Researchers looked at data from the Centers for Disease Control and Prevention on about 630,000 people who died of drug overdoses between 1999 and 2016. They separated the deaths into two categories: those with and without a specific drug indicated.
For the first category, they analyzed how contributing causes of death (such as injuries and heart problems) and personal characteristics (such as age and gender) correlated with opioid involvement. They then used these analyses to calculate the probability of opioid involvement for each unidentified drug overdose, and they found that the number of opioid deaths is likely 28 percent higher than generally reported.
The researchers also noticed that in five states—Alabama, Indiana, Louisiana, Mississippi and Pennsylvania—the number of apparent opioid deaths over the seven-year period more than doubles after taking into account their adjustments.
“Opioid deaths serve as one of the main measures of the opioid crisis, and if opioid deaths are not counted accurately, the extent of the crisis can be severely misrepresented,” says [microeconomist] Elaine L. Hill… Hill says this research highlights “the potential role of state-level medical examination systems and other policies in driving high rates of underreporting.”
The researchers say they hope government officials and other researchers will use their new prediction model to calculate estimates for future deaths and to reexamine past data.