The notices were a simple but unusual experiment — part of a growing research effort aimed at finding solutions to the opioid epidemic that is estimated to have killed almost 50,000 people from overdoses last year. They also addressed an almost astonishing gap in the American health-care system — the gulf between the care doctors provide and their knowledge about the consequences for patients. Many doctors who prescribe painkillers may believe that addiction is a problem that happens to other doctors' patients, because they never learn about their own patients who died of an overdose.
The letters were successful, although the effects were modest. Doctors who were informed of their patients' deaths were 7 percent less likely to start new patients on opioids and issued fewer high-dose prescriptions over the next three months, compared with those who did not receive a letter. In total, there was a 9.7 percent reduction in the total amount of opioids they prescribed, according to results published Thursday in the journal Science.
“What's particularly interesting to me is the personal nature of it,” said Alexander Chiu, a surgeon at Yale New Haven Hospital who was not involved in the study. “Depending on what field you're in, [the opioid epidemic] can feel a little remote. If you're not a pain doctor or a primary-care doctor, it's not quite as common to know or see your actions having a negative impact, which is what this is showing — it makes it very real. As evidence-based as we are as a profession, sometimes anecdotes can be really powerful.”
Researchers at the University of Southern California worked with San Diego County's chief deputy medical examiner on the project, which is part of an emerging thread of research into nudges and policy changes that could affect physicians' behavior, after years of top-down efforts to restrict or set guidelines on how opioids should be used or to track prescriptions.
There's evidence that the volume of prescription opioids being dispensed overall is beginning to decline, but a study published this month in the BMJ, looking at people with commercial insurance and private Medicare plans, found that the trend may not be uniform. Opioid use and the average daily dose given have not declined substantially for people covered by that type of insurance, according to the study.
The new study, and others, are focused on fixing subtle biases that may enable the rampant use of opioids — one of which could be human nature itself.
“One of the takeaways I’d like people to have is that doctors learn a lot of clinical facts, but when it comes to clinical judgment and decision-making, they fall prey to the same biases that we all do,” said Jason Doctor, director of health informatics at the University of Southern California's Schaeffer Center for Health Policy and Economics, who led the work.
Psychologists talk about a concept called the “availability heuristic” — the idea that people making decisions draw on their most recent experiences. If a doctor learns that a few months after she prescribed 30 pills of OxyContin, a patient died of an overdose, the information may make her reconsider whether the prescription is truly necessary next time.
Doctors who are reminded their actions are being reviewed may also change their behavior. A study published this month in JAMA Psychiatry found that sending letters to doctors who were high prescribers of Seroquel, an antipsychotic drug that carries a risk of side effects and is often inappropriately given to elderly people, led to decreases in prescribing, without negative effects on patients. But the evidence is mixed on this approach; an earlier Health Affairs study tried a similar tack to alert high prescribers of controlled substances that they were out of sync with their peers, but it found no effect.
Jason Doctor said that San Diego County will continue sending the letters and that several other counties have expressed interest. But letters notifying doctors of patient deaths are far from a stand-alone solution to the opioid crisis.
The decline in opioid prescribing was modest and was tracked for three months, so it is not clear how long the effect could last. The study did not track the impact of the differences in prescribing on the quality of care. And prescription opioids are only one contributor to the opioid epidemic, which is also fueled by the use of illicit opioids such as heroin.
The hope is that a multipronged approach could ultimately have a powerful combined effect. Chiu's team reported in JAMA Surgery that simply changing the default number of pills in the electronic health record from 30 pills to 12 pills per prescription could decrease prescribing; the average number of pills dispensed dropped to 20. Caleb Alexander, an associate professor at the Johns Hopkins Bloomberg School of Public Health, argues that a suite of relatively straightforward information-technology solutions, such as better integrating databases that monitor prescription drug use by patients with the medical record, could help address the crisis.
“The epidemic is a behemoth — a massive, complex epidemic. It took us 20 years to get fully into this, and it will take us 20 years to get out of it,” Alexander said. “We are at a pretty unique point in the trajectory of the epidemic — public concern is unprecedented regarding the harms that have been caused by opioids, as well as heroin and illicit fentanyl. But it’s going to take us a long time to recover, as it does from any epidemic.”