An algorithm may be able to flag veterans at high risk of suicide, according to new research — even when the signs of risk have not been obvious to doctors or other medical personnel.
“This [study] provides us with unprecedented information that will allow us to design and implement innovative strategies on how to assess and care for those veterans who may be at high risk for suicide,” said Caitlin Thompson, deputy director for suicide prevention for Veterans Affairs. “This model will advance the care provided to veterans through VA’s suicide prevention programs to allow us to better tailor our suicide prevention efforts so that we can ensure that all veterans remain safe.”
The researchers used data from the Veteran Health Administration patient population recorded between 2009 and 2011. The team split that population in half, using one group to develop the model and the other half to test if its predictions were accurate. Each group included 3,180 suicide cases and 1,056,004 other patients as a control.
Such data-gathering has been made possible by the implementation of electronic health records. In addition to increasing overall efficiency (by about 6% a year), these record systems also facilitate the analysis of hundreds of different factors using computer programs, revealing patterns and combinations not visible on a smaller scale.
The researchers emphasized that the model can’t predict individual suicides, but rather pinpoint those patients who are at a higher risk. That makes it similar to a cholesterol number, in that while high cholesterol doesn’t necessarily indicate a person will have a heart attack, it does indicate measures should be taken to address that individual’s health.
The biggest question now is what clinicians should do once they know that an individual has a high suicide risk according to the predictive model. Those steps are still being developed, according to the VHA.
The full findings have been reported in the American Journal of Public Health.