Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

New Models Improve Suicide Risk Predictions

Woman at window

(Alex Ivashenko, Unsplash)

24 May 2018. Statistical models derived from data in electronic health records and standard questionnaires for depression are shown to better predict a person’s risk of suicide following a doctor’s office or mental health clinic visit. A description of the models, developed by researchers from Kaiser Permanente and other health care plans and providers, appears in today’s issue of the American Journal of Psychiatry (paid subscription required).

A team led by psychiatrist and mental health researcher Gregory Simon at the Kaiser Permanente Washington Health Research Institute in Seattle is seeking to tap into growing collections of patients’ electronic health records to improve their ability to spot individuals at risk of attempting or completing suicides. With better predictors, say the researchers, physicians and mental health experts can intervene earlier, follow-up more often, or reach out after patients miss or cancel appointments. Data from Centers for Disease Control and Prevention show the U.S. had nearly 45,000 suicide deaths in 2016, with rates of suicide higher in rural areas, and rising in both urban and rural regions.

Current statistical models, say the authors, use fewer variables from patients’ health records, but have only moderate success at predicting future suicide activity. With today’s models, those in the top 5 percent of risk account for only about a quarter to a third of later suicide attempts and deaths. Relying on traditional screening questionnaires or clinical visits, says the team, have an even lower predictive value.

For their new models, Simon and colleagues — including associates from the Henry Ford Health System in Detroit and HealthPartners Institute in Minneapolis — gained access to 5 databases of electronic health records maintained by Kaiser Permanente and those kept by the Henry Ford and HealthPartners. The combined collections provided records for nearly 3 million patients age 13 and above, between January 2009 and June 2015, from which identifying information was removed. These records indicate the individuals made nearly 10 million outpatient visits to primary care health providers, where the clinicians made a mental health diagnosis. Another 10.3 million visits were made to outpatient mental health clinics.

The researchers developed separate models for primary care and mental health clinic visits, and tested their predictive value with more than 24,000 suicide attempts and some 1,200 suicide deaths recorded among these individuals. The models show previous suicide attempts and mental health diagnoses were strong predictors of suicide attempts within 90 days of an outpatient visit, as well as scores on standard depression assessment questionnaires, substance abuse, diagnosis of other medical conditions, prescriptions for psychiatric medications, and hospital or emergency room care.

Using these predictors, the models show individuals in the top 1 percent of suicide risk were 200 times more likely to attempt or complete suicide than people in the bottom half of the risk scale. Among individuals making primary care visits, those scoring in the top 5 percent of suicide risk accounted for nearly half (48%) and 43 percent of suicide deaths. For people visiting mental health clinics, a similar pattern emerges, with those in the top 5 percent of risk making up 43 percent of suicide attempts and 48 percent of suicide deaths.

Kaiser Permanente and its partner health systems have programs to increase awareness of mental health issues and encourage patients to talk with their health care providers about their their depression and other conditions. “But we know we could do better,” says Simon in a company statement. “So several of our health systems, including Kaiser Permanente, are working to integrate prediction models into our existing processes for identifying and addressing suicide risk.”

More from Science & Enterprise:

*     *     *

Comments are closed.