Who’s driving the US prescription crisis?

The prescription drug crisis in the US has claimed the lives of just under 1 million people in the last 25 years. According to data from the US Centers for Disease Control and Prevention’s National Center for Health Statistics, around 108,000 Americans died from drug overdoses in 2023 alone.

Among the drugs that are routinely overprescribed by US physicians are benzodiazepines, a class of antidepressants approved by the FDA that include Valium and Xanax and are used to treat conditions like insomnia and anxiety disorders.

Benzodiazepines, or “benzos” have potential side effects. Users can become dependent and experience suicidal ideation, along with a heightened risk of cognitive impairment, falls and fractures. Coming off them is tough. Withdrawal symptoms can include tremors, nausea, memory loss and even seizures. Meanwhile, overdose deaths involving these drugs have surged more than tenfold since 1999, leading the Spectator Magazine and others to surmise that the America’s benzo epidemic is “harder than heroine.”

Determining what is fuelling the crisis isn’t straightforward. Tempting as it is to picture nefarious doctors and pharma companies unscrupulously pushing controlled drugs for financial gain—think Netflix’s “Painkillers” or “Pain Hustlers”— the reality of this crisis is far more nuanced, says Assistant Professor of Strategy and Entrepreneurship, Aharon Cohen-Mohliver.

In a new paper, winner of the 2024 Award for Responsible Research in Management, he and his colleagues find that the main contributors to the benzo crisis are most likely hard working and well-intended doctors who probably believe they are acting in patients’ interests.

Identifying misuse and malpractice

Aharon and co-authors Victoria Zhang and Marissa King used what he calls “old-school sociology” to unpack the dynamics behind America’s benzo problem. Specifically, they wanted to separate doctors into different camps: those who know they are over-prescribing benzos, and “liminal” practitioners who occupy a grey area between ignorance and good intentions with unexpected consequences.

“Medical guidelines around overprescribing benzo are rarely, if ever, enforced in the US,” says Aharon. “So you have a situation where, because no one gets punished for overprescribing these drugs, doctors have to operate very much on the basis of their own judgement.”

To distinguish physicians who prescribe these drugs for other motives (selling prescriptions, for example) they looked at more than 200 million prescriptions made by 500,000+ physicians for 11,300,00+ patients between 2005 and 2008.

Cases of “excess supply” where patients were being prescribed more benzos than the maximum amount guidelines recommend they should be getting were clear, because the guidelines for these drugs are very specific (they should not be prescribed for periods longer than six weeks, typically). This allowed the researchers to estimate the number of pills that were prescribed, and purchased, but likely had no underlying medical condition that they should be treating.

Patients who received overprescribed quantities of benzos could be divided into two groups: those who received prolonged prescriptions but did not have any other signs of misuse or abuse (apart from receiving these drugs for periods that far exceed guidelines). And those patients who returned repeatedly to receive early prescriptions, who crossed state lines to receive overlapping prescriptions or who exhibited similar behaviours that are known signs of misuse or abuse. These patients often pay in cash, ask for the drugs by brand name, and ask for the maximum amount allowed by law. Doctors know how to pick up on these behaviors, particularly when they are new patients they’ve not seen before.

“The data gives us a really good snapshot of the medicine cabinets in American homes, and since most of the benzos in this crisis originate from physicians’ prescriptions, we have a reasonable estimate of the pool of supply of the drugs causing this crisis,” says Aharon.

“We can also estimate if a physician is likely to overprescribe illegally, by looking at how often they prescribe to entirely new patients who have these clear patterns of misuse or abuse. From there, it’s simply mapping how many pills come from these physicians, and how many come from physicians who rarely (or never) prescribe to patients that exhibit signs of misuse but routinely overprescribe to patients who “look” like any regular patient in the data.”

“When patients take benzos (Xanex, Valium, Clonzapam and similar drugs) for long periods continuously they can develop dependence, and withdrawal from it can be extremely difficult, and sometimes risky. When a physician prescribes these drugs for long periods they are overprescribing, and this mechanism – prescribing for continuous long periods (rather than prescriptions of high dosage) is not the first “addiction” mechanisms that doctors think about. In fact, the evidence that these drugs create dependence only emerged in the 1980’s, after they became so popular that at some point, Valium was among the most prescribed drugs in the US.”

The network effect

Crunching the data to separate out which physicians are knowingly overprescribing benzo from those who are doing so unwittingly, Aharon, Victoria and Marissa find something striking.

The “bad” doctors (those who prescribe large quantities benzos to first-time patients) account for less than 10% of the excess supply that fuels this crisis. Meanwhile, liminal over-prescribers (those who overprescribe to patients that do not show any signs of misuse or abuse) make up more than half.

In fact, almost 60% of these excess drugs are coming from what is indistinguishable from any genuinely well-intended, friendly neighbourhood doctor, says Aharon.

Why would well-intended doctors overprescribe drugs that fuel such a crisis?

Our behaviours are shaped by the social networks we have, much more than we tend to think says Aharon. When it comes to overprescribing benzodiazepines, this insight illuminates a lot of the underlying problem.

“The risks associated with overprescribing benzos only became known in the 1980s and 90s, long after these drugs were invented (and after many were already off-patent). While physicians do read guidelines and warnings, they also have strong professional identity, and if they don’t have ties to other physicians who might be more informed on just how harmful overprescribing these drugs can be, it’s easier for them to deviate from the precise guidelines – particularly if no one ever gets punished for overprescribing these drugs,” he says.

“Physicians with few ties, or no ties to other physicians simply don’t get the signals from their environment that tells them to curb their overprescribing. No one gets punished for it, and unless they can see that these drugs are harmful when prescribed for long periods, all they have to go on is information sent to them in various ways or the medications guidelines. For many who have some experience with these drugs, simply warning that they should prescribe for no longer than X weeks is probably not enough.”

To determine physicians’ professional networks, Aharon and his colleagues identify those with “patient sharing” ties: when multiple patients receive prescriptions from the same two doctors.

“Once you know the ties each physician has you can start building their professional network: who’s connected to them, and whether that contact is also connected to another one of the physician’s contacts (creating a “triangle” where each physician is connected to the others). This is a way to determine how many contacts a physician has, and how closely knit these contacts are,” he says.

Physicians who work as a part of a clinic for instance, would show up as a large cluster in which everyone has a tie to everyone else. Physicians who work by themselves in their own practice and rarely refer to the same specialists, which shows up as no patient-sharing ties would show up as “isolates”. A close look at the networks of 500,00+ physicians, reveals another striking finding: the so called “bad doctors” have many more ties, and these ties are significantly better connected to each other than their liminal “good doctors” counterparts.

“This is a bit counterintuitive for people who study misconduct because what we see is that the ‘bad doctor’ prescribers have social networks that are highly clustered with lots of ties, which is not a typical structure for someone who wants to break the law and get away with it,” says Aharon.

“That stereotypical offender should show up as having few ties and sparse networks. That’s a pretty straightforward way to minimize the chances you’ll be caught. Instead of ‘rough doctors’ operating some hidden illegal scheme, perhaps we should think of them more like Bernie Madoff who engineered the Ponzi Scheme and was notoriously well connected,” he says. “Connections bring clients, but sometimes, like in Madoff’s case, they can also help in deterring or even evading detection.”

Conversely, doctors who are most likely to be overprescribing unintentionally have much sparser social networks. They have fewer professional connections—and the connections they do have are not independently connected to each other, says Aharon. And crucially, they make up a significant proportion of the medical population working in the US today.

Lone practitioners who work in isolation, graduated from medical school a long time ago and have high workload have the highest risk of overprescribing unintentionally. They are probably well-intentioned, but also more out of touch with the latest recommendations and guidelines published by authorities. They have fewer opportunities to interact and share information with informed peers.

“Addressing the crisis could be more about encouraging doctors to forge more professional ties than about chasing the few physicians who probably overprescribe for financial gains – who represent only 2.2% of the doctors in the sample. This compares to the 16% who overprescribe but show no signs of doing so intentionally for other motives.”

For Aharon, the award, and attention these findings are receiving outside academic circles, from practitioners and medical societies in the US has a personal significance. “As a patient in long-term medical treatment myself, I’ve had first-hand experience of the side-effects of benzos, and of how incredibly hard it can be to taper off them.”

He added: “I believe it is absolutely incumbent on policymakers to empower doctors— well intended, hardworking but also sometimes too secure in their medical expertise—to connect more, form information sharing networks, and have opportunities to learn from each other. These sorts of interventions cost a fraction of the resources spent on catching the “bad guys” but can have a huge impact. And to this end, we’re optimistic that the findings in our paper will continue to have real impact.”