Abstract
Accurate attribution of heroin-related deaths, as well as the differentiation from other opioid analgesic-related deaths, is essential from a public health perspective. Heroin-related deaths involve a number of complexities where heroin-specific or non-specific metabolites and indicators (6-acetylmorphine [6-AM], morphine, and codeine) may or may not be detected. The aims of this study were therefore to develop a model for improved consistency in the attribution of heroin-related deaths and to determine areas of variation in the current decision-making processes.
A model was developed using different toxicological indicators of heroin use (6-AM, morphine to codeine ratio (M:C) or morphine alone) along with investigative evidence of heroin use (circumstances, scene and clinical findings) which were used to assign a weighted score. The combined scores for the toxicological and investigative evidence were used to determine the relative strength of association for the death being attributable to heroin according to three categories: suspected; likely; or strong. An expert panel was convened to validate the model and a series of test cases were provided to a cohort of forensic toxicologists and pathologists in order to identify sources of variation in decision-making within this group. The model was also evaluated for sensitivity and specificity by reviewing potential heroin-related cases and examining the evidence associated with the attribution of these cases to heroin or not.
Across all potential heroin-related death cases, the use of this model enabled a greater level of consistency in the attribution of death to heroin, especially in cases where 6-AM was not detected. The largest amount of variation in the attribution of a death to heroin was observed with potential intoxication-related deaths and in toxicity cases where a M:C ratio only was reported, even more than when no toxicological evidence was available. The reviewed cases highlighted the same variation in the attribution of a death to heroin, including a large number of cases that were attributed to morphine where 6-AM was not detected.
This model provides a useful tool for improved accuracy and consistency in the differentiation, attribution and reporting of heroin-related deaths. Previously challenging cases where death occurred after a significant period of time and either no 6-AM was detected or no samples were taken, are able to be captured using this model.