Second-Generation Antidepressants and Hyponatremia Risk: A Population-Based Cohort Study of Older Adults

Hyponatremia may occur after initiation of a second-generation antidepressant drug. However, the magnitude of this risk among older adults in routine care is not well characterized.

Retrospective, population-based, matched-cohort study.

In Ontario, Canada, 2003 to 2012, we compared older adults with a mood or anxiety disorder who were dispensed 1 of 9 second-generation antidepressant drugs with matched adults with comparable indicators of baseline health who were not dispensed an antidepressant drug (n=138,246 per group). A similar comparison was made in a subpopulation with available laboratory data (n=4,186 per group).

Second-generation antidepressant prescription versus no antidepressant prescription.

The primary outcome was hospitalization with hyponatremia. A secondary outcome was hospitalization with both hyponatremia and delirium.

We assessed hospitalization with hyponatremia using a diagnosis code and, in the subpopulation, serum sodium values. We assessed hospitalization with hyponatremia and delirium using a combination of diagnosis codes.

Second-generation antidepressant use versus nonuse was associated with higher 30-day risk for hospitalization with hyponatremia (450/138,246 [0.33%] vs 84/138,246 [0.06%]; relative risk [RR], 5.46 [95% CI, 4.32-6.91]). This association was consistent in the subpopulation with serum sodium values (73/4,186 [1.74%] vs 18/4,186 [0.43%]; RR, 4.23 [95% CI, 2.50-7.19]; absolute risk increase, 1.31% [95% CI, 0.87%-1.75%]). Second-generation antidepressant use versus nonuse was also associated with higher 30-day risk for hospitalization with both hyponatremia and delirium (28/138,246 [0.02%] vs 7/138,246 [0.005%]; RR, 4.00 [95% CI, 1.75-9.16]).

Measures of serum sodium could be ascertained in only a subpopulation.

Use of a second-generation antidepressant in routine care by older adults is associated with an approximate 5-fold increase in 30-day risk for hospitalization with hyponatremia compared to nonuse. However, the absolute increase in 30-day incidence is low.