Skip to main content

Why Good Doctors Prescribe Off-Label Treatments

Medicine does not always have the luxury of waiting until the mountains of needed research on any given problem are completed centuries from now. Physicians are tasked with treating real-world suffering today.
  • Author:
  • Updated:


In May 2013, in cooperation with the investigative journalism non-profit ProPublica, The Washington Postpublished a high-profile expose into physicians’ prescribing habits. The article highlighted that, although the Centers for Medicare and Medicaid Services (CMS) should know the prescribing trends of its participating physicians, the agency has not used the trove of data at its fingertips to police member prescribers. ProPublica’s enterprising reporters sought to prove the point that these data can be exploited with a mind toward keeping Medicare beneficiaries protected from suspicious clinicians. The reporters filed a successful Freedom of Information Act (FOIA) request that gave them access to prescription data of the CMS, then used the data to identify outliers who were prescribing certain psychiatric drugs.

Two questionable assumptions drove the reporting by The Washington Post: (1) Medicare has the mandate to police physicians practicing legally and in their normal scope; and (2) there is something inherently suspect or dangerous about off-label prescribing. To back up the piece’s implication that off-label prescribing carries unusual obligations, reporters turned to bioethicist Alexander Capron who advised, “when one moves beyond a single patient or maybe a couple of patients ... you’re basically saying, ‘I’m doing a study.’” The reporters then pointed out that, in a “clinical study, patients or families would be informed about side effects and risks.” I believe the combination of Capron’s declaration and the contention that CMS should investigate outlying prescribers creates chilling implications for the practice of medicine in difficult populations.

It is in these most pressing areas where clinicians come to realize that no physician in history has ever truthfully claimed first doing no harm.

As a physician tasked with managing the care of patients suffering from the effects of acquired brain injuries, I care a lot about off-label prescribing. I prescribe off-label every working day. Thus invested, I engaged Capron on the Medical College of Wisconsin’s Bioethics listserv where he confirmed to me that The Washington Post quoted him accurately. He stood behind his remarks. Capron believes that a physician who has found an off-label use of a medication to be helpful for a particular problem in a particular population has an ethical obligation to publish a research article reporting the findings to the world. He has even given a threshold outlining when the practice of medicine ends versus when the practice of unethical, non-institutional review board approved research begins. Once a physician finds a new off-label use helpful in more than two patients, he or she is engaging in unethical conduct if he or she continues treating further patients without setting up a research study. As a result, routine medicine becomes an impractical and unhelpful endeavor under these constraints, which I do not consider ethical at all.

The Washington Post story served to announce the debut of “Prescriber Checkup,” a searchable online database ProPublica compiled with their FOIA data that allows the general public to learn the top drugs any Medicare-participating physician prescribes for Medicare patients (drugs with more than 50 prescriptions). One can see his or her own physician profile by going to the website for the ProPublica database. ProPublica hopes that one day every clinician’s prescription history will be a matter of public record, whether or not a public insurance program is paying for the medications we are prescribing.

The reporters knew what they were doing when they checked up on Robert Morton, MD, an Oklahoma psychiatrist whose medical license has twice been revoked due to an opioid addiction in the 1990s. They noted that Morton, now legally back in practice, writes prescriptions for a lot of memantine hydrochloride. He prescribes it most often for patients under the age of 65, which suggests that he is not using the drug for its sole FDA-approved indication, Alzheimer-type dementia. Morton informed his interviewers that he has found that memantine helps calm overstimulation behaviors, such as rocking and flapping, in the older children and adults with autism and developmental disability he treats in facilities across Oklahoma. The fact that a small amount of the literature that discusses memantine as a treatment for difficult behaviors in autism has been accruing since 2006 did not make it into ProPublica’s story, in favor of a quote from a prominent psychiatrist who disclosed that he has no experience with the drug. The Yale psychiatrist stated that he is “dubious about the rationale” for giving an NMDA receptor antagonist to calm problem behaviors in patients with autism, an opinion from on high that appears to overrule what Morton is finding on the ground in Oklahoma.

It must be because I also try to help patients in desperate situations due to problematic behaviors that I did not read about Morton’s use of memantine with horror, but with interest. Evidence has been building in support of another drug in the NMDA receptor antagonist class called amantadine, which appears to help accelerate the return of consciousness after severe acquired brain injury. A good clinician working on the front lines might theorize that such a stimulant is worth trying, preferable to the heavier antipsychotics and benzodiazepines which are often used for the behaviors Morton identified, yet achieve their effects with a heavier toll on neural networks. The nursing director of one of the facilities that Morton serves told reporters that the memantine has “worked wonders for mood behaviors in developmentally disabled patients.” I’m pleased to see that Morton has found something that works in his similarly complex population of children and adults so impaired with harmful behaviors that they are residing in institutions rather than in the community. In practicing brain injury medicine, I know about this challenge all too well.

Medicine does not always have the luxury of waiting until the mountains of needed research on any given problem are completed centuries from now. We are tasked with treating real-world suffering today. It is in these most pressing areas where clinicians come to realize that no physician in history has ever truthfully claimed first doing no harm. None of us has ever truly understood every downstream implication of any single procedure or pill, as any bench scientist can inform the most confident clinician in his or her medical center. We must rise to a level of comfort with an intervention in building our theory as to where it may help and where it may not, and then we go forth and practice medicine with such tools as we have to deploy.

There is a natural struggle between the randomized controlled trial based medicine that spawns most FDA drug approvals and personalized medicine. Practitioners of behavioral management must trend toward the latter view if we are to be of any use to our patients. There is not a single FDA-approved drug for the treatment of cognition or behavior in acquired brain injury, and we have precious few classes of neuromodulatory drugs with which to work. All of these drugs can have unintended effects. Although the term is most often used these days to describe genetically targeted drugs, “personalized medicine” need not involve such biomarkers at all. It depends entirely on what you are treating as to what your best metric should be. When the problem is your patient slamming his or her head into a wall or striking his or her caregivers, the cessation of these behaviors is a more precise measure than any yet-to-be-developed gene assay.

I practice in a field in which the unexpected happens every day. One patient with severe TBI and with intraventricular hemorrhage may develop hydrocephalus, another may not; one penetrating brain trauma will lead to seizures, another will not. One patient with a thalamic hemorrhage may suffer debilitating central pain, others will not. One patient in a vegetative state perks up to minimally conscious with zolpidem, another does not. I sympathize with Morton when he told The Washington Post that he has not had time to write up his successes with memantine.

Once I had a patient who was minimally conscious and who spoke only when we turned her upside down. I have yet to write up this case report. I know there are underlying physiologic mechanisms that will one day explain the phenomena I see every day; happily a torrent of neuroscience research flows in this direction. In the meantime, although I can sometimes describe what I see little better than a 19th-century British naturalist on safari, I am managing to help many of my patients and their families through the wilderness.

As a rehabilitation physician, the ultimate treatment I wield is not a pill at all but the interdisciplinary therapy program at my hospital. We could study my program’s innumerable variables for the next century before changing anything we do. The literature has yet to answer even the most basic questions about the length and intensity of treatment. The correct answer to “how much and how long” is always a moving target based on an individual patient’s unique goals, resources, social setting, and evolving functional status. Perhaps this experience is why, when I read about an Oklahoma psychiatrist prescribing off-label memantine to severely developmentally disabled patients, I was not scandalized. Perhaps this is why when I read about a bioethicist whose ethical prescription would box me out of using baclofen in stroke, trazodone for sleep, propranolol for storming, intrathecal baclofen within a year of brain injury, or Botox for lower limb spasticity, all non-FDA indications, I worry for my patients.

Physicians who treat patients with acquired and developmental brain injuries know that our patients do not attract many new large randomized drug trials or new drug indications. We are used to rapidly reaching the edge of known medical science, where our practical expertise begins to fill in for the dearth of research. In these cases, we are correct to draw from our knowledge of analogous conditions that often have a greater research base simply due to a more economically favorable population for pharmaceutical and device company investment. We have a duty to keep abreast of the latest information available about the treatments we offer and a duty to discuss innovative new treatment approaches with our colleagues. We have a duty to inform our patients and their caregivers about our level of confidence in the treatments we offer, and we should encourage their participation in weighing potential treatment effects versus possible adverse effects. Such conversations should acknowledge that lower-risk treatments allow for more speculative and fewer evidence-based approaches, and more severe problems without standard approaches can justify greater risk taking under informed consent. Claims that this model of expertise-based and individual problem-oriented medical practice represents an ethical lapse deserve a strong response. When will all the dedicated physicians tackling the toughest cases start to back away for fear a national newspaper will label us rogues experimenting on vulnerable populations?

A version of this article appears in the October issue of PM&R.