Assessment of access to firearms in patients reporting suicidal ideation and risk of subsequent suicide attempts

The number of suicides has been steadily increasing in the United States in recent years. Suicide is considered a major contributor to declining life expectancy in the United States. Firearms are among the most lethal means for suicide, certainly more lethal than other common alternatives. Restricting access to lethal means for suicide through policy initiatives, such as waiting periods for firearm purchases, has been shown to be effective in reducing suicide outcomes. Counseling to reduce access to firearms by medical and mental health providers is highly recommended by several national suicide prevention agencies for patients with suicide risk, but no formal evaluations of this clinical intervention have been completed. We examined the effectiveness of firearm access assessment, which involves a provider asking a patient about access to firearms and then counseling them to restrict access if appropriate. This intervention targets people who have suicide ideation. To this end, we estimated the impact of firearm access assessment, delivered in behavioral health and medical settings, on suicide attempts and death.

Using data from a large integrated health system, we identified 8,083 individuals with suicide ideation based on PHQ-9 depression screener that was administered universally in the behavioral health settings from 2010- 2016. We used a validated Natural Language Processing program to identify receipt of the intervention in the electronic medical record. Among the patients in the sample with suicide ideation, only about 33% of these individuals received the intervention despite clinical recommendations suggesting patients with suicide risk should be screened for firearm access.

We estimated a 2-stage residual inclusion instrumental variables model to control for non-random selection onto treatment and measure the local average treatment effect (LATE) of the intervention on suicide attempts/deaths within 180 days of the ideation event. The instrument was the cumulative probability of receiving the intervention, which we calculated at the practice-level across all participants, then applied to each individual based on specific practices visited in the 30-day follow-up period. As evidenced by a first-stage F-test of 46.58 this instrument predicts receipt of counseling very well. We present evidence that the instrument is unlikely to directly affect suicide deaths based on the other processes in place at the clinics. In other words, the intervention was credibly independent of behavioral health treatment of suicide ideation. We adjusted for other factors (i.e. confounders) related to the patient’s condition and utilization. We found that firearm access assessment reduced the risk of a suicide attempt within 180 days from 3.6% to 0.83% with a relative risk of 0.23 (p = .016). The interpretation of the LATE is based on receipt of the intervention due to where an individual was treated (i.e. residence) rather than a clinical evaluation of perceived need. The LATE may be due to firearm access assessment as well as the resulting prevention strategies the often co-occur with it. Mental health providers should expand the use of counseling on access to lethal means, along with co-occurring safety planning practices, to all patients who report suicide ideation.