This article is more than 5 years old. Information may no longer be current.

Psychotropic use, psychiatric disorders strongly predict suicide risk

Machine-learning methods and data from Denmark’s single-payer health care registry suggested that psychotropic medication use and psychiatric disorders are significant factors for suicide risk, according to results published in JAMA Psychiatry. Moreover, a longer vs. shorter period of observation for many prescriptions and diagnostic variables appeared to be more important, researchers noted.

“To our knowledge, this study is the first to develop prediction models for suicide based on data from a full population,” Jaimie L. Gradus, DMSc, DSc, of the department of epidemiology at Boston University's School of Public Health, and colleagues wrote. “We found what appears to be consistency with what is known about suicide risk but also potentially important, understudied predictors for future study.”

According to Gradus and colleagues, the implementation of machine-learning methods in psychiatry allows for novel suicide risk profile development that include “broad constellations” of predictors. The researchers used population-based, prospectively recorded medical and social registry data for 14,103 individuals from Denmark who died by suicide between 1995 and 2015. As a comparison sub-cohort, they selected a random sample of 265,183 living individuals in Denmark on January 1, 1995. They used supervised machine-learning methods to develop sex-specific risk profiles for death from suicide, given the “well-established sex differences in suicide risk,” they wrote.

Across all models, the researchers reported that psychiatric disorders were the most important predictors of suicide — a finding largely consistent with previous suicide risk factor research, they noted.

Novel results of this study included a strong association between stress disorders and suicide risk among men and women in models that simultaneously evaluated depression. The consensus is firm that depression and schizophrenia are well-established risk factors for suicide, Gradus and colleagues wrote, but previous links between stress disorders and suicide risk were controversial.

They also found that antidepressants, antipsychotics, hypnotics/sedatives and medications for addictions were important for accurately predicting suicide across analyses. Although the results made clear that pharmacotherapy is important to suicide prediction, the researchers’ noncausal models were unable to determine the magnitude or direction of associations, they wrote.

For men, physical health diagnoses contributed more to suicide prediction than for women, and psychiatric diagnoses and associated medications were more important for the prediction of women’s suicide risk.

Also of note, the researchers found that in general, diagnoses and medications measured 48 months before suicide were more important indicators of suicide risk than they were when measured 6 months prior to suicide.

“Results of this study can possibly be replicated in other novel data sets and used to inform the further development of general population prediction models for suicide,” the researchers wrote.

In a related editorial, Seena Fazel, MBChB, MD, of the department of psychiatry at the University of Oxford, and Lauren O’Reilly, BS, of Indiana University, Bloomington, underscored the potential for future prediction models to alter suicide outcomes.

“Going forward, the development of prediction models in specific, high-risk groups (eg, United States Army soldiers after psychiatric treatment) or those with certain psychiatric diagnoses (eg, schizophrenia spectrum disorders) will likely have benefits of higher accuracy, more acceptability to clinicians and simpler linkage to interventions,” they wrote. “How they can be most effectively linked to evidence-based treatments needs careful examination; without such links, prediction models are unlikely to reduce suicidality outcomes.” – by Joe Gramigna

Disclosures: Fazel, Gradus and O’Reilly report no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.