Sensitivity of Death Certificate Data for Unintentional Drug Poisoning Deaths By Drug Type in Three Jurisdictions: Can It be Improved?

BACKGROUND: As drug overdoses have increased, state and local health departments have increased tracking drug overdose deaths using vital statistics data. Unfortunately, the quality of death certificate data for drug overdoses is not consistent across the U.S and in many cases lacks specificity. Heroin-involved overdose deaths, in particular, due to the complexity of morphine metabolites are often under-reported when relying on death certificate data. However, specific drug and metabolite information is often available in medical examiner and coroner toxicology reports. Participating health departments compared death certificate multiple cause codes with reported toxicology findings.

METHODS: New York City (NYC), Oregon, and King County in Washington compared toxicology data to multiple cause ICD 10 codes on death certificate for those who died in 2012 of an unintentional drug overdose. Six drug categories including benzodiazepines, cocaine, heroin, methadone, other opioid analgesics, and methamphetamines were compared. NYC did not report on methamphetamines due to low prevalence. Cases were limited to decedents 15–84 years of age that had both death certificate and toxicology data. The sensitivity of death certificates accurately reporting drug-specific deaths was calculated using toxicology data as the true positive. Positive toxicology results for ‘morphine’ was used an indicator for heroin and other opioid analgesic deaths depending on the jurisdiction and literal text description on the death certificate.

RESULTS: Among the three jurisdictions, drug-specific sensitivity on death certificates varied most widely for benzodiazepines (NYC 68%, Oregon 23%, and King County-Washington 88%). Variation continued for heroin (NYC 61%, Oregon 82%, and King County-Washington 92%), and cocaine (71% for both NYC and Oregon, with King County-Washington 95%). In contrast, drug-specific sensitivity for opioid analgesics was high (NYC 91%, Oregon and King County-Washington both 100%). Methadone-specific sensitivity was also over 80% (NYC 82%, Oregon 91%, and King County-Washington 100%). Oregon and King County-Washington reported high sensitivities for methamphetamines (90% and 97%, respectively).

CONCLUSIONS:  Using death certificates alone may result in underreporting of drug-specific mortality rates. Sensitivity varies by drug and by jurisdiction. Benzodiazepines and heroin had the lowest sensitivities. Toxicology data can provide more complete information about drug specific unintentional overdose deaths. For deaths listing morphine on the death certificate, medical examiners and coroners often have information about the source of the morphine. Standardization guidelines for medical examiners and coroners for drug overdose deaths could reduce variation and improve drug overdose surveillance. Results were used in several jurisdictions to improve the information on future death certificates.