Kidney International, Vol. 28 (1985), pp. 988—1000

NEPHROLOGY FORUM

Case presentations

A 52-year-old woman with a history of alcohol abuse (more than 6beers daily for years) was admitted to the hospital because of acuteprecordial pleuritic pain that had begun abruptly 12 hours earlier. Threeweeks prior to admission she sought medical attention, for the first timein 20 years, because of upper abdominal pain. An evaluation revealedgastritis; she was also found to have hypertension, diabetes mellitus,and chronic obstructive pulmonary disease. She discontinued ingestingalcohol and began receiving treatment wth cimetidine (300 mg 4 timesdaily), theophylline (200 mg twice daily), and clonidine, 0.1 mg, pluschlorthalidone, 15 mg, in a combination preparation. One week prior toadmission she developed increased fatigue and nausea, but she deniedvomiting or diarrhea.

Physical examination at the time of admission revealed a disheveled,thin, white female in no acute distress. The blood pressure was 190/100mm Hg with a pulsus paradoxus of 8mm Hg; the pulse was 86 beats/mmand regular. The skin was dry, no peripheral edema was present, andthe jugular venous pressure was normal. Funduscopic examinationrevealed grade-I hypertensive retinopathy. Breath sounds were dimin-ished bilaterally and a few scattered expiratory wheezes were heard. Atwo-component, scratchy pericardial friction rub was apparent at theleft lower sternal border; an S4 was also heard. Cardiac examinationwas otherwise unremarkable. The liver, palpable 3 cm below the rightcostal margin, was slightly tender.

A chest film on admission showed hyperinflated lungs withoutinfiltrates and a normal-sized heart. The electrocardiogram revealedvoltage changes consistent with left ventricular hypertrophy; frequent

Presentation of the Forum is made possible by grants from CIBAPharmaceutical Company, Geigy Pharmaceuticals, and Sandoz,Incorporated.

© 1985 by the International Society of Nephrology

premature atrial and ventricular beats also were present. There was a 2to 3 mm elevation of the ST segments in leads III and V2-V4; the QTinterval was prolonged. Admission laboratory data revealed: hemoglo-bin, 14 g/dl; hematocrit, 44%; white blood cell count, 20,800/mm3;BUN, 8 mg/dl; creatinine, 0.8 mg/dl; and glucose, 154 mg/dl. Serumelectrolytes were: sodium 116 mEq/liter; potassium, 2.3; chloride, lessthan 80; and bicarbonate, 40 mEq/liter. Arterial blood gas values on100% oxygen were pH, 7.51; PCO2, 41; and P02, 201 mm Hg. Serummagnesium was 2.2 mg/dl; calcium, 8.8 mg/dl; and phosphorus, 4.6 mg/dl. Serum theophylline level was in the nontoxic range. Serum osmolali-ty was 242 mOsmlliter, and a spot urine contained sodium, less than 10mEq/liter; chloride, less than 15 mEq/liter; and potassium, 21 mEq/liter.The CPK levels remained normal and there was no MB band.

Two hours after admission the patient had a seizure and the EKGmonitor revealed ventricular tachycardia followed by fibrillation. Car-dioversion was successful and the patient was treated with lidocaine.She subsequently developed atrial fibrillation, which persisted despitecorrection of the hyponatremia and hypokalemia with administration ofsaline and potassium chloride.

Discussion

DR. RICHARD L. TANNEN (Director, Division of Nephrology,and Professor of Internal Medicine, The University of MichiganMedical Center, Ann Arbor, Michigan): This woman presentedwith severe hypokalemia 3 weeks after the initiation of diuretictherapy. In association with the profound hypokalemia, shemanifested ventricular ectopy, which culminated in an episodeof ventricular tachycardia and fibrillation. In the course of mydiscussion, I will address the role played by diuretics in thedevelopment of her hypokalemia, and I will also consider therelationship between hypokalemia and the development of thecardiac arrhythmias.

Although it is widely recognized that diuretic therapy com-monly results in hypokalemia, considerable controversy existsregarding the clinical importance and appropriate managementof this complication [1—81. In this Forum I will (1) review thepathophysiology of diuretic-induced hypokalemia; (2) summa-rize the clinical characteristics of this condition; and (3) consid-er when potassium maintenance therapy might be indicated inpatients receiving diuretic drugs. I will use the term "potassiummaintenance therapy" to refer to the use of either potassium-chloride supplements or potassium-sparing drugs to maintainnormal potassium levels during diuretic therapy.

Pathophysiology of diuretic-induced hypokalemiaDiuretics that either inhibit reabsorption in the thick ascend-

ing limb of Henle (drugs such as furosemide, ethacrynic acid,bumetanide) or in the early distal tubule (such as thiazides,chlorthalidone) are kaliuretic and induce hypokalemia wheningested on a chronic basis (Table 1) [1]. The kaliuresis pro-duced by these agents results from increased delivery of tubular

Diuretic-induced hypokalemiaPrincipal discussant: RICHARD L. TANNEN

The University of Michigan Medical Center, Ann Arbor, Michigan

EditorsJORDAN J. COHENJOHN T. HARRINGTONJEROME P. KASSIRERNIcoLAos E. MADIAS

Managing EditorCHERYL J. ZUSMAN

Michael Reese Hospital and Medical CenterUniversity of Chicago Pritzker School of Medicine

andNew England Medical Center Hospitals

Tufts University School of Medicine

988

Diuretic-induced hypokalemia 989

fluid to the distal convoluted tubule and also possibly fromincreased sodium delivery to the distal nephron, especially thecortical collecting duct [9—12]. The effect of increased distalsodium delivery on potassium excretion may be enhanced bythe high aldosterone levels that are frequently present underconditions of diuretic use.

In addition to the kaliuresis that results directly from thediuretic properties of these drugs, at least one other pathophysi-ologic process plays a critical role in the development andmaintenance of hypokalemia. Hypokalemia is sustained inthese patients in part as a result of the accompanying chloridedepletion. The role of chloride depletion was elucidated ele-gantly by Schwartz and colleagues [13]. The major clinicalcauses of primary chloride depletion are diuretic therapy andthe loss of gastric secretions from either vomiting or nasogastricdrainage. Because a low-sodium diet is also a low-chloride diet,chloride depletion results from diuretic use because theseagents are chioruretic as well as natriuretic. As clearly demon-strated in a sequence of studies by Schwartz and coworkers, aprimary deficit of chloride results in potassium depletion withhypokalemia, metabolic alkalosis, and mild volume contraction[13]. The volume contraction, which persists despite the provi-sion of adequate dietary sodium, is not necessarily clinicallyapparent. Similarly, so long as correction of the chloride deficitis prevented—that is, by ingestion of a low-chloride diet—thealkalosis is sustained because the daily production of endoge-nous acid, otherwise more than sufficient in quantity to accom-plish correction, is balanced by renal acid excretion. Morerelevant to the present discussion is that the hypokalemia andpotassium deficiency persist despite the provision of adequatepotassium in the diet. All of the electrolyte abnormalities can becorrected upon provision of adequate chloride to restore theunderlying deficit. These experimental observations providethe cornerstone for the clinical principle that diuretic-inducedhypokalemia can only be managed successfully with the chlo-ride salt of potassium, KCI. Other potassium salts such as thoseof citrate, gluconate, and phosphate are ineffective in correctingdiuretic-induced hypokalemia secondary to chloride depletion.It should be appreciated that this dependency on the chloridesalt of potassium applies only to potassium replacement therapyand would not necessarily be the case if potassium-sparingagents were used, because such agents interfere directly withthe renal tubule's capacity to secrete potassium.

The tubular mechanisms accounting for how chloride deple-tion induces renal potassium losses are still unresolved. It hasbeen suggested that increased aldosterone secretion secondaryto volume depletion may account for the kaliuresis [14]. Al-though elevated aldosterone levels may foster more intensekaliuresis under conditions of diuretic use, it appears thatchloride depletion can result in potassium depletion even in theabsence of increased aldosterone levels [15—17]. An elevatedpH per se can increase renal potassium secretion and could playa role in the potassium loss, but the magnitude of this effect isunclear [18, 191. Finally, recent evidence suggests that theconcentration of chloride in distal tubule fluid can directlyinfluence potassium secretion. Although the precise mechanismhas not been elucidated, hypochioremia stimulates net potassi-um secretion [20].

Thus it is clear that diuretics with sites of action at either theascending limb or the early distal tubule can produce kaliuresis

Table 1. Pathophysiology of diuretic-induced hypokalemia

I. Increased delivery of fluid to distal convoluted tubule2. Increased delivery of sodium to distal nephron (? collecting duct)3. Chloride depletion

a. Low C1 concentration in distal tubular fluidb. Alkalosisc. Increased aldosterone

4. Direct diuretic-induced stimulation of K secretion (?)

and potassium depletion by at least two mechanisms. Recentexperiments in our laboratory have raised the possibility thatdiuretics, especially furosemide, might stimulate potassiumsecretion in yet another way, namely by a direct tubular effect[19]. These experiments, which utilized isolated perfused kid-neys from potassium-depleted rats, indicated that furosemidecan induce a kaliuresis of a magnitude that cannot be accountedfor entirely by an increase in either urine flow rate or sodiumexcretion. Furthermore, studies with amiloride indicated thatthis direct effect of furosemide on potassium excretion resultedfrom an increase in potassium secretion rather than from adecrease in potassium reabsorption. Whether hydrochiorothia-zide can also stimulate potassium secretion directly was notdefinitively resolved by these experiments but, if hydrochloro-thiazide has this property, its effect is much less prominent thanis that of furosemide. What, if any, role a direct effect ofdiuretics on potassium secretion might play in vivo has not beenexamined.

Clinical characteristics of diuretic-induced hypokalemiaThe clinical characteristics of the abnormality in potassium

metabolism accompanying chronic diuretic use have been de-fined carefully in the past decade [1, 2] and are summarized inTable 2. We now know that the fall in serum potassiumconcentration occurs rapidly and reaches a plateau within aboutone week after diuretic therapy is initiated [21, 22]. As we willconsider shortly, the degree of hypokalemia with standarddoses of diuretics (hydrochlorothiazide, 50—100 mg/day orequivalent doses of other thiazide diuretics; chlorthalidone, 50—100 mg/day; furosemide, 40—80 mg/day) is modest [22—44]. Lessthan 7% of patients taking thiazides and less than 1% of patientstaking furosemide exhibit a decrease in serum potassium below3.0 mEq/liter [1, 211. For this reason, it is appropriate toconsider the possibility of other causes for hypokalemia inpatients whose serum potassium falls to less than 3.0 mEq/literduring diuretic therapy. The patient today, whose serum potas-sium was 2.3 mEq/liter on admission, is a case in point. In mostinstances primary aldosteronism is high on the list of diagnosticpossibilities when hypertensive patients manifest this degree ofhypokalemia in response to treatment with diuretics. Thispatient's hyponatremia and low urinary concentration of sodi-um (less than 10 mEq/liter) and chloride (less than 15 mEq/liter)at the time of presentation, and her ability to maintain normalpotassium levels later in the hospital course, made this anunlikely possibility. Despite her denial, the impressive metabol-ic alkalosis (serum HC03 of 40 mEqlliter) along with the lowurinary chloride concentration strongly suggest the possibilityof vomiting. Alternatively, her intake of diuretic could havemarkedly exceeded the recommended dose. It is difficult todistinguish between these latter two possibilities after the fact,

990 Nephrology Forum

Table 2. Clinical characteristics of diuretic-induced hypokalemia

1. Plasma K stabilizes in one week2. 1%—7% of patients exhibit decreased plasma K to less than 3.0

mEq/liter3. Degree of hypokalemia related to Na intake

a. Very low or high Na accentuates hypokalemiab. Modest Na restriction (100 mEq/day) minimizes the degree of

hypokalemia4. Females are more susceptible to hypokalemia than are males5. Whites are more susceptible to hypokalemia than are blacks6. No clear difference in the degree of hypokalemia exists whether

treatment is for hypertension or for CHF

since they both result in chloride depletion and therefore in asimilar clinical picture.

Several other clinical features of chronic diuretic use areworth mentioning. Women seem to be more prone to hypokale-mia than men seem to be [45], and some data suggest thatwhites are more susceptible than are blacks [44]. The degree ofhypokalemia is minimized by moderate sodium restriction(—60—110 mEqiday). Both severe sodium restriction (

Diuretic-induced hypokalemia 991

Table 4. Adverse consequences of diuretic-induced hypokalemia

1. Ventricular arrhythmiasa. Increased ventricular premature beatsb. Increased risk of ventricular tachycardia and ventricular

fibrillation with acute MI2. Impaired hypertensive control3. Glucose intolerance4. Lipid abnormalities (?)

ventricular arrhythmias in stable patients with or withoutunderlying cardiac disease, and (2) an increased risk of ventric-ular tachycardia or fibrillation in the setting of acute myocardialinfarction.

It is widely appreciated that hypokalemia can provoke ar-rhythmias in patients treated with digitalis. Several recentpublications have suggested that hypokalemia per se increasesthe risk for development of ventricular arrhythmias [33, 52—771.The ventricular ectopy varies from an increase in the frequencyof unifocal ventricular premature beats to the development ofcomplex ventricular arrhythmias, including multifocal ventricu-lar premature beats, ventricular couplets, and/or ventriculartachycardia. In a recent critical review of this topic, Harring-ton, Isner, and Kassirer pointed out that a clear associationbetween hypokalemia and ventricular arrhythmias could beestablished only if: (1) other causes of ventricular ectopicactivity were rigorously excluded; (2) a firm temporal relation-ship between the development of hypokalemia and ventricularectopy were demonstrated; and (3) correction of the underlyingpotassium deficit was associated with alteration of the ventricu-lar ectopy [5]. Indeed, of the 17 reports they analyzed, only onefulfilled these three criteria [5, 52—67], and all but one were casereports or retrospective analyses of data.

Today's patient illustrates the difficulty in delineating a clearassociation between hypokalemia and arrhythmias. Hypokale-mia certainly could have been a significant factor in eitherdirectly producing or substantially increasing the susceptibilityfor the development of ventricular ectopy. However, in view ofthe underlying pericarditis and the severe hypoosmolality com-plicated by a seizure, it is impossible to delineate the preciserole, if any, hypokalemia played in the abnormal rhythm.

Since publication of the review by Harrington et al, severalprospective studies have appeared that specifically address thisquestion in patients with diuretic-induced hypokalemia [33, 55,68—77]. Table 5 summarizes these prospective studies. In threeof them, in which a total of 50 patients were examined, and alsoin the chronic phase (3 to 50 month treatment) of the largeMedical Research Council (MRC) study, in which an additional74 patients were examined, an increase in the frequency ofventricular premature beats was noted to occur with the devel-opment of diuretic-induced hypokalemia [33, 55, 70—74, 77].Approximatey 25% to 50% of patients exhibited a susceptibilityto more frequent ventricular premature beats and, furthermore,the development of complex ventricular ectopy was also notedin 15% to 30% of patients in these four positive studies. Incontrast to these findings is the absence of a significant changein ventricular ectopy in the short-term (9 to 10 week) MRCstudy of 16 patients as well as in three other studies encompass-ing 48 patients [68, 69, 71—73, 75, 76].

A similar lack of uniformity is apparent in the data collected

during normalization of plasma potassium concentration. Threestudies encompassing 35 patients reported an improvement inventricular premature beats [55, 71—74]; one study reportedimprovement only with simultaneous potassium and magne-sium therapy [70]; and one study of 16 patients found noimprovement with an increase in plasma potassium concentra-tion to 3.7 mEq/liter [68].

It is of note that Caralis and coworkers found evidence ofhypokalemia-induced arrhythmias only in a subset of patientswho had preexisting, significant cardiac disease as documentedby abnormal results on EKGs, chest films, and 24-hour Holtermonitor [74]. In all the other studies, criteria specificallyexcluded patients with angina, previous myocardial infarctions,abnormal exercise tests, or high-grade ventricular ectopy. Onlypatients without any evidence of heart disease or with mildcardiomegaly documented by a chest film, left ventricularhypertrophy by voltage criteria, or nonspecific EKG changeswere included in these other studies. Thus, the possibility existsthat a specific subset of patients, perhaps those with underlyingcardiac disease, is at particular risk; but if so, we certainly lackthe capacity to identify them uniformly.

Nevertheless, when this recent group of prospective studiesis considered in conjunction with the other reports in theliterature, the evidence seems to indicate that diuretic-inducedhypokalemia can result in ventricular ectopy and that the typeof ectopy can fall into the complex category, which hasominous implications. The data suggest that the risk of complexventricular ectopy may be higher when the serum potassium is3.0 mEq/liter or less; however, it seems clear that this complica-tion can also arise with milder degrees of hypokalemia.

Even given these conclusions, to truly assess adverse conse-quences it is necessary that we know the actual risk ofmorbidity and mortality associated with the development ofventricular ectopy. One recent study that followed patients forup to 10 years suggests that asymptomatic, healthy individualswithout underlying cardiac disease experience no increased riskof death from frequent and complex ventricular ectopy thatoccurs spontaneously [78]. Whether this prognosis would applyto patients with hypokalemia-induced arrhythmias is unknown.Furthermore it is widely accepted that ventricular ectopypredisposes to sudden death in patients with underlying isch-emic heart disease [78—80]. Because the patient populationreceiving diuretics for hypertension is at increased risk forischemic heart disease, and those undergoing therapy for con-gestive heart failure have established cardiac disease, one mightspeculate that diuretic-induced ventricular ectopy is likely to bemore ominous than is spontaneously occurring ectopy in anotherwise healthy individual.

Several recent publications have addressed the role of hypo-kalemia in the occurrence of arrhythmias at the time of acutemyocardial infarction [56, 57, 64, 81—86, 87]; this topic wasreviewed recently by Solomon [81]. An analysis of the datafrom six studies, including a total of 8327 patients, is providedin Table 6 and reveals a number of salient and reasonablyconsistent observations. First, and most important, the inci-dence of serious ventricular arrhythmias (that is, ventriculartachycardia and/or fibrillation) was increased significantly inevery study in the presence of hypokalemia (defined as a serumpotassium less than either 3.6 or 3.5 mEq/liter). In most studies,the incidence of combined ventricular tachycardia and fibrilla-

992 Nephrology Forum

Table 5. Diuretic-induced ventricular arrhythmias in hypertensive patients

Induction of hypokalemia Correction of hypokalemia

Number of Mean serum K Complex Mean serum KReference patients mEq/liter VPBSb VPBs mEqiliter VPBs

Holland et al [55, 77] 21 3.0 1' b t 4.0 bHollifield & Slaton [33, 701 13 3.4 ¶ 4.2Caralis et al [74] 16 3.5 ¶ 1 4.1MRC [7 1—73]

Acute (9—10 weeks) 16 3.6 NCb NCChronic (3—50 months)

Papademetriou et al [68]

741216

3.6

2.8

1 ¶3.53.7 NC

Papademetriou et al [69] 15 3.0 NC NCMadias et al [75] 20 3.0 NC NCLeif et al [761 13 3.0 NC NC

a Numerals in brackets are reference numbers.b Symbols: VPB, ventricular premature beat; ¶, increase; ., decrease; NC, no change.

tion during the first 1 or 2 days of hospitalization was reported[56, 57, 83, 84]; however, in the two series with the lowestincidence of arrhythmias, the data are provided in a differentfashion. Hulting only reported the incidence of ventricularfibrillation during the first 12 hours of hospitalization [64], andJohansson and Dziamski only provided data on the incidence ofmalignant arrhythmias resulting in circulatory arrest [85]. Al-though these specific events occurred less frequently than didthe combination of ventricular tachycardia and ventricularfibrillation, in general the increased frequency in the hypokale-mic group followed a similar pattern and was from two- tofourfold greater.

Whereas the likelihood of malignant arrhythmias complicat-ing an acute myocardial infarction appears to be higher in thepresence of hypokalemia, the data suggest that the frequency ofventricular premature beats is not increased [56, 57, 81, 841.Consistent with this view is one recent series (not included inTable 6) that analyzed the incidence of all arrhythmias in 103patients with an acute myocardial infarction and found nosignificant difference between hypokalemic and normokalemicpatients, although the frequency of arrhythmias was 72% withhypokalemia compared with 58% in its absence (86).

Several other important observations emerge from thesestudies. Although patients taking digitalis were not specificallyexcluded, the data in three studies were analyzed to address thepossible role of digitalis. The increased frequency of malignantarrhythmias with hypokalemia appeared to be independent ofthis agent [56, 57, 81, 84].

Evaluation of the role of diuretics suggests that they are apotential significant risk factor because of their propensity tocause hypokalemia. Indeed, in every study the incidence ofhypokalemia was approximately twofold greater or more inpatients treated with diuretic medication [56, 57, 64, 83—85].Furthermore, normokalemic patients treated with diureticswere not at increased risk for malignant arrhythmias in compar-ison with those not receiving diuretics; this finding indicatesthat it is the increased likelihood of hypokalemia rather than thediuretics per se that results in an increased risk for thiscomplication [56, 57, 83, 85].

Approximately 10% of the patients who have not receivedantecedent therapy with diuretics exhibit hypokalemia in asso-

ciation with an acute myocardial infarction [57, 83—85]. Thisraises the possibility that hypokalemia might reflect a shift inpotassium into the intracellular compartment, which might bethe result of the increased sympathic adrenal activity that canaccompany an acute myocardial infarction [88]. Indeed, thepossibility arises that hypokalemia is a marker, rather than theprimary cause of the arrhythmogenic potential, both of whichresult from an increase in cathecholamines. Nonetheless, thedegree of hypokalemia induced by epinephrine appears to bemore profound in the presence of preexisting diuretic-inducedpotassium depletion; this additive effect on lowering serumpotassium could increase the likelihood that a serious arrhyth-mia will develop [88]. Another criticism of a primary role forpotassium depletion in the induction of ventricular arrhythmiasis that in many studies the level of potassium used for analysiswas not necessarily recorded prior to the development of thearrhythmia [57, 64, 85]. The pitfall of measurement after anacute episode is emphasized by the observations by Thompsonand Cobb, who found that 49% of patients resuscitated out ofhospital for ventricular fibrillation were hypokalemic, as com-pared with 19% of patients admitted to the hospital with anacute myocardial infarction, and 9% of ambulatory patientswith coronary heart disease [53]. However, the percentage ofpatients taking diuretics did not differ among these three groupsand, in the resuscitated group, the serum potassium afterresuscitation did not differ between those who had and thosewho had not received antecedent diuretic therapy. The authorsconcluded that the high incidence of hypokalemia in the resusci-tated group resulted from the resuscitative effort itself, and theycautioned against inferring from these data that hypokalemiapredisposes to ventricular fibrillation. Nevertheless, in defenseof the suggestion that preexisting potassium depletion mightpredispose to arrhythmias, several of the reports in Table 6contained data based solely on the serum potassium level at thetime of admission to the hospital [56, 83, 84].

In none of the studies cited in Table 6 was a significantincrease in malignant arrhythmias reported in the diuretic- ascompared with nondiuretic-treated patients. However there isalso no information provided regarding which diuretic-treatedpatients received potassium maintenance therapy [56, 57, 64,83—85]. Thus, although these observations raise significant

Diuretic-induced hypokalemia 993

Table 6. Relationship of serum potassium to ventricular tachycardia(VT) and ventricular fibrillation (VF) in patients with acute

myocardial infarction

Number of

Incidence of VT/VF

Hypokalemiaa NormokalemiaReference patients % %

Solomon & Cole [57] 151 48 21Nordrehaug [56] 1035 29 17Hulting [64]b 537 8 2Dyckner et al [83] 676 50 37Reuben [841 586 21 5Johansson & Dziamski

[85]C 5342 16 7

a Serum potassium < 3.5 or 3.6 mM.b VF in first 12 hours after admission.

Malignant arrhythmias producing circulatory arrest.

concern that antecedent diuretic-induced hypokalemia might bea risk factor for sudden death in patients sustaining an acutemyocardial infarction, the data available do not definitivelyresolve this issue.

Two further studies deserve mention, however, because theydo heighten the concern that diuretic therapy might predisposetoward sudden death. Ruberman and colleagues found a two-fold increase in sudden coronary death in patients treated withdiuretics during the 5-year period following an acute myocardialinfarction [80]. Furthermore, in the Multiple Risk Factor Inter-vention Study (MRFIT), the incidence of sudden death wasincreased twofold in the hypertensive group with EKG abnor-malities receiving special intervention as compared with thosereceiving the usual care [89, 90]. Speculation has been raisedconcerning a possible role for diuretic therapy, because diuret-ics were the prescribed step-l antihypertensive drug in thespecial intervention group. Thus, a substantial body of evidenceraises serious concerns about the possibility that diuretic-induced hypokalemia might predispose toward sudden death,especially in settings of underlying cardiac disease; but theevidence is still not conclusive.

The most dramatic potential adverse effect of diuretic-in-duced hypokalemia is ventricular arrhythmia and sudden death,but there are other potential consequences of persistent, mildhypokalemia. The relationship between potassium homeostasisand blood pressure recently has been a subject of considerableattention. The data indicate that both high and low potassiumlevels can exert antihypertensive effects [91]. Indeed, potassi-um interacts with so many blood pressure regulatory systemsthat it is impossible to predict, a priori, how diuretic-inducedhypokalemia would alter blood pressure in patients undergoingtreatment for hypertension. This question was investigatedrecently in a carefully controlled, double-blind, crossover ex-amination of potassium supplementation in hypertensive pa-tients treated with diuretics [92]. Correction of hypokalemiaresulted in a small but statistically significant decline in bloodpressure. It would be premature for us to extrapolate broadlyfrom this small study on a selected group of hypertensives tothe population at large, but these data do suggest that treatmentof hypokalemia might improve blood pressure control in at leastsome hypertensive patients.

Glucose intolerance is an additional, well-defined complica-tion of diuretic-induced hypokalemia [93, 941. Whereas some

data suggest that diuretics per se can interfere with glucosemetabolism [95], the bulk of evidence indicates that hypokale-mia accounts for the development of glucose intolerance inpatients treated with diuretics [96—103]. Experimental potassi-um depletion induces glucose intolerance, and insulin-clampstudies indicate that the primary abnormality is a defect ininsulin release with no change in the insulin sensitivity of thetissue [96, 97]. Hypokalemia also can impair the conversion ofproinsulin to insulin, but this abnormality has not been repro-duced with experimental potassium deficiency [97, 104]. Diuret-ic-associated glucose intolerance does not occur when potassi-um depletion is prevented [98], correlates with the developmentof hypokalemia [100, 103], and can be corrected by normaliza-tion of the potassium deficit [99, 101, 1021.

The magnitude of potassium deficiency required for thedevelopment of glucose intolerance is modest and consistentwith the degree of depletion that develops in many diuretic-treated patients. Rowe observed glucose intolerance followingexperimentally produced potassium deficits of 1,0% to 8.4%that were documented by total body potassium counting [961.Other authors using balance techniques have observed abnor-malities in glucose metabolism with mild to moderate potassiumdepletion (that is, deficits ranging from 200 to 569 mEq) [97,105].

These observations support both the primacy and likelihoodof potassium depletion in the development of disordered glu-cose homeostasis. It has been estimated that 30% of hyperten-sive patients receiving thiazide diuretics exhibit abnormal glu-cose tolerance, and population studies with adequate controlsclearly indicate that development of glucose intolerance iscausally related to diuretic therapy [100, 101, 103]. In mostinstances, only mild glucose intolerance develops and does notrequire hypoglycemic therapy; but rarely overt diabetes be-comes manifest [4]. The potential risk, therefore, relates mainlyto whether the subtle abnormalities in glucose metabolismpredispose toward accelerated cardiovascular disease. Thisissue is unresolved in the selected population of patients withdiuretic-induced glucose intolerance.

A potential relationship between diuretic use and lipid abnor-malities is less clear than for glucose intolerance [93]. Most butnot all studies suggest that therapy with thiazide or chlorthali-done results in an increase in cholesterol and triglycerideconcentrations and in the ratio of low- to high-density lipopro-tein cholesterol [93, 101, 106—113]. Two large, well-controlledinvestigations both uncovered a significant lipid alteration inpatients treated with diuretics [106, 107]. The mechanismunderlying the abnormality in lipid metabolism and its relation-ship to hypokalemia are unknown [931. The latter issue requirescareful study, and the cardiovascular risk of the lipid abnormali-ty also needs to be delineated.

Potassium maintenance therapy, A decision regarding theappropriateness of potassium maintenance therapy must bal-ance the adverse consequences of such a treatment programagainst the litany of potential hypokalemia-induced side effects.Potassium maintenance therapy can be accomplished success-fully either with the use of potassium chloride supplements orwith the use of potassium-sparing diuretics (such as aldactone,triamterene, or amiloride). Each of these approaches can cor-rect diuretic-induced hypokalemia, but it should be recognizedthat KC1 in doses of 40 to 60 mEq/day is needed for adequate

994 Nephrology Forum

therapy [1141. The data also suggest that potassium-sparingagents may be somewhat more effective in normalizing potassi-um than are potassium supplements [211.

Regardless of which approach to potassium maintenancetherapy is used, the most serious risk is the development ofhyperkalemia, a life-threatening complication. Whereas the riskof this complication has not been well-defined for the non-hospitalized patient, 3.9% of 1910 hospitalized patients whoreceived oral potassium chloride developed significant hyperka-lemia (potassium greater than 5.5 mEq/liter) [1151. Of hospital-ized patients treated with spironolactone without potassiumsupplements, hyperkalemia occurred in 5.7% [116]. Compara-bly large studies have not been reported with the potassium-sparing drugs triamterene and amiloride, but the risks ofhyperkalemia with these drugs are well documented. Regard-less of whether potassium supplements or potassium-sparingdiuretics were used, renal impairment was an important factorpredisposing to hyperkalemia. The elderly and patients whoreceived potassium in addition to potassium-sparing drugs arealso at increased risk for the development of hyperkalemia [6,115, 1161. Although these data indicate that hyperkalemia is arisk of potassium maintenance therapy, they do not establishthe incidence of clinically significant hyperkalemia in ambula-tory patients with acceptable levels of renal function treatedselectively with either potassium supplements or potassium-sparing agents.

In addition to hyperkalemia, other adverse consequences ofpotassium maintenance therapy can occur that are drug-specif-ic. All forms of KC1 (liquid, wax matrix, and polymer-coatedmicroencapsulated products) entail a risk of gastrointestinalulceration, although the incidence of clinically significant ulcer-ation appears to be low. Spironolactone can cause estrogen-likeside effects, and triamterene can produce renal stones as well asfolate-deficiency anemia [117]. In addition, both these agents aswell as amiloride are associated with gastrointestinal sideeffects, and there also exists the risk of drug-induced allergicmanifestations.

Finally, a substantial economic cost is associated with potas-sium maintenance therapy, regardless of whether potassiumsupplements or potassium-sparing diuretics are employed. Har-rington and Kassirer estimated a few years ago that $250 millionper year is spent for potassium treatment in the U.S. [5].

Therapeutic recommendationsGiven this body of information, the physician is confronted

with the need to make a decision regarding the use of potassiummaintenance therapy. All would welcome a well-designed trialto assess the risk-benefit ratio of potassium maintenance treat-ment in order to arrive at a therapeutic strategy, but it seemsunlikely that a study of appropriate scope and design will becarried out in the near future. Hence, a treatment plan must beformulated based on the incomplete data currently available.

My personal biases are listed in Table 7. There is generalagreement that potassium maintenance treatment is indicated inpatients receiving digitalis, in those predisposed to hepaticcoma, or in those in whom serum potassium falls to less than 3.0mEq/liter [1,2,6]. In view of the well-defined risks of hypokale-mia-induced arrhythmias in patients treated with digitalis, andin view of the capacity for hypokalemia to induce hepaticencephalopathy by stimulating renal ammonia production,

Table 7. Indications for potassium maintenance therapy

1. Digitalis therapy2. Predisposition to hepatic coma3. Serum potassium < 3.0 mEq/liter4. Development of glucose intolerance5. Underlying cardiac disease6. Symptoms attributable to hypokalemia

there is a broad concensus that the risks of hypokalemia inthese settings outweigh those of therapy. Similarly, when serumpotassium falls below 3.0 mEq/liter, the likelihood of hypokale-mia-induced cardiac, muscular, and gastrointestinal symptomsare increased and, despite the absence of rigorous proof, thereappears to be no argument that the risk of therapy is justifiedunder this circumstance.

The last three indications listed in Table 7 are more contro-versial. In my view, treatment is indicated in any patient whodevelops overt glucose intolerance, because this disorder mayrepresent a pathologic sequela of hypokalemia. Whereas thelong-term risk of mild glucose intolerance is undefined, the risksof potassium maintenance therapy are small and seem appropri-ate to forestall the potential sequelae of accelerated cardiovas-cular disease.

The fifth indication, that is, to treat any patient with underly-ing cardiac disease, evokes the most heated debate. It is basedon a judgment regarding the potential risk for hypokalemia-induced ventricular arrhythmias both under stable conditionsand in the setting of an acute myocardial infarction. Myrationale is as follows: Although some studies do not reveal anincreased risk of ventricular arrhythmias in patients with diuret-ic-induced hypokalemia, several carefully performed evalua-tions have uncovered a statistically significant alteration. Fur-thermore, complex as well as unifocal ventricular arrhythmiashave been delineated. Because the failure to detect an effectdoes not prove its absence, the disparate findings can beinterpreted as suggesting that the groups selected for studydiffered, perhaps in ways we are not yet sophisticated enough toidentify. In this regard, the study by Caralis and colleaguespoints strongly to the presence of underlying cardiac disease asan important risk factor for hypokalemia-induced ventriculararrhythmia [74]. In view of the documented increased risk ofsudden death in patients with ventricular arrhythmias superim-posed on underlying cardiac disease, I believe that potassiummaintenance therapy, judiciously monitored, is a preferablealternative to hypokalemia. My convictions are strengthened bythe data regarding the risk of hypokalemia in the setting of acutemyocardial infarction. Granted that each of these studies issubject to criticism, the evidence as it currently exists suggeststhat there is, rather than is not, a real risk of sudden death inpatients with preexisting potassium depletion.

These two lines of reasoning lead to my conclusion thatpotassium maintenance therapy should be used if cardiacdisease is evident. The corollary is that I would not routinelyadvise potassium maintenance therapy for patients receivingdiuretics who do not have evidence of underlying cardiacdisease. This subgroup would seem to be less prone to arrhyth-mias, to be at less risk if arrhythmias develop, and is probablyat less risk for development of an acute myocardial infarction.

My final recommendation is to initiate a trial of potassium

Diuretic-induced hypokalemia 995

maintenance therapy in any patient with symptoms that can beascribed to hypokalemia. This includes a variety of vaguecomplaints including malaise, fatigue, muscular weakness, andmuscle cramps. Granted that all these symptoms are nonspecif-ic, but they nevertheless can be manifestations of potassiumdepletion. If therapy relieves the symptoms, the response couldbe due either to specific correction or to a placebo effect. Ineither case, long-term potassium maintenance treatment wouldseem indicated, as the patient experiences a benefit and therisks are minimal.

Questions and answers

DR. DAVID BUSHINSKY (Renal Section, Mitchell Hospital,Chicago): Many authorities recommend that antihypertensivetherapy begin with the use of diuretic agents. In view of therisks of hypokalemia, would you recommend the use of agentsother than diuretics as first-line therapy in the treatment ofhypertension?

DR. TANNEN: This issue is certainly pertinent in view of thefinding in the MRFIT study that the special intervention group(which was treated with diuretics) had an increased incidence ofunexplained sudden death [89, 90], and the report by Rubermanet al of a twofold increase in the rate of sudden death in patientstreated with diuretics following an acute myocardial infarction[801. Neither of these studies definitively incriminates diureticsas the culprit however. Furthermore, the long-term risks of theother antihypertensive therapies are not well delineated. Di-uretics are relatively safe and have few unpleasant side effects,so I still recommend them as first-line antihypertensive drugs.However, if the patient has underlying cardiac disease, Iconcurrently recommend potassium maintenance.

DR. GARY TOBACK (Renal Section, Mitchell Hospital): Whatdrug do you use for the management of diuretic-inducedhypokalemia?

DR. TANNEN: I formerly favored potassium chloride supple-mentation. However, you have to use enough KC1—40 to 60mEq/day and in some instances more—to maintain normalserum potassium levels, and recent data in the literature suggestthat potassium-sparing agents are more effective in sustainingnormal serum potassium levels than is potassium chloride. Inaddition, both triamterene and amiloride are well tolerated bymost patients. My current bias leans towards amiloride, which Ithink is a little bit more predictable and perhaps better toleratedin terms of side effects. But that is a personal bias and I thinkone can make arguments for using the other forms of potassiummaintenance therapy as well.

DR. TOBACK: I would like to hear your comments about therelationship between hypokalemia and cardiac injury. It is nottalked about much anymore, but the older pathology literaturesuggested that the ingestion of a potassium-deficient diet couldresult in myocardial necrosis, which could become strikingwhen potassium deficiency was marked [118]. In addition todeath of myocardial cells, there appears to be a proliferation offibroblasts [118]. So if there is underlying heart disease, onecould argue that there might be cells suffering from borderlineischemia and that hypokalemia might cause enough furtherdamage to make sublethally injured cells become necrotic. Iwonder if you would comment on that.

DR. TANNEN: In the 40-year-old literature that you arequoting, Dr. Toback, the patients and the animal models

studied were severely potassium deficient, and the clinicalstudies often had other complicating features [1171. Whethermilder degrees of potassium depletion can in some way sensi-tize the cell to ischemic injury, for example, has not beenaddressed to my knowledge, but I think it is a reasonable issueto consider. Furthermore, potassium has direct effects on thevasculature: a high potassium concentration is vasodilatory anda low potassium concentration is vasoconstrictive. The Sodi-Pallares solution, which contains potassium, insulin, and glu-cose, might confer benefit by delivering more potassium toischemic sites and thus produce vasodilation and improvedperfusion of ischemic areas. I am unaware, however, of anyconvincing studies utilizing this approach [811.

DR. JOHN T. HARRINGTON: In contrast to the early report ofRubini in 1961 [51j, Healy and colleagues demonstrated noeffect of diuretic-induced potassium depletion on urine concen-trating ability in 7 hypertensive patients [38]. How do youreconcile these conflicting reports?

DR. TANNEN: The absence of a significant change in eitherconcentrating ability or urine acidification in the study by Healyis difficult to reconcile with the two carefully performed studiesof dietary induced potassium depletion, which have demon-strated significant effects. It should be noted, however, that thetrends in Healy's study were in a similar direction, althoughthey failed to achieve statistical significance.

DR. SUSAN FELLNER (Renal Section, Mitchell Hospital): Iagree with your recommendations regarding potassium mainte-nance. But I think that we as physicians could do a better job inhelping to prevent potassium depletion, particularly with regardto the issue of dietary salt restriction. A real commitment to theeducation of our patients about dietary sodium might prevent atleast some of the diuretic-induced hypokalemia that we see.

DR. TANNEN: I agree. If we could get patients to ingestapproximately 100 mEq/day of sodium, we would tend tominimize the degree of hypokalemia and would probably inmany instances improve their antihypertensive control as well.However, even if we do, some patients will, for example,develop serum potassium levels of 3.5 mEq/liter, which in myview are not normal. If these people have underlying cardiacdisease, I think a normal serum potassium should be main-tained, and this probably will require some form of potassiummaintenance therapy.

DR. JORDAN J. COHEN: We all agree that there is a largegroup of individuals who have no heart disease and who do notbecome very potassium-depleted when given diuretics. If suchindividuals are truly at risk from the modest hypokalemia theymanifest, then nature has endowed us with an extremely narrowmargin for error in maintaining potassium homeostasis. Therebeing no evidence to the contrary, I prefer to believe that themammalian organism is more resilient than that. Only if onebelieves otherwise would it be justifiable to expose patientstreated with diuretics routinely—and I emphasize routinely—tothe admittedly small but definite risk of hyperkalemia frompotassium maintenance therapy. Do you agree?

DR. TANNEN: I should have mentioned whom I would nottreat. In view of our current knowledge of the risks associatedwith hypokalemia and of the costs and risks of potassiummaintenance therapy, at present I don't think the weight ofevidence is sufficient to subject the group of patients who don'thave underlying, detectable cardiac disease to potassium-main-

996 Nephrology Forum

tenance therapy. I am not sure what fraction of the hyperten-sive population that encompasses, but certainly more patientsreceiving antihypertensive therapy don't have detectable cardi-ac disease than do.

DR. COHEN: There are several bothersome issues about thereported relationship between hypokalemia and malignant ven-tricular arrhythmias in the setting of an acute myocardialinfarction. First, as you noted, no firm evidence exists for acause-and-effect relationship; hypokalemia in this setting mightbe merely a marker, especially in view of the known effects ofcatecholamines in shifting potassium intracellularly. Second,because in many of the reported cases hypokalemia was docu-mented only after successful resuscitation from the arrhyth-mias, we must consider, as you did, the question of whetherhypokalemia was the consequence of the treatment (for exam-ple, alkali infusion) or of the stress (such as catecholamines) orof something else. Finally, as you also noted, a substantialsubset of patients who are admitted with an acute MI andhypokalemia and who develop ventricular tachycardia or fibril-lation have never been exposed to diuretics. And in those whohad been receiving diuretics, we don't know how many also hadbeen treated with potassium supplements or potassium-sparingagents and developed hypokalemia anyway. But the dilemmahere is the following: If one is convinced that hypokalemia inthe aftermath of a myocardial infarction promotes malignantarrhythmias, and one wants to do something to prevent thiscomplication, one would have to give potassium supplements toeverybody in the country who is at potential risk of developingan MI, whether they were receiving diuretics or not. Of course,such a strategy is absurd.

DR. TANNEN: Let's again consider the individuals who havean acute myocardial infarction and have not been preexposed todiuretics. Presumably their hypokalemia is not the result ofantecedent potassium depletion but instead reflects an intracel-lular shift of potassium. There really wouldn't be any way topretreat people with potassium to prevent that from happeningunless we were going to propose making everybody walkaround substantially hyperkalemic. Alternatively, some havesuggested that the antiarrhythmic effects of therapy with betablockers might be mediated through inhibition of epinephrine-mediated cellular potassium uptake, but there is no firm evi-dence in this regard 181]. Thus, with the possible exception oftherapy with beta blockers, there is no prophylactic therapy forthe group with normal potassium homeostasis that is potentiallyat risk for a myocardial infarction. But we can diminish thelikelihood for hypokalemia in the diuretic-treated patients atrisk for myocardial infarction by providing them with potassiummaintenance therapy.

DR. MARSHALL LINDHEIMER (Renal Section, Mitchell Hos-pital): Let's change the discussion from the pros and cons ofpotassium maintenance therapy to another aspect of the pa-tient's problems. This patient's serum sodium level of 116 mEq/liter reminds me of a series of patients described by Dr.Fichman and coworkers more than a decade ago [119]. In thesepatients, severe hyponatremia was associated with diuretic-induced hypokalemia, and one of the causes considered was the"inappropriate" secretion of vasopressin. Would you care tospeculate on the cause of the hyponatremia in this patient?

DR. TANNEN: In this particular patient, the low urinarysodium and chloride suggested that volume depletion at least

partially accounted for the hyponatremia. Although her hypo-natremia ultimately resolved in response to concurrent treat-ment with NaCI and KC1, it didn't correct as rapidly as I wouldhave anticipated with uncomplicated sodium depletion. Noother reason for the development of hyponatremia was detect-ed. Prior to discharge from the hospital, she was able to sustaina normal serum sodium concentration in the absence of specifictherapy. Thus she might fit into the group of diuretic-treatedhypokalemic patients who appear to manifest "inappropriate"ADH secretion. The patients reported by Fichman et al devel-oped hyponatremia in the absence of detectable volume deple-tion, and these authors speculated that hypokalemia might havemodified either the osmoreceptors or the releasing mechanismfor vasopressin. Perhaps supporting their hypothesis, recentdata indicate that vasopressin levels are increased in potassium-depleted rats, and that the elevated levels appear to be adefense mechanism for protecting systemic hemodynamics[120]. On the other hand, this phenomenon has been describedmainly in the elderly population, so I always have wonderedwhether mechanisms other than hypokalemia might be at play.For example, might a subset of the elderly population havebaroreceptor mechanisms that can be activated with lesserdegrees of volume contraction than in younger individuals, sothat we don't recognize the degree of volume contraction asbeing a stimulus? I think it would be quite interesting to studythis phenomenon more carefully in terms of what we nowunderstand about the nonosmotic release of vasopressin.

DR. BRIAN DUFFY (Attending Nephrologist, Michael ReeseHospital): What is your approach to the unusual patient whobecomes severely hypokalemic during administration of smalldoses of diuretics?

DR. TANNEN: If a hypertensive patient develops diuretic-induced hypokalemia to a level less than 3.0 mEq/liter, I believeit is appropriate to evaluate for primary aldosteronism. Anotherconsideration would be renal artery stenosis; patients with thisdisorder can have significant degrees of secondary hyperaldo-steronism and develop hypokalemia on that basis, as can thosewith other forms of mineralocorticoid or glucocorticoid excess.Of course, other causes of potassium depletion also should bekept in mind, including vomiting and laxative abuse. In effect,the patient with severe hypokalemia needs to be completelyworked up. Of course this is not the case with the averagepatient taking diuretics, who is mildly hypokalemic; in thissituation, there is a very clear explanation.

DR. COHEN: If you proceed with that algorithm, you willwork up a lot of people to obtain a very low yield of patientswith primary aldosteronism. By definition, you are talkingabout patients who were normokalemic prior to the introduc-tion of diuretics, because presumably you would have alreadypursued the possibility of primary aldosteronism had they beenspontaneously hypokalemic. Thus we are considering maybe asmany as 5% of patients treated with diuretics who will becomehypokalemic to this degree. I believe the number of patients inthat subset who have primary aldosteronism will be extremelysmall.

DR. TANNEN: I agree that the yield might be quite low. Butthe workup for primary aldosteronism is not difficult. Screeningby the measurement of plasma renin is easy, and the finding ofnormal renin levels should eliminate the vast majority ofpatients from requiring further evaluation.

Diuretic-induced hypokalemia 997

DR. SERAFINO GARELLA (Renal Division, Michael ReeseHospital): You mentioned that the level of dietary sodiumintake least likely to be associated with major potassium deficitsin patients on diuretic therapy is approximatley 100 mEq/day.Would you speculate on the mechanisms responsible for thelinkage between a higher level of salt intake and the increasedlikelihood of potassium wasting?

DR. TANNEN: In the study by Ram and coworkers, potassiumdepletion was significantly greater when diuretics were takenfor 4 weeks with a dietary sodium content of approximately 200mEq/day as compared with 70 mEq/day [23]. Aldosteronelevels were higher in the group on the lower sodium intake.Thus, one could speculate that the higher sodium intake result-ed in greater fluid and sodium delivery to the distal nephron,and thereby resulted in heightened urinary losses of potassium;but the actual underlying mechanism was not subjected torigorous investigation. It is of interest that Wilcox and col-leagues recently reported no evidence of negative potassiumbalance in normal volunteers ingesting a sodium intake of 270mEq/day during 3 days of furosemide treatment [121]. Obvious-ly the two sets of studies were carried out in different fashions,but the possibility does exist that the kaliuretic response mightdiffer between normotensive and hypertensive individuals.

DR. KAI LAU (Renal Division, Michael Reese Hospital):There are at least two reasons why potassium balance studies inhumans might underestimate the true magnitude of the potassi-um deficits induced by diuretics. (1) Many hypertensive pa-tients ingesting a low-salt diet are also taking salt substitutes,which are generally potassium-rich and which we often don'tknow about because of the absence of a careful history. (2) Weare beginning to uncover increasing numbers of patients whotake variable amounts of prostaglandin inhibitors without aprescription. If this piece of history is not specifically sought,one can miss intervals of variable lengths over which renalpotassium retention can have occurred. Wouldn't these twofactors obscure the true magnitude as well as the incidence ofpotassium deficits?

DR. TANNEN: I would agree with you that either of thesephenomena could influence the degree of potassium depletionand of hypokalemia in patients taking diuretics. I suspect,however, that these two possibilities have not had a meaningfulimpact on the incidence and magnitude of potassium depletionreported in the literature. First, the data on the incidence anddegree of hypokalemia include an enormous number of pa-tients, the studies have been reported over several decades, andthe data are all reasonably consistent. Second, in the studiesthat determined actual potassium deficits using either ex-changeable or total body potassium measurements, the degreeof hypokalemia was comparable to that reported in the volumi-nous data that measured only blood levels. Thus the potassiumdeficit appears to have been measured on a representativesample.

DR. COHEN: Dr. Lau's question brings up another point. Youcommented that Lawson's data published in the '70s andcollected retrospectively from several hospitals [1151 mighthave overestimated the current risk of hyperkalemia, even inthe hospitalized patient, because we now use potassium supple-ments and potassium-sparing agents more cautiously and withmore understanding of the hazards. But it's also true thatpatients are now being exposed to several new antikaliuretic

drugs that potentially compound the risk of hyperkalemia; I'mreferring to the nonsteroidal antiinflammatory agents, the betablockers, captopril, and no doubt others.

DR. TANNEN: That's a good point. Clearly if one is going touse potassium-sparing drugs or potassium supplementation, allthe drugs that can predispose towards development of signifi-cant hyperkalemia must be kept in mind. Also, great careshould be exercised in treating insulin-dependent diabetics withany potassium maintenance regimen. Finally, the elderly andindividuals with decreased GFRs who are also at increased riskfor hyperkalemia need to be individualized so that we canassess the need for and the risks versus benefits of potassiumtherapy.

DR. JAMES BOURDEAU (Renal Division, Michael Reese Hos-pital): In experimental animals not given diuretics, chronicadministration of potassium chloride results in a phenomenonof potassium adaptation, whereby acute potassium loads areexcreted by the kidneys more rapidly than normal. Do patientswho are taking diuretics and who are being given supplementsof potassium chloride also develop potassium adaptation? Thatis, are they able to excrete an acute potassium load morerapidly than do normal individuals? If so, and if their potassiumchloride supplementation were stopped, would they be atincreased risk for developing hypokalemia?

DR. TANNEN: I would assume not. I suspect that the induc-tion of adaptation requires at least transitory hyperkalemia.However, I am unaware of any studies that have directly testedthe question you raise.

DR. BUSHINSKY: As you noted, a reasonable correlationexists between the serum potassium concentration and potassi-um balance over a wide range of potassium depletion. Yet youalso noted that the fall in serum potassium observed with threedifferent diuretic agents correlated poorly with the change inpotassium balance. Could you comment on this apparentcontradiction?

DR. TANNEN: I don't have a good answer to your question,but I would caution that the data summarized in Table 3represent a composite of multiple studies using different mea-surement techniques, and with some variation in sodium intake.In the furosemide group, which has the lowest patient number,elimination of the 12 patients who were ingesting a sodiumintake of 196 mEq/day [23] reduces the mean potassium deficitto 132 mEq. This would square better with the lower decrementin plasma potassium, but I think there are insufficient total bodypotassium data to make a firm comparison of the furosemideand thiazide data.

DR. JEROME P. KASSIRER: Dr. Tannen, whether or not onebelieves that it is appropriate to prevent hypokalemia in diuret-ic-treated patients depends in part on one's strength of convic-tion about the correlation between hypokalemia and life-threat-ening rhythm disturbances. John Harrington and I have asomewhat different view from yours about the validity of thiscorrelation. A recent study by Nordrehaug, which you did notmention, is a case in point. Nordrehaug concluded, in a study ofpatients with acute myocardial infarction complicated by seri-ous rhythm disturbances, that hypokalemia increased thechance of a serious arrhythmia [1221. Careful review of hisstudy, however, discloses that other factors, including digitalisadministration and preexisting coronary artery disease, couldhave explained the higher frequency of rhythm disturbances in

998 Nephrology Forum

the hypokalemic group. This comment raises questions aboutthe validity of the experimental design, a problem we haveaddressed before [5]. Even if one accepts the Nordrehaugreport as valid, however, we are concerned that the fraction ofpatients developing ventricular fibrillation was the same orhigher in patients with hyperkalemia as in two groups of hispatients with hypokalemia. This finding speaks to our concernthat the risks of hyperkalemia need to be balanced against therisks of hypokalemia. Can you address these issues?

DR. TANNEN: There is one very recent report by Nordrehaugand coworkers that I did not mention [1231. In this study, 60patients with an acute myocardial infarction and no treatmentwith cardioactive drugs were evaluated prospectively. Thisstudy also delineated a significant negative correlation betweenserum potassium level and ventricular tachycardia, supportingother retrospective analyses suggesting that digitalis does notexplain this relationship [56, 57, 81, 84].

You are correct in pointing out that in the study by Nordre-haug, as well as in several other studies, the incidence ofventricular tachycardialfibrillation was increased by hyperkale-mia as well as by hypokalemia [56, 57, 83, 84]. Dyckner andcoworkers suggest that the hyperkalemia is secondary to hy-poxia and acidosis and that the increase in ventricular arrhyth-mias may be related to the more serious condition of thesepatients, but the authors do not provide any data on this point[831. Certainly the cardiac risks of severe hyperkalemia areundisputable. In my view, the goal should be maintenance ofnormal potassium balance in patients at risk for myocardialinfarction in order to reduce the risk of sudden death beforethey arrive at the hospital.

Reprint requests to Dr. R. Tannen, Director, Division of Nephrology,The University of Michigan Medical Center, D3238 Medical Profes-sional Building, Box 19, Ann Arbor, Michigan 48109—0010, USA

References

1. K.ssrng JP, HARRINGTON JT: Diuretics and potassium metabo-lism. A reassessment of the need, effectiveness and safety ofpotassium therapy. Kidney mt 11:505—515, 1977

2. KOSMAN ME: Management of potassium problems during long-term diuretic therapy. JAMA 230:743—748, 1974

3. MORGAN TO: Potassium replacement: Supplements or potassium-sparing diuretics? Drugs 18:218—225, 1979

4. LAWSON AAH: Potassium replacement: When is it necessary?Drugs 21:354—361, 1981

5. HARRINGTON JT, IsNER JM, KASSIRER JP: Our national obsessionwith potassium. Am J Med 73:155—159, 1982

6. LAWSON DH: The case for a more conservative approach toprophylaxis and therapy, in Controversies in Nephrology andHypertension, edited by NARINS RG, New York, Churchill-Livingstone, 1984, pp 331—344

7. HOLLAND OB: The case for routinely normalizing serum potassi-um, in Controversies in Nephrology and Hypertension, edited byNARINS RG, New York, Churchill-Livingstone, 1984, pp 345—358

8. LEEMHUIS MP, STRUYVENBERG A: Significance of hypokalemiadue to diuretics. Neth J Med 16:18—28, 1973

9. Goon DW, WRIGHT FS: Luminal influences on potassium secre-tion: sodium concentration and fluid flow rate. Am J Physiol236:F192-F205, 1979

10. STOKES JB: Potassium secretion by cortical collecting tubule:relation to sodium absorption, luminal sodium concentration, andtransepithelial voltage. Am J Physiol 241:F395-F402, 1981

11. TANNEN RL: Potassium metabolism, in Current Nephrology (vol6), edited by G0NIcK HC, New York, Wiley, 1983, pp 151—186

12. WRIGHT FS, GWBIscH G: Renal potassium transport: contribu-

tions of individual nephron segments and population. Am JPhysiol 235:F515-F527, 1978

13. SCHWARTZ WB, VAN YPERSELE DE STRIHOU C, KASSIRER JP:Role ofanions in metabolic alkalosis and potassium deficiency. NEngi J Med 279:630—639, 1968

14. SELDIN DW, RECTOR FC JR: The generation and maintenance ofmetabolic alkalosis. Kidney mt 1:306—321, 1972

15. KASSIRER JP, APPLETON FM, CHAZAN JA, SCHWARTZ WB:Aldosterone in metabolic alkalosis. J Clin Invest 46:1558—1571,1967

16. LUKE RG: Effect of adrenalectomy on the renal response tochloride depletion in the rat. J Clin Invest 54:1329—1336, 1974

17. GABOW PA, KAEHNY WD: Aldosterone in diuretic-induced meta-bolic alkalosis in dogs. Clin Res 30: 107A, 1982

18. EVANS BM, HUGHES-JONES NC, MILNE MD, STEINER 5: Electro-lyte excretion during experimental potassium depletion in man.Clin Sci 13:305—3 16, 1954

19. TANNEN RL, GERRITS L: Response of the renal K conservingmechanism to kaliuretic stimuli: Evidence for a direct kaliureticeffect of furosemide (abstract). Clin Res 31 :752A, 1983

20. VELAZQUEZ H, WRIGHT FS, GOOD DW: Luminal influences onpotassium secretion: chloride replacement with sulfate. Am JPhysiol 242:F46-F55, 1982

21. MORGAN DB, DAVIDSON C: Hypokalemia and diuretics: an analy-sis of publications. Br Med J 280:905—908, 1980

22. WILKINSON PR, HESP R, I5sLER H, RAFTERY EB: Total body andserum potassium during prolonged thiazide therapy for essentialhypertension. Lancet 1:759—762, 1975

23. RAM CVS, GARRETT BN, KAPLAN NM:Moderate sodium restric-tion and various diuretics in the treatment of hypertension. ArchIntern Med 141:1015—1019, 1981

24. LANDMANN-SUTER R, STRUYVENBERG A: Initial potassium lossand hypokaleamia during chlorthalidone administration in patientswith essential hypertension: the influence of dietary sodiumrestriction, Eur J Clin Invest 8:155—164, 1978

25. HOLLANDER W, CHOBANIAN AV, WILKINS RW: The role ofdiuretics in the management of hypertension. Ann NY Acad Sci88:975—989, 1960

26. TALSO PJ, CARBALLO AJ: Effects of benzothiadiazines on serumand total body electrolytes. Ann NY Acad Sci 88:822—840, 1960

27. WINER BM: The antihypertensive actions of benzothiadiazines.Circulation 23:211—218, 1961

28. GIFFORD RW JR, MATTOX VR, ORvI5 AL, S0NE5 DA, ROSEVEARJW: Effect of thiazide diuretics on plasma volume, body electro-lytes, and excretion of aldosterone in hypertension. Circulation24:1197—1205, 1961

29. CHRISTENSEN M, GALSKOV A: Udbytbart Kalium hos normalesamt pateinter med hypertensio arteriolis. Nordisk Medicin24:1081—1085, 1967

30. ANDERSON J, GODFREY BE, HILL DM, MUNRO-FAURE AD,SHELDON J: A comparison of the effects of hydrochlorothiazideand of frusemide in the treatment of hypertensive patients. Q JMed 40:541—560, 1971

31. GRAYBIEL AL, SODE J: Diuretics, potassium depletion and carbo-hydrate intolerance (letter). Lancet 2:265, 1971

32. EDMOND5 CJ, JASONI B: Total-body potassium in hypertensivepatients duirng prolonged diuretic therapy. Lancet 2:8—12, 1972

33. HOLLIFIELD JW, SLATON PE: Thiazide diuretics, hypokalemiaand cardiac arrhythmias. Acta Med Scand (suppl) 647:67—73, 1981

34. STEINESS E: Thiazide-induced potassium loss not prevented bybeta blockade. Clin Pharmacol Ther 15:788—791, 1984

35. OH MS, CARROLL HJ: The renin-aldosterone system and thiazide-induced depletion of total body potassium in essential hyperten-sion. Nephron 21:269—276, 1978

36. MARONDE RF, MILGROM M, DICKEY JM: Potassium loss withthiazide therapy. Am Heart J 78:16—21, 1969

37. LEMIEUX G, BEAUCHEMIN M, VINAY P, GouGoux A: Hypokale-mia during the treatment of arterial hypertension with diuretics.CMA J 122:905—907, 1980

38. HEALY JJ, MCKENNA TJ, CANNING B, BRIEN TO, DUFFY GJ,MULDOWNEY FP: Body composition changes in hypertensivesubjects on long-term oral diuretic therapy. Br Med J 1:716-719,1970

Diuretic-induced hypokalemia 999

39. DARGIE HJ, BODDY K, KENNEDY AC, KING PC, READ PR,WARD DM: Total body potassium in long-term frusemide therapy:is potassium supplementation necessary? Br Med J 4:316—319,1974

40. LEEMHUIS MP, VAN DAMME KJ, STRUYVENBERG A: Effects ofchlorthalidone on serum and total body potassium in hypertensivepatients. Acta Med Scand 200:37—45, 1976

41. DYCICNER T, WaSTER P0: Intracellular magnesium loss afterdiuretic administration. Drugs (suppl 1) 28:161—166, 1984

42. VILLAMIL MF, YEYATI N, ENCRO MA, RUBIANE C, TAQUINI AC:Effect of long term treatment with hydrochlorothiazide on waterand electrolytes of muscle in hypertensive subjects. Am Heart J65:294, 1963

43. BERGSTROM J, HULTMAN E: The effect of thiazides, chlorthali-done and furosemide on muscle electrolytes and muscle glycogenin normal subjects. Acta Med Scand 180:363—376, 1966

44. MATERSON BJ, CARALIS PV: Risk of cardiac arrhythmias inrelation to potassium imbalance. J Cardiovasc Pharmacol 6:S493-S497, 1984

45. TONER JM, RAMSAY IE: Thiazide-induced hypokalaemia; preva-lence higher in women. Br J Gun Pharmacol 18:449—452, 1984

46. DAVIDSON C, BURKINSHAW L, MCLACHLAN MSF, MORGAN DB:Effect of long-term diuretic treatment on body-potassium in heart-disease. Lancet 2:1044, 1976

47. STERNS RH, Cox M, FEIG PU, SINGER I: Internal potassiumbalance and the control of the plasma potassium concentration.Medicine 60:339—354, 1981

48. HALL RJC, CAMERON IR: Effect of chronic dietary potassiumdepletion and repletion on cardiac and skeletal muscle structure,electrolytes and pH in the rabbit. Gun Sci 60:441—449, 1981

49. NORGAARD A, KJELDSEN K, CLAUSEN T: Potassium depletiondecreases the number of 3H-ouabain binding sites and the activeNa-K transport in skeletal muscle. Nature 193:739—741, 1981

50. TANNEN RL: The effect of uncomplicated potassium depletion onurine acidification. J Clin Invest 49:813—827, 1970

51. RUBINI ME: Water excretion in potassium-deficient man. J GunInvest 40:2215—2223, 1961

52. DANIELL HW: Arrhythmia in hypokalemia (letter). N Engi J Med284:1385, 1971

53. THOMPSON RG, COBB LA: Hypokalemia after resuscitation fromout-of-hospital ventricular fibrillation. JAMA 248:2860—2863, 1982

54. DUKE M: Thiazide-induced hypokalemia. Association with acutemyocardial infarction and ventricular fibrillation. JAMA 239:43—45, 1978

55. HOLLAND OB, NIxoN JV, KUHNERT L: Diuretic-induced ventric-ular ectopic activity. Am J Med 70:762, 1981

56. NORDREHAUG JE: Malignant arrhythmias in relation to serumpotassium values in patients with an acute myocardial infarction.Acta Med Scand S647:lOl—l07, 1981

57. SOLOMON Ri, COLE AG: Importance of potassium in patients withacute myocardial infarction. Acta Med Scand S647:87—93, 1981

58. BELLET 5, NADLER CS, GAZES PC, LANNING M: The effect ofvomiting due to intestinal obstruction on the serum potassium.Gastroenterology 12:49, 1949

59. ELEIL LP, PEARSON OH, RAWSON RW: Postoperative potassiumdeficit and metabolic alkalosis. N Engi J Med 243:478, 1950

60. WEAVER WF, BURCHELL HB: Serum potassium and the electro-cardiogram in hyopkalemia. Circulation 21:505, 1960

61. KUNIN AS, SURAWICZ B, SIMS EAH: Decrease in serum potassi-um concentrations and appearance of cardiac arrhythmias duringinfusion of potassium with glucose in potassium-depleted patients.N Engi J Med 266:228, 1962

62. DAVIDSON S, SURAWICZ B: Ectopic beats and atrioventricularconduction disturbances in patients with hypopotassemia. ArchIntern Med 120:280, 1967

63. TAMURA K, TAMURA T, YOSHIDA S, INUI M, FUKUHARA N:Transient recurrent ventricular fibrillation due to hypopotassemiawith special note on the U wave. Jpn Heart J 6:652, 1967

64. HULTING J: In-hospital ventricular fibrillation and its relation toserum potassium. Acta Med Scand S647: 109—1 16, 1981

65. SuRAwlcz B, LEPESCHKIN E: The electrocardiographic pattern ofhypopotassemia with and without hypocalcemia. Circulation8:801, 1953

66. BETTINGER JC, SURAWICZ B, BRYFOGLE JW, ANDERSON BN JR.BaLLET 5: The effect of intravenous administration of potassiumchloride on ectopic rhythms, ectopic beats and disturbances in A-V conduction. Am J Med 21:521, 1956

67. PEDERSON DH, ZIPES DP, FOSTER PR, TROUP PJ: Ventriculartachycardia and ventricular fibrillation in a young population.Circulation 60:988, 1979

68. PAPADEMETRIOU V, FLETCHER R, KHATRI IM, FREIS ED: Diuret-ic-induced hypokalemia in uncomplicated systemic hypertension:Effect of plasma potassium correction on cardiac arrhythmias. AmJ Cardiol 52:1017—1022, 1983

69. PAPADEMETRIOU V, PRICE MB, NOTARGIACOMO A, FLETCHERRD, FREIS ED: Effect of thiazide on ventricular arrhythmias inpatients with uncomplicated systemic hypertension. Clin Res32:337A, 1984

70. HOLLIFIELD JW: Potassium and magnesium abnormalities: Di-uretics and arrhythmias in hypertension. Am J Med (Electrolyte &Cardiovascular Disease Symposium) 77:28—32, 1984

71. WHELTON PK, BRENNAN P. MIALL WE, GREENBERG G, SUBRA-MANIAN B, RAFTERY EB: Thiazide induced cardiac arrhythmias.Clin Res 30:341A, 1982

72. MEDICAL RESEARCH COUNCIL WORKING PARTY ON MILD TOMODERATE HYPERTENSION: Ventricular extrasystoles during thia-zide treatment: substudy of MRC mild hypertension trial. Br MedJ287:l249, 1983

73. WHELTON PK: Diuretics and arrhythmias in the Medical ResearchCouncil Trial. Drugs (suppl 1) 28:54—65, 1984

74. CARALIS PV, MATERSON BJ, PEREZ-STABLE E: Potassium anddiuretic-induced ventricular arrhythmias in ambulatory hyperten-sive patients. Mm Electrolyte Metab 10:148—154, 1984

75. MADIAS JE, MADIAS NE, GAVRAS HP: Nonarrhythmogenicity ofdiuretic-induced hypokalemia. Arch Intern Med 144:2171—2176,1984

76. LIEF PD, BELIZON I, MATOS J, BANK N: Diuretic-induced hypo-kalemia does not cause ventricular ectopy in uncomplicatedessential hypertension (abstract). Proc Am Soc Nephrol 16:A66,1983

77. HOLLAND OB: Diuretic-induced hypokalemia and ventriculararrhythmias. Drugs (suppl) 28:86—92, 1984

78. KENNEDY HL, WHITLOCK JA, SPRAGUE MK, KENNEDY Li,BUCKINGHAM TA, GOLDBERG RJ: Long-term follow-up of asymp-tomatic healthy subjects with frequent and complex ventricularectopy. NEnglJMed3l2:l93—197, 1985

79. LOWN B, CALVERT AF, ARMINGTON R, RYAN M: Monitoring forserious arrhythmias and high risk of sudden death. Circulation(suppl III) 51, 52:111-189—111-198, 1975

80. RUBERMAN W, WEINBLATT E, GOLDBERG JD, FRANK CW,CHAUDARY BS, SHAPIRO S: Ventricular premature complexes andsudden death after myocardial infarction. Circulation 64:297—305,1981

81. SOLOMON Ri: Ventricular arrhythmias in patients with myocardialinfarction and ischaemia. Relationships to serum potassium andmagnesium. Drugs (suppl 1) 28:66—76, 1984

82. DYCKNER T, HELMERS C, LUNDMAN T, WESTER P0: Initial serumpotassium level in relation to early complications and prognosis inpatients with acute myocardial infarction. Ada Med Scand197:207—210, 1975

83. DYCKNER T, HELMERS C, WESTER P0: Initial serum potassiumlevel, early arrhythmias and previous diuretic therapy in acutemyocardial infarction. mt J Cardiol 2:146—148, 1982

84. REUBEN SR, KUAN P: The role of potassium in arrhythmias aftermyocardial infarction. mt Med Suppl 5:9—11, 1983

85. JOHANSSON BW, DZIAMSKI R: Malignant arrhythmias in acutemyocardial infarction. Relationship to serum potassium and effectof selective and non-selective /3-blockade. Drugs (suppl I) 28:77—85, 1984

86. BOYD JC, BRUNS DE, DIMARCO JP, SUGG NK, WILLS MR:Relationship of potassium and magnesium concentrations in se-rum to cardiac arrhythmias. Clin Chem 30:754—757, 1984

87. REUBEN SR. THOMAS RD: The relationship between serum potas-sium and cardiac arrhythmias following cardiac infarction inpatients aged over 65 years. CurrMed Res Opin (suppl I) 7:79—82,1982

1000 Nephrology Forum

88. STRUTHERS AD, WHITESMITH R, REID JL: Prior thiazide diuretictreatment increases adrenaline-induced hypokalaemia. Lancet1:1358, 1983

89. MULTIPLE RIsK FACTOR INTERVENTION TRIAL RESEARCH GROUP:Multiple Risk Factor Intervention Trial. Risk factor changes andmortality results. JAMA 248:1465—1477, 1982

90. SHERWIN R: Sudden death in men with increased risk of myocar-dial infarction. The MRFIT Programme. Drugs 28 (suppl 1) 28:46—53, 1984

91. TANNEN RL: Potassium and blood pressure control. Ann InternMed 98:773—780, 1983

92. KAPLAN NM, CARNEGIE A, RASKIN P, HELLER JA, SIMMONS M:Potassium supplementation in hypertensive patients with diuretic-induced hypokalemia. N Engl J Med 312:746—749, 1985

93. PEREZ-STABLE E, CARALIS PV: Thiazide-induced disturbances incarbohydrate, lipid, and potassium metabolism. Am Heart J106:245—251, 1983

94. GRUNFELD C, CHAPPELL DA: Hypokalemia and diabetes mdli-tus. Am J Med 75:553—554, 1983

95, WELLER JM, MALVIN RL: Effects and side-effects of thiazidedrugs. Med Gun North Am 53:1321—1330, 1969

96. ROWE JW, TOBIN JD, ROSA RM, ANDRES R: Effect of experimen-tal potassium deficiency on glucose and insulin metabolism.Metabolism 29:498—502, 1980

97. GORDEN P: Glucose intolerance with hypokalemia. Failure ofshort-term potassium depletion in normal subjects to reproducethe glucose and insulin abnormalities of clinical hypokalemia.Diabetes 22:544—551, 1973

98. HELDERMAN JH, ELAHI D, ANDERSEN DK, RAIZES OS, TOBINJD, SHOCKEN D, ANDRES R: Prevention of the glucose intoleranceof thiazide diuretics by maintenance of body potassium. Diabetes32:106—111, 1983

99. RAPOPORT MI, HURD HF: Thiazide-induced glucose intolerancetreated with potassium. Arch In fern Med 113:405—408, 1964

100. AMERY A, BERTHAUX P, BULPITT C, DERUYTTERE M, DESCHAEPDRYVER A, DOLLERY C, FAGARD R, FORETTE F, HELLE-MANS J, LUND-JOHANSEN P. MUTSERS A, TUOMILEHTO J: Glu-cose intolerance during diuretic therapy. Lancet 1:681—683, 1978

101. THOMAS JP, THOMSON WH: Comparison of thiazides and amilor-ide in treatment of moderate hypertension. Br Med J 286:2015—2018, 1983

102. NILSSON G: Symptomatic diabetes mellitus cured by potassiumand withdrawal of polythiazide in a hypokalemic hypertensivewoman. Acta Med Scand 208:129—131, 1980

103. MURPHY MB, LEWIS PJ, KOHNER E, SCHUMER 13, DOLLERY CT:Glucose intolerance in hypertensive patients treated with diuret-ics; a fourteen-year follow-up. Lancet 2:1293—1295, 1982

104. GORDEN P, SHERMAN BM, SIMoPouLos AP: Glucose intolerancewith hypokalemia: An increased proportion of circulating proinsu-lin-like component. J Gun Endocrinol Metab 34:2235—2240, 1972

105. SAGILD U, ANDERSON V, ANDREASEN PB: Glucose tolerance andinsulin responsiveness in experimental potassium depletion. ActaMed Scand 169:243—251, 1961

106. GRIMM RH JR, LEON AS, HUNNINGHAKE DB, LENZ K, HANNANP, BLACKBURN H: Effects of thiazide diuretics on plasma lipidsand lipoproteins in midly hypertensive patients. Ann Intern Med94:7—11, 1981

107. GOLDMAN AL, STEELE BW, SCHNAPER HE, FITZ A, FROLICH E,PERRY HM: Serum lipoprotein levels during chlorthalidone thera-py. JAMA 22:1691—1695, 1980

108. BOEHRINGER K, WEIDMANN P, MORDASINI R, SCHIFFL H, BACH-MANN C, RISEN W: Menopause-dependent plasma lipoproteinalterations in diuretic-treated women. Ann Intern Med 97:206—209, 1982

109. SCHOENFELD MR, GOLDBERGER E: Hypercholesterolemia in-duced by thiazides: A pilot study. Curr Therap Res 6:180—184,1964

110. BENGTSSON C, LENNARTSSON J, LINDQUIST 0, LINDSTEDT G,LUNDBERG PA, NOPPA H, SIGURDSSON JA, TIBBLIN E: Onmetabolic effects of diuretics and (.3-blockers. Acta Med Scand212:57—64, 1982

111. MATERSON BJ, FREIS ED, REDA D: Dose dependent metabolicpertubations by hypochlorothiazide and propranolol, in Diuretics,Chemistry, Pharmacology and Clinical Applications, edited byPUSCHETT JB, New York, Elsevier, 1984, pp 25 1—253

112. HELGELAND A, HJERMANN 1, HOLME I, LEREN P: Serum triglyc-erides and serum uric acid in untreated and thiazide-treatedpatients with mild hypertension. Am J Med 64:34—38, 1978

113. AMES RP, HILL P: Elevation of serum lipid levels during diuretictherapy of hypertension. Am J Med 61:748—757, 1976

114. SCHWARTZ AB, SWARTZ CD: Dosage of potassium chloride elixirto correct thiazide-induced hypokalemia. JAMA 230:702—704,1974

115. LAWSON DH: Adverse reactions to potassium chloride. Q J Med43:433—440, 1974

116. GREENBLATT DJ, KOCH-WESER J: Adverse reactions to spirono-lactone. JAMA 225:40—43, 1973

117. TANNEN RL: Potassium disorders, in Fluids and Electrolytes,edited by K0KK0 JP, TANNEN RL, Philadelphia, Saunders, 1985

118. FOLLIS RH, ORENT-KEILES E, MCCOLLUM EV: The productionof cardiac and renal lesions in rats by a diet extremely deficient inpotassium. Am J Pathol 18:29—40, 1942

119. FICHMAN MP, VORHERR H, KLEEMAN CR, TELFER N: Diuretic-induced hyponatremia. Ann Intern Med 75:853—863, 1971

120. PALLER MS, LINAS SL: Role of vasopressin in support of bloodpressure in potassium deficient rats. Kidney mt 24:342—347, 1983

121. WiLcox CS, MITCH WE, KELLY RA, FRIEDMAN EA, SOUNEYPF, RAYMENT CM, MEYER TW, SKORECKI KL: Factors affectingpotassium balance during furosemide administration. Clin Sci67: 195—203, 1984

122. NORDREHAUG JE, VON DER LIPPE G: Hypokalemia and ventricu-lar fibrillation in acute myocardial infarction. Br Heart J 50:525—529, 1983

123. NORDREHAUG JE, J0HANNESSEN K-A, VON DER LIPPE G: Serumpotassium concentration as a risk factor of ventricular arrhythmiasearly in acute myocardial infarction. Circulation 71:645—649, 1985