Effects of low-dose thiazide diuretics on fasting plasma glucose and serum potassium-a meta-analysis

This study is a meta-analysis of the metabolic profile (fasting plasma glucose and serum potassium) of low-dose thiazide and thiazide-like diuretics. The meta-analysis involved 10 randomized controlled clinical trials with a total sample size of 17,636 and 17,947 for the potassium and glucose arms respectively. The random effect model was used to calculate the odds ratio with 95 percent confidence interval. The cumulative mean change of fasting plasma glucose was +0.20 mmol/L (+3.6 mg/dL) for the diuretic arm versus +0.12 mmol/L (+2.2 mg/dL) for the comparator arm. The cumulative mean change of serum potassium was -0.22 mmol/L (-0.22 mEq/L) for the diuretic arm versus +0.05 mmol/L (+0.05 mEq/L) for the comparator arm. The aggregate odds ratio for having higher fasting plasma glucose in subjects on low-dose thiazide versus non-thiazide antihypertensive was 1.22 (1.11 to 1.33; P < .01). The odds ratio for having a lower serum potassium in subjects on low-dose thiazide versus non-thiazide antihypertensive was 0.36 (0.27 to 0.49; P < .01). The magnitude of the observed change in fasting plasma glucose associated with low-dose thiazide diuretic use, while statistically significant, does not appear to place patients at clinically significant risk. On the other hand, the observed change in serum potassium was also statistically significant, and may be clinically significant in patients whose baseline potassium concentration is low or low-normal, and could predispose at-risk patients, such as those with ischemic heart disease, to ventricular arrhythmias.