Magnesium and Potassium Balance

Magnesium repletion and its effect on potassium homeostasis in critically ill adults: Results of a double-blind, randomized, controlled trial

Objectives: The aims of this study were to evaluate the safety and efficacy of magnesium replacement therapy and to determine its effect on potassium retention in hypokalemic, critically ill patients.

Design: A prospective, double-blind, randomized, placebocontrolled trial.

Setting: A surgical intensive care unit (ICU).

Patients: A total of 32 adult surgical ICU patients were admitted to the study on the basis of documented hypokalemia (potassium of <3.5 mmol/L) within the 24-hr period before entering the study. Patients were randomized to receive either placebo (n = 15) or magnesium sulfate (n = 17). One patient from each group was excluded from the study due to failure to complete the full series of doses.

Interventions: Patients received a “test dose” of either magnesium sulfate (2 g, 8 mmol) or placebo (5% dextrose in water) infused over 30 mins every 6 hrs for eight doses. The next scheduled test dose was held if hypermagnesemia (magnesium of >2.8 mg/dL [>1.15 mmol/L]) was documented at any time during the study. Routine replacements of potassium and magnesium continued during the duration of the study, when clinically indicated, for serum potassium concentrations of 3.5 mmol/L or serum magnesium concentrations of <1.8 mg/dL (<0.74 mmol/L).

Measurements and Main Results: Age, weight, and Acute Physiology and Chronic Health Evaluation II scores were recorded on entry into the study. Just before administration of each test dose, blood was drawn for magnesium and potassium, bicarbonate, pH, and glucose determinations, and an aliquot of the preceding 6 hrs urine collection was sent for magnesium and potassium determinations. Serum calcium, phosphate, urea nitrogen, and creatinine concentrations were measured daily. The amounts of magnesium and potassium administered via parenteral nutrition, tube feeding, and replacement infusions were calculated for each 6-hr interval. The amounts of magnesium and potassium excreted in the urine were similarly assessed.

The groups showed no differences with regard to age, weight, Acute Physiology and Chronic Health Evaluation II scores, or initial serum magnesium concentration. Initial potassium, bicarbonate, pH, calcium, phosphate, glucose, blood urea nitrogen, and creatinine values were not different between groups. Patients receiving magnesium sulfate showed a statistically significant increase in serum magnesium concentration at 6 hrs when compared with placebo, as well as with itself at time 0 (p < .0001), a difference maintained throughout the study. Compared with the placebo group, the total amount of elemental magnesium administered was significantly greater in the treatment group (1603 ? 124 vs. 752 ? 215 mg [65.7 ? 5.8 vs. 30.8 ? 8.8 mmol], p < .0001), as was urine magnesium excretion (1000 ? 156 vs. 541 ? 68 mg [41.0 ? 6.4 vs. 22.2 ? 2.8 mmol] p < .0001). However, the net magnesium balance (total magnesium in - total urine magnesium) was signficantly more positive in the treatment group (612 ? 180 vs. 216 ? 217 mg [25.1 ? 7.4 vs. 8.9 ? 8.9 mmol], p < .005). The treatment and control groups had the same serum potassium concentrations and did not receive different amounts of potassium (245 ? 39 vs. 344 ? 45 mmol, respectively, p = .06), although the treatment group required less potassium replacement/6 hrs by 30 hrs compared with itself at time 0 (p < .05). Despite the same serum potassium values, the net potassium balance for 48 hrs was positive in the treatment group (+72 ? 32 mmol) and negative in the control group (-74 ? 95 mmol, p < .05). There were no complications associated with the magnesium sulfate administration.

Conclusions: Magnesium sulfate administered according to the above regimen safely and significantly increases the circulating magnesium concentration. Despite greater urine magnesium losses in the treatment group, this group exhibited significantly better magnesium retention. In addition, within 30 hrs of entry into the study, the treatment group exhibited a net positive and statistically significant (p < .05) improvement in potassium balance compared with the control group. (Crit Care Med 1996; 24:38-45)

Hypomagnesemia is a frequent finding in hospitalized patients that occurs in ~65% of medical intensive care unit (ICU) patients with normal serum creatinine concentrations. Often, serum magnesium does not precisely reflect total body stores of magnesium. Although a low serum magnesium concentration predicts the presence of intracellular depletion, a normal serum magnesium concentration can exist in the face of a clinically important intracellular magnesium deficiency. In addition, hypomagnesemia is often found concurrently with hypokalemia and hypochloremia. Further, magnesium repletion has been shown to facilitate correction of potassium and calcium eficiencies.

The human body contains ~0.33 g/kg (~1.32 mmol/kg) of magnesium. Only 1% of the body’s magnesium is found in the extracellular fluid compartment, approximately two thirds being in the bone, and the remaining third is found in cardiac and skeletal muscles. The usual daily intake provides 200 to 350 mg of magnesium/day, 30% to 50% (~100 mg) of which is absorbed through the intestine by both passive and active transport mechanisms. The kidney is active in magnesium homeostasis, filtering ~2.5 g of magnesium daily. Only 5% of this amount is actually excreted (~100 mg/day), and the remainder is reabsorbed primarily in the ascending loop of Henle, leading to a balanced homeostasis. In the face of magnesium deficiency, reabsorption can be nearly complete, although this process can be impeded by catecholamines, diuretics, aminoglycosides, and many other interventions that are administered to critically ill patients. In patients receiving total parenteral nutrition, 1 g of magnesium salt (100 mg of elemental magnesium = 4 mmol) per day has been recommended to avoid deficiency. However, the acuity of the patient population was not taken into account.

Magnesium is a required cofactor for most adenosinetriphosphatases (ATPases), since it is the ATP-magnesium++ complex that is bound and hydrolyzed by enzymes. Consequently, severe magnesium deficiency can have important detrimental effects on cellular function by decreasing the ATPase activity at the sodium-potassium cellular pump, as well as limiting oxidative phosphorylation, fat and protein synthesis, and DNA and RNA metabolism. Deficiency of this cation can be associated with weakness, tremors, muscle fasciculations, hypertension, cardiac dysrhythmias, electrocardiographic changes, atherogenesis, and sudden death. Prolonged deficiency has been associated with evidence of skeletal cardiomyopathy. The exacerbation of bronchospasm, gentamicin-induced ototoxicity, cyclosporine-induced nephrotoxicity, and possibly neurotoxicity, agitation, psychosis, dementia, and the seizures of delirium tremens have all been linked to hypomagnesemia. Magnesium depletion has also been correlated with increased concentrations of aldosterone and thromboxane, and insulin resistance, all of which improve after magnesium infusion.

Hypokalemia is also a common finding in hospitalized patients.. Diuretics, amphotericin, the penicillins, aldosterone, hypomagnesemia, and hypercalcemia contribute to the renal loss of potassium. Gastrointestinal losses of potassium occur with nasogastric suctioning, vomiting, diarrhea, and enteric fistulas. Intracellular shifts also result from increased catecholamine concentrations, correction of acidosis, or from the administration of insulin to treat hyperglycemia. Respiratory alkalosis is frequently found in the critically ill population due to overzealous mechanical ventilation and/or agitation, further affecting the distribution of potassium. To further exacerbate the problem, reduced intake of potassium is also common in ill patients unless careful attention is given to nutrition and/or supplementation.

Hypokalemia in the face of coexisting hypomagnesemia may be refractory to treatment. While magnesium replacement infusions are usually benign, potassium replacement infusions carry a significantly higher risk of cardiac dysrhythmia, phlebitis, pain, and tissue necrosis in the event of extravasation. It was shown. that after potassium replacement infusions in hypokalemic patients with coincident low-normal to subnormal magnesium concentrations, urine potassium excretion increases and the serum potassium concentration trends back toward pretreatment concentrations within 1 hr of completion of the infusion.

The current study was undertaken to evaluate the efficacy and safety of generous administration of magnesium in critically ill patients aimed at establishing and maintaining serum magnesium concentrations at high-normal to mildly increased values. In addition, the intent of the study was to evaluate whether this more aggressive replacement regimen would improve potassium homeostasis when compared with patients who had magnesium replaced only after documentation of hypomagnesemia.


The study protocol was approved by the Human Investigation Committee. Informed consent was obtained.

Patient Selection

Surgical ICU patients noted to have a circulating potassium concentration of <3.5 mmol/L within the past 24 hrs, and who had indwelling arterial catheters and radiologically confirmed central venous catheters in place were considered for the study. Exclusion criteria included age <18 yrs, oliguric or anuric renal failure, presence of the need for dialysis therapy, expected discharge from the ICU in <48 hrs, and the presence of any neuromuscular disease (in a nonmechanically ventilated patient) that might increase the patient's sensitivity to the muscle-relaxant effects of hypermagnesemia.

Study Design

Patients were enrolled by the ICU attending physician or fellow in a prospective, randomized, double-blind, placebo-controlled study. Randomization by way of a random number generator was performed by a single hospital pharmacist, who also filled and dispensed the “test dose” syringes of either placebo or magnesium sulfate. The treatment group received 2 g of magnesium sulfate in 50 mL of 5% dextrose in water over 30 mins every 6 hrs for eight doses. The control group received an infusion of 50 mL of 5% dextrose in water over 30 mins, following the same schedule. The test dose was held for a serum magnesium concentration of >2.8 mg/dL (>1.15 mmol/L), measured as a part of clinical care, and the scheduled dosing administration resumed when the serum magnesium concentration had returned to the normal range (1.8 to 2.8 mg/dL [0.74 to 1.15 mmol/L]). Study laboratory values were unavailable to the investigators until after completion of the study. Both groups continued to receive the standard potassium replacement therapy (potassium chloride, 20 mmol in 100 mL of 5% dextrose in water over 1 hr) for a serum potassium concentration of <3.5 mmol/L, and magnesium sulfate (2 g in 50 mL of 5% dextrose in water over 30 mins; 2 g of magnesium sulfate = 200 mg of elemental magnesium) for serum magnesium concentration of <1.8 mg/dL (<0.74 mmol/L), respectively, documented on care-related laboratory studies.

Specimen Acquisition

Just before beginning the test dose infusion, specimens were collected for serum magnesium, potassium, pH, glucose, and bicarbonate determinations. A baseline urine specimen for magnesium and potassium measurement was obtained before the first infusion. Subsequently, before the initiation of each test dose infusion, an aliquot of urine from the previous 6 hrs accumulation was sent for magnesium and potassium measurements. All study specimens were collected and processed by the bedside nurse. Daily calcium, phosphate, creatinine concentrations, and blood urea nitrogen values were recorded. Serum and urine magnesium and potassium concentrations were measured by colorimetric and potentiometric assays, respectively, on an Ektochem 700 Clinical Analyzer (Eastman Kodak, Rochester, NY). Continuous electrocardiogram was monitored with the ectopy record activated (Marquette, Marion, OH). Total non-test dose milligrams of elemental magnesium and millimoles of potassium administered were calculated for each 6-hr period from the clinically indicated replacement infusions, intravenous solutions, parenteral nutrition, and tube feedings. All calculations and analyses used the total magnesium and potassium (test dose plus nontest dose) amounts administered during each 6-hr period. Administration of insulin, inotropes, diuretics, and kaliuretic drugs such as amphotericin B were also recorded.

Statistical Analysis

Quantitative data were compared, using repeated-measures analysis of variance and Student’s paired t-test. A p .05 was considered significant. Values are expressed as mean ? SEM, unless otherwise noted.


From September 1991 until July 1993, 32 patients were enrolled in the study, 17 in the treatment group and 15 in the placebo group. One patient in each group failed to complete the study due to sequentially missed doses and processing of laboratory specimens (treatment), and to triage out of the surgical ICU before completion of the study (placebo). Patient characteristics, including primary diagnoses, demographic data, and potentially confounding drugs administered to each group are noted in Table 1 . There were no differences between the groups with regard to age, weight, or Acute Physiology and Chronic Health Evaluation II scores.

Serum magnesium concentrations on entry into the study were the same in both groups (1.91 ? 0.06 vs. 2.09 ? 0.12 mg/dL [0.78 ? 0.02 vs. 0.86 ? 0.05 mmol/L]). After the first administered test dose, the serum magnesium concentration was significantly higher in the treatment group (p < .0001) compared with the control group as well as compared with itself at time 0. This relationship was maintained throughout the study (Fig. 1) . Of the possible eight scheduled test doses, the treatment group received 5.7 ? 0.4 doses, and the control group received 7.2 ? 0.4 “doses” (p < .01). The number of doses administered to the treatment group tended to decrease over the course of the study. At 6 hrs, only 12% of the treatment patients had the test dose withheld for a magnesium concentration of >2.8 mg/dL (>1.15 mmol/L), while at 36 and 42 hrs, 50% of the patients had the “dose” withheld (Fig. 2) .

The total elemental magnesium administered over 48 hrs (test dose,

TABLE 1 — Characteristics of patient groups

Treatment Group (n = 16) Control Group (n = 14)
Primary Diagnosis
Sepsis 7 7
Multiple trauma 2 2
s/p GI bleeding 2 2
Postoperative 2 0
Pancreatitis 3 3
Demographic Data
Age (yr) 60.5 ? 15.6 a 52.1 ? 17.7 a
Male/female 8/8 10/4
Weight (kg) 88 ? 32 a 88 ? 24 a
APACHE II 16.9 ? 3.5 a 17.0 ? 4.8 a
Confounding Drugs Administered
Diuretics 10 9
Insulin 7 7
Amphotericin 3 2
1-3 mug/kg/min 2 2
>3 mug/kg/min 1 1
Other vasoactive drugs 1 1
s/p, status post; GI, gastrointestinal; APACHE II, Acute Physiology and Chronic Health Evaluation II score.

a Values are expressed as mean ? SD.

replacement infusions, intravenous solution, parenteral nutrition plus tube feedings) was significantly greater in the treatment group than in the control group (1603 ? 124 vs. 752 ? 215 mg [65.7 ? 5.8 vs. 30.8 ? 8.8 mmol], p < .0001). This amount is equivalent to 9.6 ? 0.8 vs. 4.3 ? 1.1 mg/kg/day (range 5.6 to 16.7 vs. 0 to 11.9, p = .005). While the urinary magnesium excretion increased significantly in the treatment group compared with the control group (total excreted/48 hrs = 1000 ? 156 vs. 541 ? 68 mg [41.0 ? 6.4 vs. 22.2 ? 2.8 mmol], p < .0001) (Fig. 1) , the treatment group still retained significantly more magnesium as determined by milligrams of elemental magnesium administered minus the number of milligrams excreted in the urine (612 ? 180 vs. 216 ? 217 mg [25.1 ? 7.4 vs. 8.9 ? 8.9 mmol], p < .005) (Fig. 1) .

There was no difference between the treatment and control groups’ serum potassium concentrations on entrance into the study (3.4 ? 0.1 vs. 3.3 ? 0.1 mmol/L). Of note, both groups’ mean serum potassium concentrations increased significantly during the 6-hr interval after the first test dose was given, to 3.7 ? 0.1 (p < .01) and 3.8 ? 0.1 (p < .0001) mmol, respectively. After this increase, the concentrations did not change (Fig. 3) . The amount of potassium administered to the treatment and control groups, as calculated from replacement infusions, intravenous solutions, and parenteral nutrition plus tube feedings, did not differ significantly (245.1 ? 38.9 vs. 344.4 ? 45.4 mmol, p = .06). In contrast, the control group excreted significantly more potassium than the treatment group (418 ? 99 vs. 173 ? 30 mmol, p < .0005) over 48 hrs (Fig. 3) . The net potassium balance (potassium administered minus potassium excreted in urine) over the 48-hr period was significantly better in the treatment group (+72 ? 32 vs. -74 ? 95 mmol, p < .05) (Fig. 3) .

Bicarbonate and pH, calcium, phosphate, blood urea nitrogen, and creatinine values were the same on entry into the study. The initial mean serum glucose concentration in the treatment group was significantly higher than in the control group (188.1 ? 19.6 vs. 142.1 ? 8.7 mg/dL [10.5 ? 1.1 vs. 8.0 ? 0.5 mmol/L], p < .05), but both mean values fell within the clinically

Figure 1. Serum magnesium (Mg++) (mg/dL; 1 mg/dL = 0.41 mM) values and magnesium excreted in the urine over time. Magnesium balance represents the cumulative difference between magnesium administered and that excreted in the urine. Open circles, treatment group; solid squares, control group. *p < .05; # p < .005;? p < .0001. Values are expressed as mean ? SEM.

acceptable range for this population of critically ill patients. There were no significant variations in any value during the course of the study, when compared with study entry values (Table 2) .

A number of medications known to alter homeostasis of potassium and magnesium were administered to the study population as part of their medical management. These medications included diuretics, insulin or catecholamine infusions, and amphotericin. There were no differences between the groups with regard to use of these therapies (Table 1) . There was also no correlation between the number of diuretic doses administered and the total urine magnesium excretion over 48 hrs (r = .206, r2 = .04; p = .48) (Fig. 4) .

Of note, only three patients in the treatment group exhibited clinically important ventricular ectopy (frequent premature ventricular beats, pairs of premature ventricular beats, and/or short runs of ventricular tachycardia) during the first 12 hrs of the study. These three patients had 52%, 87%, and 96% decreases, respectively, in frequency of premature ventricular beats during the last 12 hrs of the study. On the other hand, five patients in the control group had an increase in the frequency of ectopy, comparing the same time frames. All other patients had little or no ventricular ectopy. No patient required acute intervention for treatment of dysrhythmias.

There were no complications associated with the administration of the magnesium. Only two patients had serum magnesium values of >4.0 mg/dL (>1.6 mmol/L); the highest magnesium concentration was 4.5 mg/dL (1.8 mmol/L). Neither of these patients showed clinical evidence of nausea, hypotension, or myocardial depression.


Hypomagnesemia is a common finding in hospitalized patients, and particularly in patients entering ICUs. Chernow et al. found that 61% of postsurgical patients had subnormal circulating magnesium concentrations on ICU admission. In addition, the severely hypomagnesemic patients had more hypokalemia and a higher mortality rate than similar acuity patients with normomagnesemia. Rubeiz et al. found that critically ill patients with hypomagnesemia had twice the mortality rate and a more rapidly fatal course compared with normomagnesemic patients.

Disease states may alter the distribution of magnesium through intra-extracellular shifts, altered binding, chelation, and sequestration in bony stores. Furthermore, malnutrition can limit the body’s ability to replenish intracellular stores of magnesium. Both alpha- and beta-adrenergic receptor stimulation causes a temporary efflux of magnesium from most cell types. In addition, free fatty acids released by beta-adrenergic receptor-mediated lipolysis chelate magnesium, decreasing free magnesium concentrations. Stress hormones, cellular ischemia, acidosis, and depletion of cellular adenosine 5′-triphosphate (ATP) associated with hypoperfusion can cause release of magnesium from intracellular binding sites, an increase in the concentration of intracellular ionized magnesium, and the subsequent efflux of magnesium from the cell.

Figure 2. Number of test doses administered over time to the control group (solid bars) compared with the treatment group (open bars). *p < .05; **p < .01 both comparing the control with the treatment group; ?p < .005 when comparing the treatment group with itself at time 0. Values are expressed as mean ? SEM.

Print This Post Print This Post

Pages: 1 2