“Firearm injury is a critical problem in our society that profoundly impacts people’s lives,” said Dr. Hechter, a research scientist in the Division of Epidemiologic Research in the Department of Research & Evaluation. “We believe it is our responsibility to use our public health expertise to contribute to a solution.”

Her multidisciplinary team successfully applied for funding to study risk factors for firearm injury and clinical opportunities for prevention.

In April 2023, in the journal Academic Pediatrics,¹ they published their first findings: an analysis of nearly 2 million pediatric Kaiser Permanente members that highlights risk factors associated with children’s chances of experiencing a firearm injury.

“Pediatric firearm injury is a public health issue that I feel very strongly about helping to address,” said the study’s lead author, Sonya Negriff, PhD, a research scientist in the Division of Behavioral Research in Research & Evaluation. “Our findings bolster confidence in known risk factors and could help target prevention efforts among the most vulnerable youth.”

A new endeavor for an experienced team

Most prior research on firearm injury had focused on fairly narrow sets of risk factors using data from small numbers of patients. Larger studies considering a broader constellation of risk factors had been lacking.

To help fill that gap, the team assessed electronic health record data from 1,889,182 children younger than 18 who had received any care at Kaiser Permanente Southern California from 2010 through 2018.

With this data, they investigated potential links between firearm injuries and age, gender, Medicaid status, mental health, substance use disorder, medical comorbidities, and race and ethnicity. Importantly, the study combined those individual-level factors with neighborhood-level data on educational achievement levels in each child’s surrounding residential environment.

“Our team had not worked with firearm injury data before,” said Margo Sidell, ScD, MSPH, the collaborative biostatistician research scientist and programming lead for the Division of Behavioral
Research at Research & Evaluation. “It was very important to us to take the time needed to validate our measures and ensure quality control of the data.”

“We couldn’t have done this work without Margo and our programmer Richard Contreras,” Dr. Negriff added. “They are essential for projects like this.”

Children’s risk factors revealed

The analysis showed that children ages 12 to 17 were at significantly higher risk of non-self-inflicted firearm injury than children ages 5 and younger. Being male, being Black or Hispanic (versus white), being a Medicaid recipient, living in a neighborhood with lower education levels, and having a substance use disorder were also linked to higher risk of non-self- inflicted firearm injury.

The only factors significantly associated with self- inflicted firearm injury were being male and being ages 12 to 17. However, Dr. Negriff cautioned, this does not contradict prior studies that have linked mental health to self-inflicted firearm injury.

“Self-inflicted firearm injury is quite rare, which limits our ability to detect statistical associations,” she said. “That is why it is important to accumulate evidence from various different studies like ours.”

These findings underscore the need to screen pediatric patients for risk of either self-inflicted or non-self-inflicted firearm injury and could help inform efforts to further reduce risk. For instance, in addition to individual screening, medical centers in higher-risk locations could provide services that address risk factors at a broader population level.

“It is important to consider the contextual environment in which non-self-inflicted firearm injuries occur and the systemic factors that contribute to environments where, for instance, kids may face higher risk of firearm injury from just walking to school,” Dr. Negriff said.

Building knowledge to inform action

Now, the team is finishing up a similar investigation of risk factors for firearm injuries in adults, as well as a study of trends in firearm injury rates over time with data updated through 2020.

They are also deepening their research with artificial intelligence. In 2022, Fagan Xie, PhD, a principal architect and manager of the Clinical Informatics team in Research & Evaluation, led the development of an algorithm that detects suicidal ideation and attempts in freely written clinical notes, ensuring such instances are counted even when not formally coded in people’s electronic health records.

This summer, the team will present a machine- learning-based model that uses more than 170 predictors in a patient’s record to assess potential risk of self-inflicted firearm injury. Construction of the model was led by Hui Zhou, PhD, MS, a collaborative biostatistician research scientist and the biostatistician lead in the Division of Epidemiologic Research at Research & Evaluation, in close collaboration with Claudia Nau, PhD, a research scientist in the Division of Behavioral Research.

“This model essentially creates a risk score that we are hoping can be adapted for use in the clinic,” Dr. Hechter said. “It could serve as a tool to help support more targeted screening.”

The importance of such tools was reinforced by the CDC’s recent announcement that, as of 2020, firearm injuries surpassed car accidents as the leading cause of death for U.S. children and adolescents.

“Firearm injury is a major component of children’s public health in the United States,” Dr. Hechter said. “Many people on our team have children ourselves, and we hope our work can help prevent some of the tragedies we hear about all too often.”

¹Negriff S et al. Acad Pediatr. 2023;23(3):604-609.