Study: VA Database Could Help Measure Patient Suicide Risk

June 12, 2015 in News

A database developed by Department of Veterans Affairs and NIH researchers could use predictive modeling to help identify veterans at high risk of suicide, according to a study published Thursday in the American Journal of Public Health, the New York Times reports.

Background

The database — which researchers began developing four years ago — uses electronic health records to help flag veterans who are at risk of suicide. The system searches for patterns among millions of EHRs using 381 variables, such as:

  • Diagnoses;
  • History of suicide attempts; and
  • Prescribed medications.

To test whether the system could accurately predict suicide risk among veterans, researchers created an algorithm based on data from half of the patient population between 2008 and 2011. The algorithm then was tested on the other half of the patient population from that time period (Philipps, New York Times, 6/11). The study included records on more than 3,000 VA patients who had committed suicide (Zoroya, USA Today, 6/11).

According to the Times, the study marks the first time predictive risk was used to analyze veteran suicides on a large scale.

Findings

Overall, predictions for the test cohort aligned with actual suicide rates, according to Robert Bossarte, director of epidemiology at VA’s Office of Public Health.

Meanwhile, the study found that VA, which currently relies on doctors and other staff to flag high-risk patients, was missing a majority of such individuals (New York Times, 6/11). Such practices only identified about 33% of patients with the highest risk of suicide.

Using the database, researchers were able to identify about 600 individuals with suicide rates 60 to 80 times higher than other VA patients (USA Today, 6/11).

The database also identified a subset of 1% of the veterans with the highest risk scores who were 12 to 14 times more likely to commit suicide than the other VA patients.

Implications

Bossarte cautioned that while the database can be used to flag high-risk individuals, it cannot necessarily predict suicide. Instead, he said the database can help identify cases for which VA “should take measures to lower the risk.”

Meanwhile, Caitlin Thompson, deputy director for suicide prevention at VA and study co-author, said the department is still considering how to best use the data (New York Times, 6/11). However, she noted that the database could be “a game changer in terms of suicide prevention overall and not just for the VA population” (USA Today, 6/11).

House Subcommittee Discusses VA Suicide Data

In related news, a House Committee on Veterans’ Affairs subcommittee during a hearing on Wednesday heard testimony from stakeholders about issues, such as software errors, that led to erroneous VA suicide data, Modern Healthcare reports.

During the hearing, Randall Williamson, director of health care at the Government Accountability Office, gave an update to a November 2014 GAO report that reviewed VA’s antidepressant prescribing practices and the accuracy of the department’s suicide data.

According to Modern Healthcare, stakeholders discussed how VA after the report uncovered a “software-mapping error” that led to the incorrect documentation of major depressive disorder diagnoses among veterans. Meanwhile, some dates of veterans’ deaths had been incorrectly entered in the electronic templates used for VA’s Behavioral Health Autopsy Program, which made it difficult accurately determine what treatment had been provided (Robeznieks, Modern Healthcare, 6/11).

Be the first to like.
VN:F [1.9.22_1171]
Rating: 0.0/5 (0 votes cast)

Article source: http://feedproxy.google.com/~r/Ihealthbeat/~3/lPJ_BBNmemA/study-va-database-could-help-measure-patient-suicide-risk

Be Sociable, Share!
Bookmark and Share

Leave a reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>