MEASURING AND REPLETING SERUM POTASSIUM IS AN entrenched practice in treating patients with acute myocardial infarction (AMI), is recommended as routine (Class I) in practice guidelines, and is often codified in hospital admission templates and electronic orders. Achieving a target potassium level requires substantial resources. Frequent blood sampling is necessary to monitor potassium concentration and renal function to determine appropriate dosing according to standardized algorithms. Intravenous potassium replacement is slow, requires strict infusion control in a separate line to prevent potentially lethal overdoses, and when administered peripherally, may be painful and lead to thrombophlebitis. Oral potassium is unpalatable and often insufficient to achieve the target concentration. Moreover, complications of usual care to treat MI may worsen renal function and increase the risk of hyperkalemia, especially in elderly patients. How did potassium repletion become so ingrained into standard treatment of patients with AMI? A critical review of the evidence reveals only multiple, relatively small observational studies coupled with plausible pathobiology. Sixty years ago, hypokalemia was reported to lower the fibrillation threshold in isolated rabbit hearts treated with calcium infusion. During myocardial ischemia in patients, adrenergic stimulation activates the Na/K-ATPase pump and drives potassium into cells thereby reducing extracellular potassium. The regional hyperpolarization from potassium flux exacerbates electrical heterogeneity within the myocardium and creates substrate for ventricular arrhythmias. Therefore, repletion of extracellular potassium in the setting of ischemia would seem a reasonable strategy to reduce ventricular electrical irritability. Subsequently, during the 1980s, several studies ranging from 60 to 1400 patients with AMI supported a relationship between hypokalemia and the risk of ventricular arrhythmias. The incidence of ventricular fibrillation (VF) or ventricular tachycardia was especially high with serum concentrations of less than 3.5 mEq/L ( 8% in some reports) and was uncommon at levels of greater than 4.5 mEq/L. Unlike for magnesium repletion, there are no randomized trials of potassium administration for arrhythmia prevention in patients with AMI. Based on the available data and acknowledging a lack of randomized trials, current practice guidelines, including the American College of Cardiology/American Heart Association guidelines for management of ST-elevation MI, recommend (Class I) maintaining potassium levels of greater than 4.0 mEq/L, although with the lowest level of evidence (C). Others advocate a target of greater than 4.5 mEq/L. These guidelines do not suggest an upper level above which might be undesirable. These recommendations are based on experimental work and observations from small nonrandomized studies (mostly 20 years ago) using predominantly surrogate end points. Moreover, the treatment and natural history of AMI have since changed substantially, leading to a significant decline in lethal arrhythmias (which prompted the creation of coronary care units) by one-third or more over the past decade. For example, from 1995-2005, the risk of VF decreased more than 2-fold among 7472 post-AMI patients (4.2%-1.9%). Another large international registry ( 50 000 patients) showed that the risk of VF post-MI declined by almost one-third (6.4%4.4%) between 1999-2005. How important then is measuring serum potassium in the currentmanagementofAMIandwhat is theoptimalpotassium level? The study by Goyal et al in this issue of JAMA provides meaningfulnewdataabout theriskofhypokalemiaandhyperkalemia ina largecontemporarycohort. In this studyof38 689 patients with AMI, they reported a U-shaped relationship betweenadmissionandmeanpostadmissionpotassiumlevelswith in-hospitalmortality,adjustingforbaselinecharacteristics,medication use, and postadmission procedures and complications includingacutekidneyinjuryandthedevelopmentofheart failureandcardiogenicshock.The lowest-riskpatientswere those withpotassiumlevelsof3.5 to4.0mEq/L.Consistentwithpreviousreports,patientswithhypokalemia( 3.5mEq/L)onadmissionweremore likely todevelopsubsequentcardiacarrest and those with potassium levels of less than 3.0 mEq/L were at increasedriskof in-hospitalmortality.Admissionpotassium levels of greater than 5.0 mEq/L were also associated with increased mortality, but not VF or cardiac arrest. The interpretation of these results becomes more complicated when examining the relationship between postadmission potassium levels and outcomes. Even though severe postadmission hypokalemia ( 3.5 mEq/L; n=804; 2.1%) and hyperkalemia ( 5.0 mEq/L; n=1091; 2.9%) occurred