Suicide is the tenth leading cause of death in the United States - and in some age groups, it's even higher - the second leading cause of death for those 10-34, and the fourth for those 35-54. Suicide among veterans continues to rise despite attempts to prevent it through resilience training, stress reduction, crisis centers, and a host of other interventions.
Despite widespread clinical screening, suicide is hard to predict, but a new George Mason University College of Health and Human Services study has found a way.
Dr. Farrokh Alemi, professor of health informatics, led the study published in Health Services Research in September in the Drivers of Health theme issue.
Alemi and colleagues analyzed U.S. Department of Veterans Affairs electronic health record (EHR) data for a population of more than five million veterans. They looked at both medical and social determinants of health (SDOH) data, although SDOH data was less frequently reported.
"We found that social determinants of health, such as homelessness, divorce, unemployment, are predictors of suicide or self-harm, but they are not necessarily the cause. Mental illness causes both these adverse life events and suicide," explains Alemi. "Programs that reduce social risk factors may not reduce suicide and self-harm, unless they also address the root cause— mental illness," explains Alemi.
They found that SDOH can predict suicide and self-harm on their own, yet medical history, such as diagnosis of major depression, psychosis, or history of self-harm, was an even better predictor than SDOH. The combined medical history and SDOH taken together were not a better predictor of suicide and self-harm than medical history alone.
"Frustration with the rise in suicides have led many to call for abandoning suicide prevention efforts. It seems to not work," explains Alemi. "Our study shows that risk modification efforts could be improved, and will have a larger impact, if they focus on medical issues such as use of antidepressants. Modifying SDOH will improve wellness but may not reduce suicide."
The researchers recommend additional study among larger groups of women and ethnic and racial groups to test the accuracy of these predictions in those groups, as this study used veteran data which includes primarily Caucasian male participants. They also encourage health care providers and claims processors to report SDOH, as these data may be able to help predict suicide, self-harm, and countless other conditions in the future. Finally, they point out that improving wellness, reducing isolation, is an important goal by itself, independent of whether it reduces suicides.
As a supplement to the study, the researchers also developed a Social Determinants of Illness index based on EHRs to assist in predicting impact of stressful life events on suicide.
More information: Farrokh Alemi et al, Relative accuracy of social and medical determinants of suicide in electronic health records, Health Services Research (2020). DOI: 10.1111/1475-6773.13540
Journal information: Health Services Research
Provided by George Mason University