MIRECC / CoE

Dr. Claire Houtsma Published in Nature Mental Health

Congratulations to Dr. Houtsma on her article, “Predicting firearm suicide among US Army Veterans transitioning from active service”, being published in Nature Mental Health. Dr. Houtsma participated in a brief interview about her article with us.

Q. You’ve been published in the highly esteemed journal, Nature Mental Health. How do you feel about earning this achievement?

Absolutely thrilled! I would not have achieved this without the amazing opportunities afforded to me through the VA-STARRS Researcher in Residence Program offered by the VA Office of Research and Development (ORD) and Suicide Prevention Research Impact NeTwork (SPRINT). This funded position gave me the chance to work with incredible researchers such as Ron Kessler, Brian Marx, and Chris Kennedy who are involved in examining data from Army Soldiers through the Study to Assess Risk and Resilience in Servicemembers – Longitudinal Study (STARRS-LS). It has been a privilege to have access to this data and to use it to try and prevent suicide among service members and Veterans.

Q. Can you give us an overview of your article?

This study involved development of a firearm-specific machine-learning model to compare against a suicide method-agnostic model (i.e., a model designed to predict suicide regardless of method) to determine which is more effective at predicting firearm suicide risk among recently separated Soldiers—a period known to carry significantly elevated suicide risk for Veterans. Using data on ~1,400 firearm-related suicides, it was determined that the best model for predicting firearm suicides depends on the intervention threshold: a method-agnostic model works best at high thresholds (i.e., 100 suicides/100,000), while a firearm-specific model is more effective at lower thresholds (i.e., 75 to 90 suicides/100,000). Prediction models for firearm suicide prevention among recently separated Soldiers need to balance the nature of firearm suicides, characterized by brief transition periods and high lethality, with the health care system's workload. Therefore, firearm-specific interventions might need to be deployed at lower thresholds compared to general suicide prevention strategies to optimize resource allocation. These findings offer valuable insights that may pave the way for further advancements in the field.

Q. What key points do you want readers to take away from your article?

Using the firearm-specific prediction model means using a lower risk threshold and, in turn, intervening with a greater proportion of the target population. This carries implications for workload burden. However, it is important to keep in mind that the transition from thinking about suicide to attempting suicide can be very brief, and using a firearm in a suicide attempt is almost always fatal. Therefore, using a firearm-specific prediction model to identify and intervene with those potentially at risk may be an effective way to prevent firearm suicide in this population.

Q. Anything else you want our readers to know?

If other early career researchers are interested in this type of work, they should consider applying for the Researcher in Residence program and explore other opportunities through SPRINT!

Last updated: January 27, 2026