The study, led by Danta D. R. Bien-Aimé and colleagues, was guided by the Andersen Behavioral Model for Vulnerable Populations, a framework that organizes potential drivers of healthcare use into predisposing characteristics, enabling resources such as insurance and neighborhood support, and need factors such as co-occurring diagnoses. The researchers estimated Firth bias-reduced logistic regression models to identify factors associated with any primary care, behavioral health, or emergency department use documented at any time during the study window, supplemented by a negative binomial regression modeling the count of emergency department visits. Firth’s method is a bias-reduction technique suited to small samples and rare outcomes, an appropriate choice given the modest cohort size.

The most consistent signals emerged from clinical need factors. Anxiety disorders and non-OUD substance use disorders were associated with higher odds of both primary care and behavioral health use, suggesting that men carrying additional psychiatric and substance use burdens were more likely to connect with routine and specialty services. Neighborhood-level public assistance was associated with higher odds of primary care use, hinting that community socioeconomic context may shape access to preventive care even in a safety-net system designed to reduce financial barriers. Most other measured predisposing and enabling factors, including demographic and insurance characteristics, were not significantly associated with service use.

When emergency department use was modeled as a count rather than a binary outcome, depression and non-OUD substance use disorders were associated with higher emergency department visit rates. This distinction matters methodologically: a binary model captures whether someone ever used the emergency department, while a count model captures intensity of use, and the two approaches surfaced overlapping but not identical need factors. The findings suggest that psychiatric comorbidity and polysubstance use mark men who cycle through acute care most frequently, even as the cross-sectional design prevents conclusions about whether these conditions cause higher utilization or result from it.

A striking negative finding ran through the results: documented receipt of medications for opioid use disorder (MOUD) was low across the cohort. MOUD is a cornerstone of overdose prevention, and men with criminal-legal involvement face elevated overdose risk alongside substantial barriers to initiating and sustaining treatment. The low documented receipt in this safety-net sample aligns with the study’s framing of this intersection as understudied and underserved, although the measure reflects documented receipt within the linked records rather than any treatment obtained outside the system.

The linked-data design is a notable strength. By combining electronic health records with arrest and jail records under HIPAA-compliant de-identification procedures, the researchers assembled a picture of a population often invisible to analyses relying on a single data source. Men with OUD who move through jails, courts, and community supervision churn across institutions, and their care episodes fragment accordingly. Linking correctional and healthcare records captured utilization patterns that either system alone would miss, while neighborhood-level data added a geographic dimension to the individual-level clinical picture.

Nevertheless, limitations shape how the findings should be read. The sample comprised 159 men served by a single urban safety-net healthcare system, so results may not generalize to women, rural populations, or regions with different safety-net structures. The cross-sectional design captures associations rather than trajectories, binary outcomes reflect any documented use across the full window rather than timing or continuity of care, and unmeasured factors such as housing instability or social support could account for some observed associations. The modest sample size limits statistical power, which the Firth correction partially addresses but cannot eliminate.

Within those constraints, the central takeaway is that behavioral health and substance use-related need factors were more consistently associated with service utilization than demographic, insurance, or most neighborhood-level factors. That pattern carries practical implications for safety-net systems: screening for co-occurring anxiety, depression, and non-OUD substance use may help identify men at highest risk of high-cost emergency department reliance, while the near-universal emergency department contact rate positions those settings as potential engagement points for OUD treatment initiation. The authors’ observation that healthcare utilization was common, particularly emergency department use, frames the emergency department less as a failure of the system and more as the de facto front door through which this high-risk population already enters care. Future longitudinal work with larger, more diverse samples will be needed to establish causal pathways and to test whether connecting men with OUD and CLI to MOUD and behavioral health services can shift care away from emergency departments and toward settings where overdose risk can be addressed directly.

Subject of Research: Healthcare utilization patterns among men with opioid use disorder and criminal-legal involvement

Article Title: Factors associated with primary care, behavioral health, and emergency department utilization among men with opioid use disorder and criminal-legal involvement: a cross-sectional study

Article References: Bien-Aimé, D. D. R., Mazen, S., Bitere, C., Parry, G., Progovac, A. M., & Flores, M. W. (2026). Factors associated with primary care, behavioral health, and emergency department utilization among men with opioid use disorder and criminal-legal involvement: a cross-sectional study. BMC Health Services Research. https://doi.org/10.1186/s12913-026-15514-x

Image Credits: AI Generated

DOI: 10.1186/s12913-026-15514-x

Keywords: opioid use disorder, criminal-legal involvement, emergency department, primary care, behavioral health, safety-net healthcare, MOUD, cross-sectional study, health services research, substance use disorders, Andersen Behavioral Model, overdose risk