Opioids may be killing thousands more people each year than we previously thought

Opioids have already spawned the deadliest drug overdose crisis in US history — but they may be killing even more people than we think.

According to a new study published in the American Journal of Preventive Medicine, actual opioid deaths for 2014 were 24 percent higher than previously estimated. When pulled out alone, heroin deaths were also 22 percent higher than previously estimated. That would bump up the total number of all opioid overdose deaths in 2014 from nearly 29,000 to more than 35,000.

Even more alarming was how much some specific states under-counted their opioid overdose death rates. For example, the corrected opioid death rates in Pennsylvania, Indiana, Louisiana, Alabama, and Mississippi were more than 100 percent higher than each state originally estimated. Most New England states generally reported accurate rates.

The study helps explain why some of the geographic patterns you would expect to see in some states don’t show up with the official rates, since it turns out that some states are just not reporting the full figures.

Researcher Christopher Ruhm of the University of Virginia created several maps to show just how much underreporting is going on in some states. On the left, you see the maps for the official 2014 death rates for all opioids and just heroin. On the right, you see the corrected death rates. The maps on the right make it much clearer that the opioid epidemic is following geographic patterns — suddenly showing more distinct trends in the Midwest, South, and West than previously understood.

This has important implications for policy. Pennsylvania, for example, looks much worse off in reality than it did under the official numbers. The corrected figures, then, could lead policymakers to perhaps consider that the state has a more urgent opioid crisis to tackle than previously believed. And that could lead to a boost in opioid-specific policy interventions, such as a boost in medication-assisted treatment or harm reduction efforts.

The poor data could also make it difficult to identify common characteristics among the worst-off states, which can in turn make it harder to address the root causes of drug addiction and overdoses to avoid future, similar crises.

The study looked at the death certificates used to calculate the official statistics. It didn’t identify more total drug overdose deaths than previously estimated for 2014, which totaled more than 47,000. Instead, it looked at the existing count of poisoning deaths to see when opioids weren’t correctly marked as the cause of death.

Typically, deaths are marked by local coroners or medical examiners through a system made up of codes from the International Classification of Diseases, 10th Edition (ICD-10). If a medical examiner marks a death as immediately caused by an opioid overdose through the proper ICD-10 codes, that death is eventually added to the US’s total for opioid overdose deaths.

But there’s no national standard for what counts as an opioid overdose, so it’s left to local medical officials to decide whether a death was caused by an opioid overdose or not. This can get tricky when officials lack the proper tools to diagnose some opioid overdoses, rely on different standards, and look at cases in which multiple conditions can mask the true cause of death.

Ruhm tried to get through this hurdle by combing through death certificates to estimate how many were likely underestimating opioid deaths. Since they’re estimates and the death certificates likely contained some errors that Ruhm couldn’t correct for, he acknowledges that his numbers aren’t perfect — but they’re likely much more accurate than the official numbers we have today.

And they tell us that an already grim story is even worse than we thought.

For more on the opioid epidemic, read Vox’s explainer.