Serum potassium testing in diuretic-treated outpatients. A multivariate approach.

We evaluated risk factors for abnormal serum potassium (K+) concentrations in 5817 outpatients on chronic diuretic therapy. Hypokalemia (K+ less than 3.5 meq/L) occurred in 18.8 percent at the last clinic measurement. Hyperkalemia (K+ greater than 5.5) was rare (0.5%) and was even less common in those also taking KCl or a K+ sparing drug (0.2%). Hypokalemic patients were younger, healthier, more often hypertensive, more commonly taking a thiazide diuretic, and more likely to have been hypokalemic in the past. Hyperkalemic patients were older, more often suffering from heart or renal failure, and more commonly taking furosemide. Logistic regression analysis yielded eight variables that contributed significantly to the prediction of hypokalemia: mean and most recent serum K+, thiazide use, the use and duration of use of potassium-sparing diuretics, age, and most recent serum sodium and bicarbonate determinations. The regression equation allowed us to identify a group of patients (63% of the total) at low risk for hypokalemia (K+ less than 3.5 in 9.4%; K+ less than 3.0 in 0.5%) and another group (37% of the total) at high risk for hypokalemia (K+ less than 3.5 in 35%; K+ less than 3.0 in 5%). Information about the risk of hypokalemia can be used to decide when to measure serum potassium concentration in individual diuretic-treated outpatients.