Cumulative Incidence of Thiazide-Induced Hyponatremia : A Population-Based Cohort Study

Abstract

Background: According to drug labels, the frequency of thiazide-induced hyponatremia is unknown or uncommon to very rare (that is, <1 in 10 000 to <1 in 100), but the exact burden remains unclear.

Objective: To estimate the increase in the cumulative incidence of hyponatremia using thiazide diuretics compared with nonthiazide antihypertensive drugs in routine clinical practice.

Design: Population and register-based cohort study using target trial emulation.

Setting: Denmark, 1 January 2014 to 31 October 2018.

Participants: Two target trials were emulated among persons aged 40 years or older who had no recent prescription for any antihypertensive drug, had no previous hyponatremia, and were eligible for the studied antihypertensive treatments. The first target trial emulation compared new use of bendroflumethiazide (BFZ) versus a calcium-channel blocker (CCB). The second target trial emulation compared new use of hydrochlorothiazide plus a renin-angiotensin system inhibitor (HCTZ-RASi; that is, combination pill) versus a RASi alone.

Measurements: Two-year cumulative incidences of sodium levels less than 130 mmol/L using stabilized inverse probability of treatment-weighted survival curves.

Results: The study compared 37 786 new users of BFZ with 44 963 of a CCB and 11 943 new users of HCTZ-RASi with 85 784 of a RASi. The 2-year cumulative incidences of hyponatremia were 3.83% for BFZ and 3.51% for HCTZ-RASi. The risk differences were 1.35% (95% CI, 1.04% to 1.66%) between BFZ and CCB and 1.38% (CI, 1.01% to 1.75%) between HCTZ-RASi and RASi; risk differences were higher with older age and higher comorbidity burden. The respective hazard ratios were 3.56 (CI, 2.76 to 4.60) and 4.25 (CI, 3.23 to 5.59) during the first 30 days since treatment initiation and 1.26 (CI, 1.09 to 1.46) and 1.29 (CI, 1.05 to 1.58) after 1 year.

Limitation: The study assumed that filled prescriptions equaled drug use, and residual confounding is likely.

Conclusion: Treatment initiation with thiazide diuretics suggests a more substantial excess risk for hyponatremia, particularly during the first months of treatment, than indicated by drug labeling.

Primary funding source: Independent Research Fund Denmark.