Variable Classification of Drug-Intoxication Suicides across US States: A Partial Artifact of Forensics?

BACKGROUND The 21st-century epidemic of pharmaceutical and other drug-intoxication deaths in the United States (US) has likely precipitated an increase in misclassified, undercounted suicides. Drug-intoxication suicides are highly prone to be misclassified as accident or undetermined. Misclassification adversely impacts suicide and other injury mortality surveillance, etiologic understanding, prevention, and hence clinical and public health policy formation and practice. OBJECTIVE To evaluate whether observed variation in the relative magnitude of drug-intoxication suicides across US states is a partial artifact of the scope and quality of toxicological testing and type of medicolegal death investigation system. METHODS This was a national, state-based, ecological study of 111,583 drug-intoxication fatalities, whose manner of death was suicide, accident, or undetermined. The proportion of (nonhomicide) drug-intoxication deaths classified by medical examiners and coroners as suicide was analyzed relative to the proportion of death certificates citing one or more specific drugs and two types of state death investigation systems. Our model incorporated five sociodemographic covariates. Data covered the period 2008-2010, and derived from NCHS's Multiple Cause-of-Death public use files. RESULTS Across states, the proportion of drug-intoxication suicides ranged from 0.058 in Louisiana to 0.286 in South Dakota and the rate from 1 per 100,000 population in North Dakota to 4 in New Mexico. There was a low correlation between combined accident and undetermined drug-intoxication death rates and corresponding suicide rates (Spearman's rho = 0.38; p<0.01). Citation of 1 or more specific drugs on the death certificate was positively associated with the relative odds of a state classifying a nonhomicide drug-intoxication death as suicide rather than accident or undetermined, adjusting for region and type of state death investigation system (odds ratio, 1.062; 95% CI,1.016-1.110). Region, too, was a significant predictor. Relative to the South, a 10% increase in drug citation was associated with 43% (95% CI,11%-83%), 41% (95% CI,7%-85%), and 33% (95% CI,1%-76%) higher odds of a suicide classification in the West, Midwest, and Northeast, respectively. CONCLUSION Large interstate variation in the relative magnitude of nonhomicide drug-intoxication deaths classified as suicide by medical examiners and coroners in the US appears partially an artifact of geographic region and degree of toxicological assessment in the case ascertainment process. Etiologic understanding and prevention of drug-induced suicides and other drug-intoxication deaths first require rigorous standardization involving accurate concepts, definitions, and case ascertainment.