Abstract 2754: Relationship of 1 Year Hypokalemia Incidence to Clinical Outcomes in the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT)
ALLHAT, a clinical trial of antihypertensive therapy, randomized older hypertensive subjects at high risk of cardiovascular disease (CVD) to initial therapy with chlorthalidone (C; n=15,255; 12.5–25 mg/day) vs. amlodipine (A; n=9,048; 2.5–10 mg/day) or lisinopril (L; n=9,054; 10 – 40 mg/day). Potassium (K+) supplements were available. Participants with initial and year 1 serum K+ measurements were stratified by K+ at 1 year. Post 1 year CHD and other CVD incidence were compared between those with K+<3.5 (hypoK+) and K+ 3.5–5.4 mmols/L (normoK+). Cox proportional hazard ratios (HR) for hypoK+/normoK+ were estimated, adjusting for important baseline characteristics (age, race, sex, history of diabetes, CHD, and atherosclerotic CVD, cigarette smoking, atrial fibrillation, SBP, and estimated year 1 glomerular filtration rate). Because HR estimates differed significantly across treatment groups, drug-specific analyses were done. Among participants, 19,731 had known baseline and year 1 K+ values and were normokalemic at trial entry: C, 9,159; A, 5,371; and L, 5,201. Baseline characteristics in this subset did not differ significantly from all participants. At 1 year, incidence of hypoK+ (<3.5 mmol/L) overall was 6.85% (n=1351); incidence in C (12.94%) differed significantly from A (2.10%; p<0.001) and from L (1.02%; p<0.001). After the first year, CHD occurred in 7.92% of the hypoK+ group and in 7.86% of the normoK+ group (n=17,982). Overall, the respective incidence of CHD, stroke, and heart failure did not differ significantly between hypoK+ and normoK+ groups. For drug-specific comparisons, CHD incidence for hypoK+ vs normoK+ differed significantly in A (HR=2.56, p=0.001;A/C interaction p<0.01). Corresponding HRs for C and L were 1.02 (p=0.851) and 1.20 (p=0.800), respectively (L/C interaction p=0.801). Combined CVD outcome HR was significantly lower in hypoK+ compared to normoK+ in C (HR=0.86, p=0.034), higher in A (HR=1.62, p=0.011;A/C interaction p<0.01), and non-significantly different in L (HR=1.40, p=0.253;L/C interaction p=0.080). These data confirm the increased incidence of hypoK+ associated with C compared to the other agents, but provide no evidence that diuretic induced hypoK+ is associated with increased cardiovascular risk.