Suicide Attempts in the 12 Months Following Incident Prescriptions of Sedative-Hypnotic Medications in a Large Healthcare System

Document Type

Poster Presentation

Publication Date



Introduction: Both sleep medications in general and insomnia have been associated with suicide risk. We sought to ascertain the frequency of suicide attempt (SA) by patients using any such medications and to provide SA rates by individual medication.

Methods: Subjects were receiving care in the US Veterans Health Administration (VHA) with an index (first time) prescription for a sleep medication in 2011, no SAs and no sleep medication prescriptions in 2010, and no additional sleep medication prescribed for 12 months after the incident prescription. Medications included benzodiazepines, z-drugs, ramelteon, sedating antidepressants, and antihistamines in dosages (e.g., 50–100mg of trazodone) and dosing schedules (e.g., excluding 1 day prescriptions) consistent with insomnia treatment. SA frequencies in the 12 months after the index prescription were tabulated by medication. Data were obtained from the VHA Corporate Data Warehouse and the Suicide Prevention Application Network, which captures all VHA veteran SAs known to VHA. SA incidence was calculated according to accepted methods and presented as events per 100,000 person years (PYs) with a 95% confidence interval (CI).

Results: 226,482 VHA users had an incident sleep medication prescription, representing 3.4% of VHA users. Among 15 medications identified, the most commonly prescribed were trazadone (24.4%), zolpidem (18.4%), hydroxyzine (15.3%), lorazepam (11.6%), and mirtazapine (8.1%). The total of recorded SAs within one year of incident prescription was N=454, representing 207 SAs per PYs (95% CI=188–227). Those with the highest rate of SA per PY had been prescribed mirtazapine (330), trazadone (296), hydroxyzine (188), zolpidem (178), and lorazepam (180). The lowest rates among medications with at least 10,000 incident prescriptions was observed for temazepam (99), diazepam (117) and amitriptyline (131).

Conclusion: The rate of SA following incident sleep medications was similar to that observed in other large veteran cohorts, though there was variability of SA rates across individual medications. An observation is that the incident prescription rate is low compared to expected rates of incident insomnia.

Support (If Any): VA Center of Excellence for Suicide Prevention.

Disclaimer: The authors’ views or opinions do not necessarily represent those of the Department of Veterans Affairs or the US Government.


Poster presented at the 2017 Sleep Conference in Boston, Massachusetts, June 2017.

Abstract published in Sleep: Official Publication of the Sleep Research Society, 2017, Volume 40, Abstract Supplement, A129:

This document is currently not available here.

Additional Files