The Person and the Place: Community Data in Suicide Risk Prediction

If you only look at the patient, you miss part of the story. If you only look at the neighborhood, you miss the person. Suicide prevention planning needs both views.

A new study led by Dr. Matthew Castner, a DrPH graduate of the Johns Hopkins Bloomberg School of Public Health, focuses on the second view: how well information about communities can help identify where suicide-related outcomes are more likely. "Assessing the utility of health access data and social determinants of health in ecological suicide prediction models" was published online in Social Psychiatry and Psychiatric Epidemiology in July 2026 (Castner et al., 2026). The paper grew out of Dr. Castner's dissertation research and is his first peer-reviewed publication.

Co-authors include Center Medical Director Dr. Paul Nestadt, Center Director Dr. Holly Wilcox and Center faculty member Dr. Hadi Kharrazi.

What the study asked

"Ecological" models look at areas rather than individuals. Many such models start with basic demographic information about a community. The team asked whether adding three other kinds of area-level information would improve predictions:

• clinical conditions in the community,
• health care access, and
• social determinants of health, the conditions in which people live, work and grow up.

The outcomes studied included suicidal ideation or attempt, psychiatric hospitalization and suicide death.

What they found

Adding clinical information produced the largest improvement in the models. Health care access and social determinants of health also improved the models, but by less.

One access measure stood out. The availability of emergency mental health services in an area was associated with lower rates of psychiatric hospitalization and of suicide death.

Why it matters

Much of the information used in these models is publicly available at the area level. That means health departments and communities could use similar data to see where prevention resources may be needed most, and plan accordingly.

The finding on emergency mental health services is also a reminder that the care available in a community is part of the picture, not only the characteristics of the people who live there.

Limits

Because the study looks at areas, its findings describe patterns across communities. They cannot tell us about the risk of any individual person, and an association between an area's characteristics and its outcomes does not show that one causes the other. Prediction models also depend on the quality of the data that go into them.

The Center congratulates Dr. Castner on his first peer-reviewed publication.