A latent class analysis of accidental polysubstance overdose deaths in Allegheny County, Pennsylvania

2020 
Background:The overdose mortality rate in Allegheny County has risen from 18.5/100,000 in 2008 to 63.3/100,000 in 2017. Despite most accidental drug poisoning deaths involving more than one drug, patterns in polysubstance overdose deaths are not well understood. Methods: Polysubstance overdose deaths were identified from Allegheny County Office of the Medical Examiner (ACOME) data. Drugs present at post-mortem toxicology were categorized using a modified list from the Council of State and Territorial Epidemiologists(CSTE) overdose analysis tool. Latent class analysis (LCA) of decedent-level factors including demographics and drug categories present at post-mortem toxicology were used to identify common patterns within polysubstance overdose deaths. Results: Among n = 3,749 toxicology reports from the Allegheny County Office ofthe Medical Examiner (ACOME) covering case years (cy) 2008 to 2018, n=2,864 involved more than one drug. LCA identified five latent class (LC) decedent groups associated with polysubstance overdose deaths: featured co-occurring fentanyl and fentanyl analogs (LC1); benzodiazepine presence with other opioid presence (LC2); ages 25-34 years and heroin presence (LC3); female sex, antidepressant presence, and amphetamine presence (LC4); and ages 45-54, black race, cocaine presence, alcohol presence, and fentanyl presence (LC5). LC1 was the largest class, representing 38.7% of all polysubstance overdose deaths. Conclusion: Among polysubstance overdose deaths, five sub-populations were identified. These findings match findings in the literature and suggest surveillance targets. Public health significance: This analysis is among the first to characterize accidental polysubstance overdose death in Allegheny County, PA. The composition of the sub-populations identified in this analysis will inform future overdose surveillance policy and outreach.
    • Correction
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    0
    Citations
    NaN
    KQI
    []