Interruptive Alerts Boost Suicide Risk Screenings by Over 10x, Study Finds

Credit: Adobe Stock/Sandu

Research published in JAMA Network Open found that interruptive computer alerts — pop-up windows that required immediate attention — were much more effective for screening for suicide risk compared with less intrusive noninterruptive alerts. These alerts led doctors to screen for risk in-person in 42% of patient visits compared with only 4% with noninterruptive alerts.

Investigators emphasized that the findings suggest that more direct prompts could help increase the identification of suicide risks during routine medical care. However, neither group reported any cases of suicidal ideation or attempts during the study.

Although traditional suicide risk identification focuses on patient-reporting, face-to-face screening and support network reporting, a few recent studies have used statistical risk models to enhance clinical judgment. However, investigators noted potential barriers to clinical decision support (CDS) include a potential alert burden for a stigmatized clinical condition coupled with a scarcity of data on the best practices for integration.

“Screening within health care encounters remains a priority for the field, given the incidence of suicide shortly after such encounters; in some cases, suicidal self-harm occurred the same day in which at-risk patients were seen,” wrote a team of investigators led by Colin G. Walsh, M.D., associate professor of Biomedical Informatics, Medicine and Psychiatry at Vanderbilt University. “Screening remains particularly relevant in primary and non-mental health specialty care, the most common point of contact in the year prior to death due to suicide, with 77% of those who die by suicide seen in primary care in the preceding year.”

To determine the effectiveness of risk model-driven CDS on assessing suicide risk, investigators performed a comparative effectiveness randomized clinical trial between August 2022 and February 2023 within the Department of Neurology across the divisions of Neuro-Movement Disorders, Neuromuscular Disorders and Behavioral and Cognitive Neurology at Vanderbilt University Medical Center.

Patients were recruited into the study at visit check-in and randomly assigned to either the interruptive or noninterrupted CDS groups. In the interruptive alert cohort, an on-screen pop-up and patient panel icon were simultaneously displayed. The noninterruptive alert cohort displayed with patient panel icon, but not the pop-up. The patient summarization panel showed an “elevated suicide risk score,” which doctors could choose to hover over, activating a pop-up identical to the interruptive alert.

The primary end point was the ultimate decision to assess a patient’s risk of suicide in person, while other end points included the rates of suicidal ideation and attempts in both groups.

In total, 561 patients with 596 encounters were randomized. The mean age of patients was 59.3 years and approximately half (52%) were women.

After adjustments, the interruptive CDS resulted in significantly higher numbers of decisions to screen for suicide risk (121 of 289 encounters) with documented assessments compared with the noninterruptive CDS (12 of 307 encounters) as well as in comparison with the baseline rate in the previous year (8%).

Investigators mentioned limitations including the possibility that suicide risk may have been evaluated regardless of the alert. Additionally, they were unable to assess any changes in rates of suicide attempts or deaths as none occurred during the study. Results may have been impacted by performance bias, although the team noted the noninterruptive alerts resulted in a lower decision-to-screen rate when compared with the previous year’s baseline.

“A larger-scale trial focused on the effectiveness of the overall system powered sufficiently against standard of care to reduce suicide events is indicated,” investigators concluded. “Future research might also further iterate CDS design, test these systems in more diverse settings and integrate preventive recommendations beyond risk assessment.”