Prediction Models for Suicide: Not Yet Ready

Peter M. Yellowlees, MBBS, MD


May 24, 2019

This transcript has been edited for clarity.

This is the Medscape Psychiatry Minute. I'm Dr Peter Yellowlees.

Prediction models that use algorithms on large-scale data sources have the potential to improve the identification of patients at heightened suicide risk, but are they useful?

To answer that question, a team of investigators[1] from the Defense Health Agency in Maryland performed a systematic review of 17 cohort studies assessing the use of these prediction models across five countries and in more than 14 million participants. Their goal was to evaluate the diagnostic accuracy of these models in predicting suicide and suicide attempts, and to simulate the effects of implementing suicide prediction models using population-level estimates of suicide rates. They found that although the research quality of the included studies was generally high and classification accuracy was good (≥ 0.80 in most models), the predictive validity associated with a positive result for suicide mortality was extremely low (≤ 0.01 in most models), as were simulations of the results across a variety of population assessment characteristics.

This important study serves as a useful baseline marker of the lack of capacity of historical clinical data systems to be used to help predict suicide, a relatively rare event in comparison with the number of participants in these studies. Several new suicide prediction models are currently being developed for use across enterprise-level healthcare systems by the US Department of Defense, US Department of Veterans Affairs, and Kaiser Permanente. We now know that these important studies are possible from a methodologic perspective, and that suicide classification is quite accurate. What the current investigators have to focus on is improving the predictive validity of the numerous factors contributing to suicide so that the electronic health records that most of us use can really add value to our clinical practice.

Thank you for listening to this Medscape Psychiatry Minute. Do continue to enjoy your practice.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.