as Stevens‐Johns on syndrome or a necrotizing vasculitis) in its original presentation can cautiously receive a thiazide or a loop diuretic. ⁷⁷

Severe necrotizing pancreatitis is a rare, but serious and potentially life‐threatening, complication of thiazide therapy. Acute allergic interstitial nephritis with fever, rash, and eosinophilia, although an uncommon complication of diuretics, is one that may result in permanent renal failure if the drug exposure is prolonged. ⁷⁸ Allergic interstitial nephritis may develop abruptly or some months after therapy is begun with a thiazide diuretic or, less commonly, it can occur with furosemide. ⁷⁹ Ethacrynic acid is chemically dissimilar to the other loop diuretics and can be safely substituted in diuretic‐treated patients who experience any of these allergic complications.

CARCINOGENESIS

Twelve clinical studies, three cohort (1,226,229 patients with 802 cases of renal cell carcinoma) and nine case‐controlled studies (4185 cases of renal cell carcinoma and 6010 controls), have evaluated the association between the use of diuretics and renal cell carcinoma. In all case‐controlled studies, the odds were greater for patients being treated with diuretics to develop renal cell carcinoma (average odds ratio of 1.55). The risk of renal cell carcinoma appeared to be related not to the average daily diuretic dose, but rather to the duration of the diuretic treatment. Unlike the association between diuretics and renal cell carcinoma, no association has been found between diuretic therapy and breast cancer. The issue of renal cell carcinoma occurring with diuretic therapy at the current time remains one incompletely resolved. ⁸⁰ , ⁸¹ , ⁸²

ADVERSE DRUG INTERACTIONS

Loop diuretics can potentiate aminoglycoside nephrotoxicity. ⁸³ By causing hypokalemia, diuretics increase the risk of digitalis toxicity. ⁸⁴ Plasma lithium (Li⁺) concentrations can increase with thiazide therapy with significant volume contraction due to the associated increase in Li⁺ absorption. ⁸⁵ However, some diuretics, such as chlorothiazide or furosemide, with significant carbonic anhydrase inhibitory, can increase Li⁺ clearance, thus leading to a fall in blood levels. ⁸⁶ , ⁸⁷ Whole‐blood Li⁺ should be closely monitored in patients administered Li⁺ in conjunction with diuretics. Nonsteroidal anti‐inflammatory drugs can both antagonize the effects of diuretics and predispose diuretic‐treated patients to a form of functional renal insufficiency. The combination of indomethacin and triamterene may be particularly dangerous, in that acute renal failure can be precipitated. ⁸⁸

CONCLUSIONS

Diuretic‐related side effects occur more commonly with loop and/or K⁺‐sparing agents and are less so with thiazide‐type diuretics, as they are currently used in the treatment of hypertension. Diuretic‐related side effects, particularly with loop and/or K⁺‐sparing agents, are not uncommon causes for hospitalization due to hypotension, renal impairment, electrolyte disturbances, and gout in the context of hospital admissions prompted by adverse drug reactions. ⁸⁹ Many diuretic‐related side effects can be avoided or effectively tempered by selection of the lowest dose necessary for effective BP and/or volume control.