Background & Aims

The opioid epidemic continues to represent a significant threat to the public health. Over the past decade, opioid overdose deaths in Illinois have tripled, with 3,013 deaths in 2021. The increasing role of synthetic opioids impacts this trend from both prescription and illicit drug supply. This study sought to establish predictive models for opioid-related overdoses in which to develop a patient overdose risk.

Methods

Surveillance data from the IL Department of Public Health and prescription drug monitoring program data from the IL Department of Human Services from 2019-2022 were merged, matched, and redacted before delivery to investigators. Specific datasets included emergency department (ED), hospital admissions, opioid prescriptions, and overdose fatalities. ICD-10 codes were binned into broad categories. Investigators utilized Python 3.12.1 for Windows (Python Software Foundation, Wilmington, DE, USA). Variables were selected based on the Risk Index for Overdose and Serious Opioid-induced Respiratory Depression (RIOSORD) screening tool. All variables included in the RIOSORD were not available for analysis. The sensitivity and specificity of sampled models were assessed using logistic regression, decision tree fitting, support vector machine methods, and multilayer perception (neural net) methodology.

Results

Following cleaning and data preparation, 129,135 unique patient cases were identified. All cases involved either ED presentation or hospital admission for opioid overdose. A total of 9535 (7.4%) opioid overdose deaths occurred. A modified RIOSORD model (excluding prescription information on benzodiazepines and antidepressants) in combination with a history of leaving the emergency department or hospital before discharge (against medical advice) via decision tree analysis yielded the highest predictive accuracy (97.3%) with a sensitivity of 0.1% and a specificity of 99.9%. This model additionally demonstrated high predictive accuracy (84%) for repeat overdose (sensitivity 55.6%, specificity 91.7%).

Conclusions

Attempts to identify those at the highest risk for opioid-related mortality continue. This modeling provides valuable public health predictive risk estimates to allow for early intervention when patients experiencing opioid overdose present for care at hospitals in Illinois.

References

1.Zedler BK, Saunders WB, Joyce AR, Vick CC, Murrelle EL. Validation of a screening risk index for serious prescription opioid-induced respiratory depression or overdose in a US commercial health plan claims database.

Presenting Author

Chris Herndon

Poster Authors

Christopher Herndon

PharmD

Southern Illinois University Edwardsville

Lead Author

E-mail

Carrie Butts-Wilmsmeyer

PhD

Southern Illinois University Edwardsville

Lead Author

E-mail

Stacey Hoferka Jensen

MPH

Illinois Department of Public Health

Lead Author

E-mail

Dejan Jovanov

MPH

Illinois Department of Public Health

Lead Author

E-mail

Topics

  • Epidemiology