Thiazide Diuretic-Induced Change in Fasting Plasma Glucose: a Meta-analysis of Randomized Clinical Trials

Abstract

Background: Prior meta-analyses measuring thiazide-induced glycemic change have demonstrated an increased risk of incident diabetes; however, this measure's definition has changed over time.

Aim: To determine the magnitude of change in fasting plasma glucose (FPG) for thiazide diuretics.

Data sources: A research librarian designed and conducted searches in Medline®, EMBASE, and EBM Reviews-Cochrane Central Register of Controlled Trials (inception through July 2018) and International Pharmaceutical Abstracts (inception to December 2014).

Study selection: Randomized, controlled trials comparing a thiazide or thiazide-like diuretic to any comparator reporting FPG were identified. Trials enrolling < 50 participants, those with a follow-up period of < 4 weeks, and conference abstracts were excluded.

Data extraction: Independent duplicate screening of citations and full-text articles, data extraction, and assessment of risk of bias was conducted.

Data synthesis: Ninety-five studies were included (N = 76,608 participants), with thiazides compared with placebo, beta-blockers, calcium channel blockers, renin-angiotensin-aldosterone-system inhibitors, potassium-sparing diuretic, and others alone or in combination. Thiazide diuretics marginally increased FPG (weighted mean difference 0.20 mmol/L (95% CI 0.15-0.25); I² = 84%) (1 mmol/L = 18 mg/dL). Results did not change substantially when considering dose or duration, comparing thiazides with placebo or an active comparator, or using thiazides as monotherapy or combination therapy, even when combined with a potassium-correcting agent.

Conclusion: Thiazide diuretics have a small and clinically unimportant impact on FPG.

Keywords: fasting plasma glucose; meta-analysis; randomized controlled trial; thiazide diuretic.

Figure 1

Summary of study retrieval and identification for meta-analysis.

Figure 2

Summary of risk of bias for trials included in meta-analysis.

Figure 3

Summary of forest plot demonstrating the weighted mean difference in fasting blood glucose (1 mmol/L = 18 mg/dL) in patients treated with thiazide diuretics overall and in sensitivity and subgroup analyses.