Suicide and death from opioid misuse is a national public health emergency with veterans suffering twice the suicide rate of the general population, according to a recent VA National Suicide Data Report. With resources from a unique R01 grant co-funded by Veterans Affairs (VA) and the National Institutes of Health (NIH), Jyotishman Pathak, Ph.D., Frances and John L. Loeb Professor of Medical Informatics, professor of psychiatry, and chief of the Division of Health Informatics in the Department of Population Health Sciences, and colleagues are beginning work on a study at the heart of this issue.
Together, the researchers will apply machine learning algorithms on electronic health record data from the VA and multiple academic medical centers across New York City, Chicago and the State of Florida to identify veterans on chronic opioid therapy (COT) at high risk of death from suicide or accidental opioid overdose, as well as suicidal ideation and attempt. The particular combination of methods being applied makes the study one of the first of its kind.
As principal investigator, Dr. Pathak will work with Co-PI, David Oslin, M.D., director of the VISN 4 Mental Illness Research, Education, and Clinical Center (MIRECC), associate chief of staff for behavioral health at the Philadelphia VA Medical Center, and professor of psychiatry at University of Pennsylvania Perelman School of Medicine. Other experts contributing to the study are from Weill Cornell Medicine, University of Florida, Northwestern University, Pittsburgh VA Medical Center, and VA Salt Lake City Health Care System, among others. The innovative nature of the collaboration brings together private and public health care institutions to leverage several governmental resources to address one of the nation's most pressing public health challenges.
“This project is a collaboration across a wide range for experts in informatics, mental health services research, substance use research and healthcare policy,” said Dr. Pathak. “The nature of the problem that we are trying to address requires such a team science approach.”
Patients with chronic pain often carry additional comorbid risk factors for suicide, such as post-traumatic stress disorder (PTSD), depression and anxiety. Reports show that nearly half of veterans suffer from chronic pain, and the suicide rate among younger veterans (ages 18-34) has steadily risen to more than three times the national average.
While national initiatives have attempted to decrease the number of veterans with opioid prescriptions for pain management and address suicide risk, they have primarily analyzed VA data alone. However, care fragmentation from receiving opioids from both VA and non-VA sources puts these veterans at greater risk for potentially unsafe prescription opioid use and a high risk of suicide. Combining the data will widen the scope of “dual user” risks and potentially provide evidence for more effective initiatives.
The research is especially timely considering the recent passage of the VA MISSION Act, which is designed to further increase Veteran patient access to non-VA opioid treatment.
To complete the study, Dr. Pathak and colleagues will integrate large-scale, real-world clinical VA and non-VA data to study the risk of deaths (suicide and accidental opioid overdose) and suicidal ideation and attempts in veterans on COT. The researchers have electronic health records (EHRs) on over 10 million patients from VA’s Corporate Data Warehouse and over 34 million patients from three PCORI-funded clinical data research networks. These three networks represent patients in New York City (data was provided by Weill Cornell Medicine’s partner, INSIGHT CRN), Chicago and Florida between 2010 and 2018. The records will help identify patients receiving COT and establish links between veterans who received both VA and non-VA services.
Using EHRs, the researchers will also incorporate risk factor detection, which has not yet been done widely.
“Studies suggest that, on average, 18 US veterans die by suicide every day,” said Dr. Pathak. “Several factors, such as insomnia, depression, anxiety, sexual victimization, gun ownership and substance use disorders, have been shown to contribute to suicide risk among service members and veterans. This project will study such risk factors.”
The researchers plan to use the five-year grant to explore future clinical implementation of risk models within the VA.
Overall, Dr. Pathak and his collaborators have one main goal: “We would like to develop a generalized suicide risk prediction model that can be eventually deployed for point-of-care clinical decision support to proactively identify veterans at risk of suicide.”