Christopher J. Ruhm, PhD, of the University of Virginia, Charlottesville, conducted an analysis using improved estimates of state-level opioid and heroin overdose fatalities during 2014 and of changes over time during 2008-2014. The analysis suggests that, in 2014, national opioid mortality rates were actually 24% greater than reported and that heroin overdose fatality rates were 22% higher than reported (Am J Prev Med. 2017 Aug 7.).
To correct for this, Dr. Ruhm extrapolated the drug category in these unspecified cases by using information from death certificate reports where at least one specific drug category was identified as the cause of death. The variables considered included sex, ethnicity, marital status, education, day of the week indicators, location of death (such as hospital inpatient or ED) or arriving at the hospital already deceased, and interactions among these variables.
“The corrected rates provide the best currently available information on geographic variation in opioid and heroin involved fatality rates,” Dr. Ruhm noted.
The differences between corrected and reported rates of fatal opioid and heroin overdose varied significantly between states; in some cases, the correct rates were more than double the reported rates.
In Pennsylvania, for example, the corrected rates of opioid and heroin overdose fatalities were 108% and 107% greater than the reported rates, respectively; in Mississippi, the corrected rates were 107% and 139% higher than reported. Alabama, Indiana, and Louisiana also showed large disparities between corrected and reported rates.
Dr. Ruhm also noted significant variability between the states in whether a specific drug was mentioned on the death certificate or not. In Rhode Island, Connecticut, and New Hampshire, a drug category was mentioned for more than 99% of drug poisoning deaths; but in Pennsylvania, Indiana, Mississippi, Louisiana, and Alabama, this was reported in only about half of these incidents.
There was a general trend toward underreporting in the South, while rates of nonreporting were lower in parts of the Northeast and the West.
Dr. Ruhm suggested that additional training and standardization could be helpful in states with low specification rates, “particularly because this is a bigger problem when death certificates are completed by coroners rather than medical examiners and in states without centralized oversight.”
Overall, the growth in opioid-involved drug deaths from 2008 to 2014 was similar between the reported and corrected rates. However, in the case of heroin-related mortality, the increase was underestimated by the reported data by about 18%. Again, this difference varied between states, with substantial underestimates seen in Pennsylvania, Indiana, New Jersey, Louisiana, and Alabama.
“Understanding the inaccuracies resulting from the lack of specificity of drug involvement on death certificates is also important because federal policies often target states believed to have especially severe opioid or heroin problems,” Dr. Ruhm explained. “More fundamentally, geographic disparities in drug poisoning deaths are substantial, and a correct assessment of them is almost certainly a prerequisite for designing policies to address the fatal drug epidemic.”
Dr. Ruhm declared no conflicts of interest.