METHODS: Two high-rate, high-population counties were selected. Vital statistics reported 72 unintentional overdose deaths from County A and 36 from County B (n=108) in 2014. Experienced NVDRS abstractors contacted the ME in each county and arranged to abstract those cases on-site or from copies of records. Police reports were not reviewed in this pilot. Abstractors shared data review and entry duties.
RESULTS: The overdoses were related to substance abuse in 65% of the cases (n=70), overmedication 19% (n=20) and unknown or missing 16% (n=17). No previous overdose was reported in 92% (n=99). 91% (n=98) of the cases reported no current or history of substance abuse treatment. An abuse history of prescription opioids, heroin or both was reported in 17% (n=18), 29% (n=31), and 8% (n=9), respectively. Naloxone was not administered or mentioned in 81% (n=87). No bystanders were present 54% (n=58) of the time. Prescriptions for opioids were poorly documented; the number of opioid prescriptions in the 30 days preceding injury was unknown in two-thirds of the cases.
CONCLUSIONS: Overall completeness and interpretation issues relegated about half of the questions lacking in usefulness. For other potentially useful variables, many definitions were unclear or difficult to pigeonhole among variable value choices. Several questions lacked an “unknown” response item leaving interpretation in doubt. Skip logic in the web entry system prohibited access to some victim characteristic questions in the suicide section that would have been useful. The weaknesses are substantial enough that we cannot recommend use of the optional module without substantial revision. Linking PDMP to death data remains an important alternative for connecting prescription drug use history to overdose cases. Crucial questions about unintentional overdoses requires more coordination with MEs and coroners or resources for “shoe leather” epidemiology using medical record reviews and family interviews to uncover crucial information.