Opioid-related overdoses could be 28 percent higher than reported because of incomplete death records, researchers found in a study published Thursday.

More than 400,000 people in the United States have died of opioid overdoses since the turn of the century, a quarter of them in just the past six years. But University of Rochester researchers found that between 1999 and 2016, about 100,000 more people died from opioids who were not accounted for — potentially obscuring the scope of the opioid epidemic and affecting funding for government programs intended to confront it, Elaine Hill, an economist and senior author of the study, told The Washington Post.

The discrepancies were most pronounced in several states, including Alabama, Mississippi, Pennsylvania, Louisiana and Indiana.

“We thought we would find underreporting, but we were definitely not prepared to find how spatially determined it is,” Hill said.

The federally funded study, published in the medical journal Addiction, used data on death rates per county and state from an agreement that Hill and co-authors Andrew Boslett and Alina Denham have with the Centers for Disease Control and Prevention. (The Post has the same arrangement with CDC.)

The researchers found that the records were least consistent in poorer communities. On average, the people whose records were not counted were white females in the 30 to 60 age range.

The incorrect records could be attributed to several factors, Hill said. Limited resources in counties can delay toxicology reports, limit drug testing and even prevent the completion of autopsies.

Counties can opt to appoint medical examiners, while others elect coroners. Coroner positions do not require professional training, whereas medical examiners are appointed and have board certification in a medical specialty, according to the National Academy of Medicine.

Hill said the team is not the first to notice how death records are not always well-kept in the United States.

In the midst of the researchers’ year-long look at death data, comedian John Oliver aired a segment on coroners and medical examiners on his late-night HBO program. Oliver argued that coroners were too inexperienced to deal with the influx of bodies including overdose victims, citing an Atlanta Journal-Constitution investigation that found coroners who were grocers, farmers, handymen and hairdressers.

“In an ideal world, no one would die. I’m certainly never going to, and at the risk of sounding too prescriptive, I don’t think you should, either,” Oliver joked. “But in a slightly less ideal world, we’d phase out coroner’s offices and replace them with medical examiners.”

Some coroners may not even rule overdose deaths as drug-related, said Stanford University professor Keith Humphreys, citing a USA Today investigation in 2015 that found at least one Tennessee county that reported overdose deaths as caused by “natural causes.”

“Every coroner is a king and a queen with how they report overdose deaths,” Humphreys said in an interview.

Humphreys said that previous studies have found deviations in reporting but that researchers are still working on quantifying the consequences of those disparities.

In addition to varied investigative methods, inaccurate data can also come from the stigma associated with drug use, Hill said. Grieving families can decline an autopsy.

Officials tracking the crisis have also faced difficulties, as the type of opioids causing the most deaths has changed over the years.

Beginning in 2013, when fentanyl, a synthetic opioid, was increasingly found in illicit drugs, not all death investigators were looking for it in toxicology testing, The Post reported.

“It’s incredibly frustrating when you feel like, given the resources of the United States, that we can’t harness those resources in ways that help us see around the corner,” Michael Botticelli, the White House drug czar in the Obama administration, told The Post.

Hill and Boslett first stumbled upon the discrepancy while studying the economic, environmental and health impacts of shale gas development and coal mining. Counties with natural resource development were among the most affected by the opioid crisis, Hill said.

The team of researchers hopes to next look at county-by-county data in reporting opioid deaths, but they face barriers. For one, there’s no centralized place for what type of death investigations each county does.

“The question we’re talking about hasn’t been asked yet,” Boslett said.

Hill said they will publish the corrected data on her website in the near future for scientists or anyone else who is interested.

“We’re hoping that this paper contributes to the reality that we do need to improve reporting methods for target areas that are in the most need,” Hill said. “Just the sheer volume in numbers in our paper shows this is a really large and national issue that we hope attention is brought to.”

Read more: