Using Social Networks (and some old school sociology) to Understand the Prescription Drug Crisis

Nearly 1 million Americans have died from #drug overdoses since 1999.

That figure boggles the mind. These are pharmaceutical drugs, approved by the FDA and prescribed by doctors and yet there is widespread abuse and misuse of these drugs.

In a paper just published at Administrative Science Quarterly , my brilliant co-authors Victoria Zhang and Marissa King and I decided to apply some old-school #sociology to study who are the physicians that prescribe controlled drugs, with the potential for such widescale abuse.

Our setting is #benzodiazepines, rather than #opioids. Statistically, this helps us avoid problems of unobserved marketing efforts by pharma companies and really narrow in on the physicians' practices. That is because benzo's have been off-patent for years now, so there's much less financial incentives associated with prescribing them compared to Opioids. But yet, they are addictive, they contribute to that huge number of deaths cited above and they are prescribed widely by physicians.

The most common explanation for the overprescribing driving the drug crisis is the one you see on Netflix and read in books. Broadly, it’s that pharma companies and physicians are “pushing” these drugs for some financial motives. Let’s call this “deviance”.

But there is another, less nefarious explanation.

It could be that caring physicians prescribe excessive quantities of these drugs. These physicians are not 'pushing drugs', they are just bending the rules – potentially thinking they’re helping their patients. That is what we call a #liminal practice; not clearly wrongful but unambiguously violating prescribing guidelines.

Using more than 200 million prescriptions we run some basic econometrics to assess how much excess supply of drugs there is in the US at any given point in time. Think of “excess supply” as drugs that were prescribed, but for periods that far exceed guidelines and are thus probably not being put to the best medical use (potentially even just lying in a medicine cabinet).

When we do that, we find a striking fact: “corrupt” physicians account for less than 10% of the excess supply of drugs. Liminal prescribers (your friendly, kind physician who’s prescribing more than they should) account for more than half.

The important point here is that the existence of guidelines but absence of enforcement allows overprescribing to exist in a moral grey zone: clearly not right because unambiguously phrased guidelines indicate it should not be done, but not clearly wrong because punishment remains an unrealized outcome. If that sounds familiar, it is because this happens in many cases (accounting rules, for example).

Next, we try and understand how the social networks of physicians influence which type they’ll be – deviant, liminal or rule-abiding. To do this we construct the patient sharing network of more than 500,000 physicians. We find two striking facts:

1.    “corrupt” physicians look a bit like Berney Madoff. They have lots of ties, and their networks are highly clustered. They are not some lonely physician in an obscure corner of the patient-sharing network. There are also very few of them (2.2% of physicians)

2.    Liminal prescribers on the other hand have very few ties, and the ties they have are not “closed” (tied to each other). There are many more of these liminal prescribers (~16% of physicians).

This paints a very different picture of the drug crisis. It may be fuelled not by “pill mills” and corrupt physicians but by well-intending physicians who lack the network ties to learn about the harmful effects of these drugs before they start prescribing them!

To illustrate this point, we calculate the effect of adding one tie to a rule-abiding physician on the first (“corrupt”) type of physician and the second (“liminal”) type.  Adding a single social connection to an isolated liminal prescriber reduces the chances that they’ll overprescribe by 9.8%. That is massive. The full range is in the figure below (blue line). Adding a connection to a physician who’s already prescribing corruptly only increases their rates of overprescribing (orange line).

Other findings align with this basic explanation. For example, when we measure how much workload physicians have we find that it increases liminal prescribing (perhaps because of lack of attention) but reduces deviant prescribing (perhaps because physicians who sell prescriptions need to be careful with whom they are dealing).

We use physician relocation and some other robustness tests to check that prescribing is driven by the network and not the other way around (which is completely plausible but does not seem to happen in our setting).

The good news is that the interventions needed to curtail liminal prescribing are less harsh (less punitive), and might be easier to implement than trying to identify and catch pill mills and corrupt pharmacies. Their impact on the crisis might also be outsized.

The full paper, with more detailed findings, is here, and it is open access.

#drugabuse #prescriptiondrugs #socialnetworks