HEROIN DEATH RATES FAIL TO CAPTURE SEVERITY OF CRISIS
Flawed data-gathering makes finding solutions a daunting task.
In an epidemic, numbers matter. Laws, science and budgets can turn on records churned out by hospitals, medical examiners and treatment centers.
But even the best-known national drug databases are beset by blind spots and drawbacks, The Palm Beach Post found, skewing the true scope of the worst opioid epidemic in U.S. history.
Among them:
■ Heroin-related deaths are undercounted, and have been for years, even as a federal drug classification system has been overcounting the number of deaths linked to prescription opioids.
■ Just as the wave of prescription opioid deaths was peaking in 2011, and with a wave of heroin deaths emerging, a federal agency abruptly shut down a key means of tracking the scope of the crisis and has yet to replace it.
■ Federal drug death data used by researchers and policymakers is, in some parts of the country, collected by people with little to no medical experience in drug identification, or not compiled at all. No drug is listed on death certificates in at least one of every five overdose deaths in America.
■ Drug treatment and coroner reports are counted in different ways by states and federal agencies, often creating conflicting statistics.
When The Post calculated the national shift from prescription opioids to heroin beginning in 2011, it compared multiple sources of information to help offset weaknesses in any single set of numbers.
It isn’t that the various federal and state records on drug deaths don’t have important information. They all offer a window into which drugs are burning through the country.
But vulnerabilities are baked into the numbers from the start.
For instance, states are given flexibility to decide how they count admissions to drug treatment centers for one federal agency’s widely used database, called TEDS — and how quickly they provide it. The government’s figures are slow to update. For example, Florida had yet to provide full 2014 numbers by this year, federal records show.
The National Survey on Substance Abuse and Health, another crucial source, groups reports of treatment
for amphetamine abuse, including street meth, into the same category as treatment for prescription pills.
Then there’s WONDER — or Wide-ranging Online Data for Epidemiologic Research — the Centers for Disease Control and Prevention’s massive drug-death data set. It’s compiled from death certificates, many of them from local medical examiners or coroners.
WONDER would work best if all medical examiners and coroners used similar tests in similar ways to detect drugs found in overdose victims.
That doesn’t happen. Controversy over death determinations is not new. In 2011, when the popular “CSI” television series was spinning off variations on its theme of high-minded scientific forensic units, a joint investigation by ProPublica, Frontline and NPR shot holes in the idea that medical examiners uniformly employ cutting-edge science to reach unassailable conclusions.
But getting the numbers right is key to getting policy and treatment right.
“We had to get programs that we knew would work here,” said Jim Johnson, a veteran cop and former head of the Huntington, W.Va., Office of Drug Control Policy. “That’s why it was so important that we got up-to-date with the data to see what the problem was.
“If you don’t know what the problem is, how can you find a solution?”
Proving it’s heroin
Even basics like chemistry and drug classifications can add up to skewed drug death numbers.
Once in the body, heroin rapidly degrades into morphine. A person may have overdosed on heroin, yet lab results will show only morphine in their system.
WONDER elevates this to a national problem. It lumps morphine deaths into a broad category of prescription drugs, including oxycodone, but it counts heroin deaths separately. Because heroin deaths misclassified as morphine inflate the number of prescription opioid deaths, the number of deaths tied to oxycodone and similar opioids is overcounted while the number of heroin deaths are undercounted.
Tracking oxycodone-related deaths is important, too. Oxycodone is not the only widely abused prescription opioid. It’s the powerfully addictive drug most closely identified with the prescription opioid epidemic.
The CDC’s hands are largely tied. The WONDER categories reflect an international disease classification system created by the World Health Organization. And WHO works at a glacial pace: It took years to revise the classification system now in use.
Yet distinguishing deaths tied to prescription opioids and deaths tied to heroin is crucial. The current drug crisis is sometimes referred to as just one opioid epidemic. But heroin and prescription opioids are different drugs requiring different solutions, said epidemiologist Jon Zibbell, a senior public health analyst at research group RTI International who has tracked the fallout from drug epidemics for decades.
“Our response ... is different depending on the drug people are doing,” said Zibbell. “For example, we know how to shut off the supply of prescription opioids. We work with doctors on safer prescribing.
“But illicit drugs are another animal.”
Tracking a molecule
It can take some investigation to figure out whether an overdose was linked to heroin and not the result of prescription morphine. Was there heroin paraphernalia nearby, such as a burned spoon with heroin residue? Was there a history of heroin use?
Asked in 2016 how many of the morphine deaths his office handled in 2015 were actually tied to heroin, Palm Beach County Medical Examiner Dr. Michael Bell told The Post: “Almost all.”
Not every medical examiner or coroner investigates, though, or has the expertise, budget or staff to do so.
When they don’t, they can get an inaccurate picture of a drug problem.
In 2014, researchers decided to take a closer look at 112 morphine deaths in Allegheny County, Pa., home to Pittsburgh.
That many deaths tied to morphine, a prescription opioid, would suggest the county had a prescription pill problem.
In fact, it had a heroin problem: 74 of the 112 “morphine” deaths were heroin-related, researchers found.
Of 20 deaths where both heroin and morphine were listed on the death report, 19 of the people took heroin, not morphine.
The DEA’s footnotes confirm that researchers and public health officials acknowledge heroin deaths likely are being improperly categorized as morphine deaths. No one knows for certain how many.
However, there is evidence suggesting that the undercounting is not footnote-sized.
In 2011, the number of Floridians who died after using heroin grew from 58 to 62, the Florida Medical Examiners Commission found. Some considered the number too small to be significant.
But unlike the CDC, Florida has a category to specifically track morphine-related deaths. And the number of people who died the same year with morphine in their system shot up by 106 to 747.
Citing the heroin-to-morphine quandary, the medical examiners’ 2016 report cautioned that the failure to distinguish between the two could lead to “a substantial over-reporting of morphine-related deaths as well as significant under-reporting of heroin-related deaths.”
Indiana’s flawed test
In its 2012 study for the state Legislature, the Indiana Pain Society reported that the CDC had a “dizzying array” of metrics and definitions to count death rates, “and to make matter worse, the definitions have changed over time,” the frustrated authors wrote.
But the real problem was that Indiana coroners were categorizing causes of death using the wrong medical code.
At one point in Bartholomew County, nine of every 10 deaths were wrongly categorized, the report concluded.
The statewide result of wrong coding: a significant undercounting of how many people were dying from opioids in that region.
In 2013, the National Association of Medical Examiners wrote that some coroners and medical examiners resorted to simple urine drug tests known to return false results. The test had no place in determining a cause of death, the study’s authors wrote.
That’s in cases where coroners or medical examiners or doctors bother to list a drug at all: Between 19 percent and 22 percent of overdose death cases in 2013 and 2014 had no information about the drug on the death certificate.
Other researchers concluded the missing drug information profoundly distorted what was really going on. Heroin deaths in New Jersey and Louisiana were undercounted, for instance, even as overdoses from all opioids were overcounted in Connecticut and South Carolina.
Bad numbers can lead to bad conclusions. Indiana would have ranked 39th nationally in 2009 for the number of narcotic-involved deaths based on official drug death records. The numbers suggested Indiana was escaping the worst of the prescription opioid and heroin epidemics.
That was because in many cases, coroners were misclassifying drugs, the Pain Society report found.
The true prescription opioid death rate was much higher, the report concluded, and would have ranked Indiana 16th in the nation for opioid deaths, placing it among the hardest-hit states.
Autopsies optional
Similarly, in Arkansas, WONDER data from 2010 to 2016 indicated that the state inexplicably escaped the heroin and prescription pill epidemic: It recorded no more than 13 deaths a year from heroin and no more than 144 a year from the category of opioids that includes oxycodone.
But in Arkansas, there is no requirement that coroners perform autopsies on drug-related deaths. Requirements to be a coroner are bare-bones: A person needs only to be 18, a registered voter, have no felony record and a resident of the county where the coroner’s office is located.
“When you are talking about determining the manner of death, you would think someone would need to have some kind of a medical background, or investigative background,” said Kevin Cleghorn, president of the Arkansas Coroners Association.
Cleghorn is hopeful that a new reporting system offered to the state’s network of county coroners will provide better data more quickly. Some county coroners are already on board.
But Arkansas coroners are elected. As constitutional officers, they cannot be forced to use the new system. They cannot be forced to conduct autopsies on drug deaths. That would require changing the state constitution.
Other states’ coroner and medical examiner systems have their own challenges, including double duty. In Texas, justices of the peace are the coroners, according to the CDC. In Georgia towns of 5,000 or less, a person can be both the mayor and the coroner. In North Dakota, the sheriff, county manager or state attorney determines how someone died.
The big goof
Even the best state-supplied data can go awry.
The Florida Medical Examiners Commission is part of the Florida Department of Law Enforcement. Among the nation’s patchwork system of charting drugs linked to death, the commission’s annual report on drug fatalities stands head and shoulders above other states, and has for years.
So it was unexpected to find a significant error in the 2011 annual report, one which stood uncontested for six years until a Post analysis uncovered it.
It was a simple mistake: 2011 drug-related deaths in Palm Beach County for all drugs — heroin, oxycodone, cocaine and others — had been uploaded into a state computer system twice. The number of deaths linked to drugs was actually half of what was reported.
“There is no other way to say this,” said an FDLE official after The Post shared its findings. “We goofed.”
Prompted by the discovery, the commission re-ran several years of data to rule out other errors.
Other states publish drug death data, but using it to piece together a picture of what is happening nationwide is dicey. “Every state does it differently,” epidemioligist Zibbell said.
In 2013, Oklahoma reported the top 25 drugs detected in overdose deaths. Kentucky’s 2016 list of drugs found in overdose cases took seven pages.
In Colorado in 2016, the state’s drug monitoring agency reported it didn’t have access to toxicology reports in drug overdoses and so could not definitively report on the full number of drugs involved.
In Washington, the Department of Health calculated overdoses two different ways, as the state struggled to differentiate heroin and morphine overdoses. At least one calculation likely underestimated heroin-related deaths, the agency wrote.
Death of DAWN
Emergency room records also can spot the beginnings and scope of an epidemic but, in 2011, the federal Substance Abuse and Mental Health Services Administration, or SAMHSA, killed DAWN, the Drug Abuse Warning Network, a national system that continuously monitored drug-related visits to hospital emergency rooms.
It was a bad year to kill a tracking measure for a drug epidemic. In 2011, the prescription pill epidemic was claiming thousands of lives. A lethal transition from pills to heroin was emerging.
“The cost of the program was an issue,” said Phillip Walls, a SAMHSA spokesman.
DAWN was halted with the pledge that something better would soon replace it. But years passed, heroin and fentanyl overdoses began packing emergency rooms, and there was no replacement.
It’s only now, seven years later, that SAMHSA has received $10 million, not for a new surveillance tool but to bring back DAWN. “We are currently working through the contracting process,” Walls said.
SAMHSA, the American College of Toxicology and the National Institute of Justice have thrown their weight behind better and more consistent drug death investigations.
The CDC has launched a new, limited drug death system providing information faster than WONDER, although the numbers are only provisional.
For now, though, said James Hall, a nationally recognized epidemiologist with Nova Southeastern University, “All it really comes down to is that we have to go with the data that we have.”