℞ for an Epidemic

An unprecedented surge of opioid prescriptions launched an epidemic of substance use disorder in the United States. Harvard Chan School researchers are piecing together how it happened—and how to stop it from happening again.

By Chris Sweeney

The statistic is shocking: For every 100 people living in Alabama’s Fourth Congressional District in 2016, there were 166 opioid prescriptions. “It’s staggering,” says state Senator Andrew Jones, whose constituents are largely based in the district. “When you tell most people, there’s a look of disbelief on their face.”

Like many Americans, Jones, a Republican who was born and raised in Alabama and elected to state office in 2018, was well aware of the lethal toll that highly addictive prescription opioids had inflicted across the nation. He had followed the news about soaring overdose rates and heard wrenching stories of families losing loved ones. But it wasn’t until he came across a recent study by a trio of Harvard T.H. Chan School of Public Health researchers that he grasped the extent to which physicians in his own backyard were fueling the crisis.

The study—led by SV Subramanian, professor of population health and geography—analyzed and compared the rate of opioid prescribing in each of the nation’s 436 congressional districts. It was the first time these data were mapped into congressional boundaries, and the findings showed that Alabama’s Fourth had the highest prescribing rate in the nation, with districts in Arkansas, Kentucky, Mississippi, Oklahoma, Tennessee, and Virginia trailing close behind.

A Map with Policy Implications

Congressional districts don’t always translate neatly into the field of public health, where data are often collected and analyzed at county and city levels. The outlines of a congressional district can resemble a Rorschach test, and it’s not unusual for a single district to spill across multiple counties, cities, and towns. Alabama’s Fourth, for instance, cuts across more than a dozen counties and spans the state’s northern width, from Fort Payne in the east to Red Bay in the west.

Yet from a health policy perspective, congressional districts can be enormously influential. They’re an important link between the local, state, and federal instruments of government, and they’re where the work of policymakers and policy implementers intersects.

Subramanian has long been intrigued by the idea of aligning health indicators with political districts. It’s a way, he says, to engage politicians on an array of health issues that directly affect the constituents they’re elected to represent. “For all my career, I’ve wondered why in the U.S. and India—the countries I’ve had the most exposure to—so much of the data used in public policy, such as life expectancy and poverty levels, are never reported at the geographical unit that’s represented by an elected official,” he says. “Even in a very mature democratic country like the U.S., where you have tons of data down to ZIP codes and census tracts, health indicators are never presented at the congressional district level.”

While data may be plentiful, actually fitting them into the unshapely geography of congressional districts is complex, tedious work. The idea rattled around for years in the back of Subramanian’s head, but time and funding never worked in his favor. “I knew I had to wait for the right opportunity,” he says, “and the right set of students to work with.”

In 2017, Subramanian met a then-SM student named Jack Cordes, who had studied geography as an undergraduate at the University of North Carolina (UNC) at Chapel Hill. For his capstone work at UNC, Cordes analyzed the geography of opioid-related mortality in the state by drug type. He compared 15 years of data on overdoses associated with prescription opioids and those associated with street heroin. Cordes found that the prescription overdoses were heavily concentrated in the rural western part of North Carolina, while heroin overdoses were more concentrated in the urbanized eastern areas. That was expected, he says. What wasn’t expected was the way in which prescription opioid overdoses became more evenly spread throughout the state as the years went by. “That was my first foray into studying the opioid epidemic,” Cordes says, “and it was really interesting to see how different drugs had different spatial patterns in terms of mortality across the state and across time.”

Subramanian recognized that Cordes had a continuing interest in substance use disorders and the technical chops for the nitty-gritty data science required to map a health indicator by congressional districts. Subramanian also turned to the expertise of Lyndsey Rolheiser, who was then a postdoctoral fellow at the Harvard Center for Population and Development Studies and is now an assistant professor in the department of urban planning at Ryerson University. The three agreed that homing in on opioid prescribing rates was a worthwhile endeavor—not only because it was an urgent public health issue, but also because there were robust data from the U.S. Centers for Disease Control and Prevention (CDC).

SV Subramanian, professor of population health and geography (left), and Jack Cordes, SM ’19, innovatively mapped the opioid epidemic by congressional district in the United States—sparking conversations and renewed concern among elected officials.
SV Subramanian, professor of population health and geography (left), and Jack Cordes, SM ’19, innovatively mapped the opioid epidemic by congressional district in the United States—sparking conversations and renewed concern among elected officials.

The trio published the study in July 2018 in the American Journal of Public Health, and it made a splash with the media. The popular political website Axios turned the findings into an interactive map, while a reporter from U.S. News & World Report sought a response from legislators in high-prescribing districts. Just as Subramanian anticipated, overlaying the data onto political districts sparked conversations and renewed concern among elected officials, including Jones.

“I always knew it was a serious problem here,” Jones says. “But to learn that we were ground zero for prescribing rates was eye-opening.”

A Singular Epidemic

This isn’t the country’s first opiate epidemic. Waves of opium, morphine, and heroin use can be found throughout the 19th and 20th centuries. But, says Mary T. Bassett, director of the Harvard François-Xavier Bagnoud (FXB) Center for Health and Human Rights and FXB Professor of the Practice of Health and Human Rights in the Department of Social and Behavioral Sciences, this one is unusual and poses distinct public health challenges.

“This is a typical epidemic in the sense that it’s on a broad scale and persistent. So we know that we can apply public health interventions to help prevent, mitigate, and save lives,” Bassett says. “But what’s unusual with this epidemic is that the doorway to addictive drugs was thrown open by health workers writing legal prescriptions. It’s clear now that the enormous increase in prescription painkillers was encouraged by the pharmaceutical industry.”

Volumes have been written by now on the pharmaceutical industry’s culpability in the crisis. Purdue Pharma’s launch of OxyContin (oxycodone) in the late 1990s, and its subsequent marketing efforts to brand the drug as a safe, nonaddictive pain treatment, kick-started the trend. In the years following, other drugmakers introduced similarly powerful and addictive pain medications and spent heavily promoting these products to doctors through advertisements, conference presentations, and “detailing,” the process of sending sales reps to doctors’ offices day in and day out for one-on-one meetings. The industry’s efforts paid off, and throughout the early 2000s, the number of prescriptions for opioids climbed higher and higher year after year. By 2006, there were 72.4 opioid prescriptions for every 100 people in the United States. By 2010, that figure climbed to 81.2 prescriptions.

Part of the problem was the rise of so-called “pill mills,” storefront clinics where spurious doctors doled out opioid prescriptions to anyone who walked through the door. In one case, a physician in Tennessee is alleged to have prescribed some 500,000 hydrocodone pills, 300,000 oxycodone pills, and 1,500 fentanyl patches over the course of three years. In another instance, a network of three clinics in Jacksonville, Florida, wrote out prescriptions totaling 3.2 million opioid pills in a six-month span.

But there were also countless well-intentioned physicians who, following their medical training, prescribed opioids for patients because they were taught that it was the best way to manage unabated pain. “When I started medical school in 2011, it was at the height of this movement toward thinking of chronic opioid therapy as the compassionate response to patients with chronic pain. Patients who had poorly managed pain were seen as being opioid deficient,” says Michael Barnett, assistant professor of health policy and management. “That was something I internalized. I thought that opioid therapy was underused and that we shouldn’t hold it back from suffering patients.”

Barnett’s perspective shifted when he started his residency and began encountering patients who were on chronic opioid therapy. They were consistently some of the most challenging patients to manage because their pain was often not well-controlled despite the opioid therapy. Many had complex mental health needs, and many were somewhere on the spectrum of opioid use disorder—an official psychiatric diagnosis that includes physical and psychological addiction. “I began to wonder whether chronic opioid therapy was really helping people,” Barnett says, “because it seemed to create so many problems for so many of the patients on it.”

Barnett wasn’t the only medical professional worried about opioids. But changes in clinical practice come slowly, and the drugs continued to saturate every corner of the country through legal and illegal pathways. According to evidence recently presented by the U.S. Drug Enforcement Administration, a handful of pharmaceutical companies pushed 76 billion oxycodone and hydrocodone pain pills into the U.S. between 2006 and 2012. During the same period, there were nearly 100,000 fatal overdoses associated with prescription opioids.

The severity of the epidemic became impossible to ignore, and everyone from police chiefs to academics to activists was working on ways to stem the flow of prescriptions. Some states launched prescription-drug databases. Others imposed strict prescribing rates. “In New York City, we took a page from the pharmaceutical industry and did what we called ‘public health detailing,’” explains Bassett, who was commissioner of the New York City Department of Health and Mental Hygiene before joining the FXB Center. “We sent people out to talk with physicians and health care providers about the risks of opioids and encourage judicious prescribing. It’s the same strategy the Sackler family, which owns Purdue Pharma, and all the other companies used to sell opioids.”

Barnett focused his efforts on studying how doctors prescribed opioids. “There was a lot that we didn’t know about how opioids were used, and there were a lot of assumptions among doctors about how opioid dependence develops,” he says. “One of the assumptions that stood out and still persists is that if you give someone a short-term prescription for opioids to manage acute pain, addiction won’t develop. That’s what led me to my first major study on opioid prescribing.”

In that study, published in 2017 in the New England Journal of Medicine, Barnett and colleagues examined data on opioid-prescribing habits of emergency physicians from 2008 through 2011. Because patients don’t choose their emergency physicians, they were naturally randomized to doctors with a wide variety of prescribing habits. Barnett compared two groups of patients. The first included 215,678 patients who received treatment from doctors considered low-intensity opioid prescribers. The second group included 161,951 patients who received treatment from high-intensity prescribers. The findings were clear as daylight: Patients of high-intensity prescribers were significantly more likely to become long-term opioid users.

“We saw that any opioid prescription, even if it’s short and one-time, carries a real risk that patients might be on these drugs for several months or more,” Barnett says. “It’s not a side effect to minimize.”

After years of steady growth, the surge of opioid prescribing began to abate. In 2013, the prescribing rate dipped from 81.3 prescriptions per 100 persons in the U.S. to 78.1 prescriptions. In 2014, it dropped to 75.6 per 100 persons and continued inching its way down. By 2017, the rate was 58.7 per 100 persons, the lowest it had been in more than a decade, according to the CDC.

Astonishing Variability in Prescribing Rates

While the drop in prescribing is welcome news, progress hasn’t been consistent. When mapping prescribing rates by congressional districts, Subramanian and his team used 2016 data—a period when rates were well into their national decline. In many ways, the results confirmed what was known: Prescription opioids had insinuated themselves into every part of the country, but rural, economically strapped, and predominantly white parts of the country were especially hard hit.

“I was astonished by the variability we saw,” Subramanian says. The 10 highest-prescribing districts all had more than 110 prescriptions per 100 people. The lowest opioid prescribing rates had fewer than 30 prescriptions per 100 people and were concentrated in congressional districts near urban centers.

As anyone who has paid attention to the opioid crisis knows, prescription drugs are just one part of the challenge. Use of heroin and synthetic opioids such as fentanyl has skyrocketed—and so have fatal overdoses associated with them. In 2017, more than two-thirds of the nation’s 70,000 drug overdoses involved opioids. And Jones, the Alabama state senator, is quick to point out that other drugs such as methamphetamine and cocaine have been a persistent burden for years in many of the same parts of the country that have been hit so hard by opioids.

For Barnett, this complex pattern illustrates a bigger problem. “The overall epidemic that we’re facing is really an epidemic of addiction,” he says. “Addiction to many different substances is common throughout the U.S., and unfortunately we have very poor systems to deal with it and treat it.”

Barnett has recently turned his attention to understanding the barriers to treatment that people with opioid use disorder encounter. In a “secret shopper” study published this year, study participants called health care providers to ask about access to buprenorphine-naloxone (buprenorphine), a highly effective, evidence-based treatment for opioid use disorder. Each provider was called twice—once by a caller posing as a Medicaid enrollee and once by a caller pretending to be an uninsured patient. Callers who said they were current heroin users and seeking treatment were denied an appointment 38 to 46 percent of the time. Moreover, only 50 to 66 percent of clinicians booking a new appointment allowed buprenorphine to be prescribed on the first visit, and there was an overall scarcity of clinicians accepting new patients. Barnett says the study highlighted roadblocks at every step of the process for people interested in treatment with buprenorphine, from finding a clinic that prescribed it in a timely manner to finding clinics that accepted public insurance.

Treating Stigma as Well as Addiction

Treatment for opioid use disorder is complex, multifaceted, and still evolving. Barnett points out that many people with opioid use disorder may need to be maintained on treatments such as buprenorphine-naloxone or methadone for life to prevent addiction from creeping back in. To make headway with the country’s epidemic of substance use disorder, patients will need a continuum of evidence-based treatment and equitable access to that care. Equally important, the culture needs to dispel the stigma that addiction and addiction treatment carry.

Recent polls analyzed by Robert Blendon, the Richard L. Menschel Professor of Public Health and professor of health policy and political analysis at the Harvard Chan School and Harvard Kennedy School, offer a troubling picture of how stigma manifests in the U.S. “Almost six in 10 people say they’re not willing to work closely on a job with someone being treated for opioid addiction. Fifty-five percent say they would prefer not to live next door to someone who’s being treated for opioid use,” Blendon says. “And 40 percent think it’s OK for employers to fire people being treated for opioid use if they lapse.”

Blendon is also concerned about a sharp disconnect surfacing in his polls between Americans’ perception of treatment and the reality of it. While many people consider opioid use disorder an illness that can be treated—a promising sign—far fewer are open to the idea that treatment may last years, or even a lifetime. And while most people don’t know if there is an effective long-term treatment for opioid addiction, of those who think there is, many believe the addict can be successfully treated in less than a year. “When people think of effective treatment, they expect individuals to be able to get off opioids and return to their normal lives,” Blendon says. “They’re going to be disappointed if reality doesn’t live up to their expectations.”

The moralism and shame that looms over addiction is one of the greatest challenges facing the field of public health, Bassett says. “People with substance use disorders face stigma, their families face stigma, and their communities face stigma, which may make it less likely for people to seek available treatment,” she says.

There is, however, no simple blueprint for overcoming stigma—which is tinged with social, racial, and economic elements. Erasing stigma requires cultural shifts at the individual level and at the institutional level, and such changes can often take years. Bassett adds that overcoming stigma also requires a hard look at history.

“When there was a heroin epidemic in the ’70s and ’80s, it was localized in black and brown communities, and a lot of the stigma that emerged around methadone and effective treatment for opioid use had to do with the highly racialized attitude toward addiction. Addiction was criminalized,” she says.

To that end, Bassett is particularly heartened to see an about-face in how some police forces have shifted their stance on substance abuse, treating it as an illness rather than a crime. She hopes this attitude persists no matter the complexion of the sufferer. “People need help, not handcuffs,” she says. “That has always been the appropriate response.”

Chris Sweeney is senior media relations manager at the Harvard T.H. Chan School of Public Health.