Social Determinants of Health Impact Suicide Rates Among Veterans Compared to Nonveterans Over the Past Decade

Over the past decade, suicide rates have consistently been higher among veterans compared to nonveterans, and the rate among veterans has been increasing at a faster pace than among nonveterans, according to a study published in JAMA Network Open.

Social determinants of health (SDOHs), encompassing factors like socioeconomic status, access to healthy food, education, housing, and physical environment, play a significant role in predicting suicidal behaviors, including ideation, attempts, and death. Social disruptions, such as relationship breakdowns, financial instability, legal problems, and childhood adversity, trigger suicidal behavior. However, comprehensive SDOH information has been lacking in large population-based databases, with structured data primarily used for analysis. Study results revealed that unstructured data contain approximately 90 times more SDOH information than structured data.

Researchers sought to explore the association between 9 SDOH factors and the risk of suicide using both structured data (ICD codes, stop codes) and unstructured data (clinical notes processed through natural language processing) from the US Veterans Health Administration's Electronic Health Records (EHR) system. The study employed a nested case-control design and risk-set sampling to match controls to cases, facilitating the study of associations between SDOHs and suicide risk in a rare event context.

Base cohort included all veterans with VHA records between October 1, 2010, and September 30, 2015. Cases comprised veterans who died by suicide during this period, matched to living controls. The study used a multitask learning framework based on RoBERTa, a pretrained language model, to extract 13 factors, including eight SDOHs, from clinical notes. The analysis considered exposures to these factors in the two years preceding the index date.

Results showed that SDOHs were strongly associated with an increased risk of suicide. Factors like legal problems, violence, and social isolation had significant estimated effect sizes, with odds ratios exceeding 2. Combining both structured and unstructured SDOH data yielded similar results. Combinations of SDOHs were also associated with an increased risk of suicide.

Researchers highlighted the importance of addressing SDOHs in suicide prevention efforts and the potential of natural language processing to extract valuable information from clinical notes for risk assessment. It also emphasized the need for further research to integrate SDOH information into health care systems seamlessly. However, the study acknowledged limitations, including the representativeness of the VA population and potential residual confounding in the analysis.

"We strongly believe that analyzing all available SDOH information, including those contained in clinical notes, can help develop a better system for risk assessment and suicide prevention. However, more studies are required to investigate ways of seamlessly incorporating SDOHs into existing health care systems," said researchers.

Reference

Mitra A, Pradhan R, Melamed RD, et al. Associations between natural language processing–enriched social determinants of health and suicide death among US Veterans. JAMA Network Open. 2023;6(3):e233079-e233079. doi:10.1001/jamanetworkopen.2023.3079