There is general agreement–but not an official statement that I could find–that in all comers, K <3.0 should be repleted. In patients with a history of past cardiac surgery, heart disease, and definitely in the post-MI population, K<3.5 should be repleted for a goal of 4.0. When there is acute concern for torsades or other arrhythmia, there is again general agreement but no official consensus that the goal is raised to >4.5.
Remember action potentials? The ins and the outs with K, Na, and Ca with the alphabet soup of channels? (Brief review in the first section of this editorial.) In the short-term, having a low serum K affects repolarization and has a chain effect on the action potential, causing increased automaticity, excitability, and QT prolongation, potentially triggering fatal arrhythmia. In the long-term, hypokalemia is associated with cardiovascular mortality in patients with underlying heart disease, arrhythmias like RBBB, and heart failure.
NB: Kind of supporting this are findings that higher doses of thiazide diuretic are linked to sudden cardiac death. Some argue that the mortality benefit of ACE inhibitors and beta-blockers in heart failure comes in part from an ability to better stabilize potassium levels. (Beta-blockers keep potassium extracellular through beta-2 receptors.)
Many studies linking hypokalemia to arrhythmia were relatively smaller studies (it seems like anything fewer than n=5,000 is not impressive in general cardiology) done in the 1980s, with mixed patient populations of mostly acute MI, hypertension (looking specifically at thiazide diuretics), or heart failure. These studies implied that the higher the potassium (K>4.5) the better:
A more recent population-based cohort study of post-MI patients in JAMA (n>38,000), on the other hand, showed the following:
Because the lowest rate of mortality was found in the group with K 3.5-4.5, a goal of 4.0 is generally set for post-MI patients, which was extrapolated to any patient with heart disease. A similar distribution was found in patients who also had renal disease, and this study based on data from MERLIN-TIMI 36. This is a good reminder that hyperkalemia is linked to increased cardiovascular mortality, too.