How FDA-Approved Prescribing Information Lags Behind Real-World Clinical Practice

Once a drug has been approved for some use it can be legally prescribed for any use. New uses for old drugs are often discovered. When physicians learn of these new uses, prescribing practices move beyond the uses that the FDA has evaluated and permitted. In Outdated Prescription Drug Labeling, Shea et al. compare off-label uses for cancer drugs that are graded as “well-accepted” by the National Comprehensive Cancer Drugs & Biologics Compendium (NCCN) with the labelled, “FDA-approved” uses. What they find is that most drugs have multiple off-label uses that are significantly different from FDA approved uses.

Our analysis of the NCCN Compendium and FDA drug labels for 43 cancer drugs approved between 1999 and 2011 identified hundreds of off-label uses, most of which were strongly supported by NCCN expert panels.

…Additionally, of the 253 off-label uses, 165 (65.2%) were categorized as “new indications,” meaning they were in disease settings not represented on labels

In my work on off-label prescribing (and with Klein) I have emphasized that the off-label world offers a window on what the larger world would look like with much less FDA control over new drug approvals. Notice that even today it’s physicians and the private approval process, as represented by the compendia, that determine actual prescribing and payment.

We found that 4 of the 5 largest private payers, as well as Medicare, cover over 90% of uses listed on the NCCN Compendium (uses graded 1 and 2A), suggesting widespread acceptance of these uses by diverse stakeholders. While standards for FDA approval differ from standards for coverage determinations, these findings indicate that the gulf between labeled uses and covered uses may be needlessly wide.

To bring FDA labeling up to real-world practice the authors recommend “a collaboration between the FDA and the developers of clinical guidelines and drug compendia to evaluate existing evidence about approved drugs and suggest updates to labeling.” In other words, the decentralized, private approval system should be used to determine which new uses of old drugs are safe and effective and those determinations should then be adopted by the FDA. I agree. But if private practices can be used to approve new uses for old drugs, why shouldn’t similar procedures be used to approve new uses for new drugs?


Here is a recent article that found that the evidence NCCN uses to generate its recommendations is often very weak--randomized clinical trials are the basis for just 23% of indications they recommend outside the FDA labeling. The authors conclude, "The NCCN’s reliance on lower quality of evidence may lead to false inferences concerning the efficacy of toxic, costly cancer drugs or the magnitude of their benefit.”

Alex's argument that the payers cover compendia-recommended indications is completely meaningless. By law, Medicare must cover compendia recommended cancer indications. ( Then the private payers just follow Medicare coverage decisions for these physician administered drugs.

So we have this system where NCCN uses poor evidence to recommend more indications, then Medicare must cover it, and the private insurers follow. So of course the drug makers have little to no incentive to go back to FDA with solid evidence of effectiveness and safety to expand their labeling. Drug spending continues to increase, especially in oncology, because there is no check on what Medicare and insurers will cover. NCCN is a poor gatekeeper. And anecdotally, the other compendia are even worse, often with financial conflicts of interest.

Come now., what do you have against Laetrile, a most promising cancer therapy being stymied by the cold hand of the FDA?

(1) The post's point is that the randomized trials the FDA insists on are slow and unnecessary. Sometimes it's obvious something works, even without formal trials. So that;s a feature, not a bug.

(2) the private insurers don't HAVE to follow Medicare, do they? They'd make exceptions to following Medicare if they could provide better health care cheaper.

The way laws are written that define adequate coverage and other regulations on insurers means that if the insurers did not follow Medicare, they would either run afoul of coverage requirements or a state insurance commissioner might just say they have to cover it.

So it may be that the "if they could provide better health care cheaper" is not actually possible due to interactions with other state and federal laws.

1) this is not true 2) when the largest payer in the country covers a bunch of things and you don't, that's the easiest way to lose customers real fast, even if you're offering premiums a couple percent lower than your competitors.

So the affront is to the process.

What about the people who are receiving these drugs? Are some people getting well? Is the state of the art advancing?

Could these questions even be contemplated if the process is not affronted?

Some are getting well, many aren't, and for most we actually don't know. Cancer patients are often on a whole cocktail of therapies, making knowing what works and how well pretty much impossible without the level of controlled study FDA typically asks for. You don't do these types of studies for fun. It would be cruel to deny people of therapies that work better than existing treatments if you know they are truly better. The opposite is true as well..

Thanks Jan. After reading the post, I was logging in to make very similar comments.
The post overlooks the fact that off-label use of oncology drugs are based on poor evidence, and do net harm to patients.
I say this as an oncologist.

Anti-science comes in many forms, the most dangerous being ideologues who would substitute their ideology for science. Physicians are not scientists, and their history of indiscriminately prescribing inappropriate drugs does not engender confidence in their ability to determine off-label uses of drugs. The ideologues might argue that the devil (i.e., Medicaid) made them do it, but the reality is that the devil didn't do it, the physicians who indiscriminately prescribed opioids did it.

What if drug companies game the system: If they have a drug that works for some use, such as acne, and they suspect it might work for skin cancer, but the testing would take longer and cost more, wouldn't the strategy be to get approval for acne. Then you would later see the off label usage that Alex sees.

Another in the endless post from Professor Tabarrok urging an end to the current FDA approval process. It's important to distinguish between follow on indications for already approved drugs where knowledge about the drugs benefit/risk ration is known and an unapproved drug where it is not. The new drug approval process is of necessity rigorous as one does not know the full safety profile because of the small number of patients enrolled in the clinical trial. Additionally, companies cannot study every single indication the drug may end up being used for as that would significantly increase the length of the trial process. IIRC, the Pfizer antibiotic trovafloxacin was studied in a a large number of infection indications for which it was approved only to discover that acute liver injury could not be successfully managed. It's well known that oncology drugs may have more use than the indication(s) that it was approved for.

Jan's point in the first post on the thread is spot on.

"But if private practices can be used to approve new uses for old drugs, why shouldn’t similar procedures be used to approve new uses for new drugs?"

The critical difference is that FDA government bureaucrats are near angelic public servants with few human imperfections, self-interest, or laxity in their noble mission. OTOH all private companies are populated by greedy schemers who are usually quite willing to sacrifice public safety for their personal profit. One can always trust government employees... but must always always be highly suspicious of private business personnel, especially in medical/health issues.

Even if you don't go that far, there is something far simpler.

Complex systems work because they aren't totalitarian in nature, and people make them work by finding ways to make them work. To change the rules of the FDA is almost impossible. No one can seriously posit that the procedures of the FDA are optimum.

If the FDA had absolute control and things worked as written, the death toll would be shocking.

'prescribing practices move beyond the uses that the FDA has evaluated and permitted' is easily contradicted by quoting the words of the Bartley J. Madden Chair in Economics at the Mercatus Center - 'Once a drug has been approved for some use it can be legally prescribed for any use.' Once approved by the FDA, all uses are legal, as noted.

Again, this is simple not precisely accurate, while attempting to create a narrative - 'What they find is that most drugs have multiple off-label uses that are significantly different from FDA approved uses.' If such uses are found relevant and beneficial to the bottom line of a pharma company (who cares nothing about actual improvement among users), then those uses will be approved too. Which is almost irrelevant, considering that it is fully legal to prescribe a FDA approved drug for any purpose.

'To bring FDA labeling up to real-world practice the authors recommend “a collaboration between the FDA and the developers of clinical guidelines and drug compendia to evaluate existing evidence about approved drugs and suggest updates to labeling.” '

Lovely idea, and one can be confident that the pharma industry will fully back it as long as it involves costs only to taxpayers, and not pharma companies. At least, one would assume that such a position would be uncontraversial to anyone familiar with public choice economics..

If expanded label claims for therapeutic targets based on clinical trials are allowed, I'd be all for that if it's traded for an end to off-label precriptions. I don't think that's what the professor had in mind. The current situation is the Wild West -- once it's approved for one thing, you can use it for anything.

A good example is human growth hormone (HGH). It's useful for a small number of children with a genetic mutation that creates a deficiency of this hormone. But for decades there has been a cult of quack doctors prescribing it as a sort of "anti-aging" medicine. Decades have passed without any evidence to support that use, but it is legal to prescribe if you're an MD or DO. Plenty of doctors in Florida and southern California have lucrative practices doing exactly that. If we get rid of off-label prescribing, that form of quackery goes away.

"widespread acceptance of these uses by diverse stakeholders": what about white stakeholders?

Whites prefer to hold knives over stakes.

It's worth noting that cancer drugs are really a special case. For one thing, in many cases patient populations are small, and that makes running clinical trials very difficult and too expensive to pay off for Pharma companies. For another, we tolerate a whole lot of side effects from cancer drugs that would be unacceptable in other conditions.

Off-label prescribing is much less common outside cancer therapy, though of course it does happen. FDA certainly tries its best to pressure companies into running trials and submitting applications when an off-label use becomes widespread.

I think most doctors would prefer to use drugs where there is strong scientific evidence of safety and efficacy. That's really the FDA standard.

" That’s really the FDA standard. "

FDA Standard is extreme caution and forcible suppression of doctor & patient choice (liberty) in medical issues.

The major, overwhelming personal risk to FDA bureaucrats is their formal 'approval' of drugs/devices that might harm someone -- there is little personal risk in formal 'disapproval' or long approval delay for effective drugs/devices.
FDA bureaucrats have sharply different incentives than doctors/patients.

'FDA Standard is extreme caution and forcible suppression of doctor & patient choice (liberty) in medical issues.'

Laetrile continues to be oppressed, thus ruining a glorious profit making opportunity.

Just to echo comments made by Alan Goldhammer. The question at the end bait private markets replacing the fda for new drugs ignores the large costs and rigor that go into determining the side effect profile for a new drug. Private companies have far less incentive to pay attention to, report, and address the many risks that may come with a new drug. The Fda serves this role of demanding a certain level of safety before approval.

Insurnace companies have extremely strong incentives to avoid approving drugs that hurt their customers, unless, perhaps, the drugs kill them at an early stage of an expensive ailment. The difference from the FDA is that insurance companies also ahve strong incentives to approve effective drugs, whereas those incentives are weak for the FDA. The insurance companies look at both sides.

Good point. A few thoughts. Not evidence based. Insurance companies definitely don't want to approve drugs causing morbidity and mortality that they have to pay for if the benefits of the drug in reducing medical costs is less than the cost of adverse effects. I don't know if they have the same incentive to track rate but fatal harms that could accrue if the majority of patients are fine and the overall cost is decreased. In those cases wouldn't they be aligned to ignore the few patients harmed for net gain.

People can have differing opinions on that balance and whether it would be ok for harm to accrue for some but most patients would like to know and make the decision for themselves. Not sure insurance companies would meet the standards of identifying all those rare outcomes.

This is a rather curious statement. Insurance companies to my knowledge only want to minimize the probability of using expensive drugs. Many of the new biotech drugs cost upwards of $2K/month and don't have the best safety profiles (just listen to the litany of side effects on the DTC television advertisements). Insurance companies rely on the FDA approved drug label for safety and efficacy and in the absence of that would be forced to gather all this information on their own. One of the unfortunate effects of this would be to drive up drug costs which are already quite high. Professor Tabarrok's approach of using some third party standards organization doesn't fully address this issue. Data has to be collected, analyzed and then reviewed. Moving this to some third party is not going to markedly negate those costs.

No no no. Insurance companies have incentives to give their customers/patients what they want, which is usually hope, even if it is unproven and futile. Cancer drugs are typically measured by how much they extend survival, not ability to cure, so the question is whether the drug is effective in that metric compared to existing therapies, not whether it's "safe." When the largest payer in the country covers a drug for a particular terminal indication, from a business perspective you can't say no. Even if the only reason Medicare covers it is because it is mandated to by law.

In order to more easily promote the benefits of newly discovered uses of previously discovered compounds, there should be a streamlined and simplified process for people aware of coverups related to problems caused by a pharmaceutical.

For example, if patently fraudulent statistical manipulations that could not possibly be explained by accident resulted in covering up boys growing breasts or twice as many heart attacks in elderly patients ... it should be easy, not hard, for an individual to profit handsomely from drawing attention to fraudulent or even fabricated statistics related to such things (e.g.,

Off-label use in NCCN compenida are typically backed by evidence, albeit at a lower level than required (typically phase II vs.
phase III studies). A manufacturer may not want or need to gain US label approval to support use of their agents in off-label populations. The chief advantage of FDA approval from a pharma manufacturers perspective is the ability to promote as only in label information can be discussed by the sales force and included in marketing collateral.

I am going to assume not many people read the abstract or the original paper. Even if you just read the abstract, they note that most of these new uses were just using the cancer drug on a different kind of cancer.

"and 65% were in cancer types not previously represented on labeling"

The fact that the drug was approved by the FDA after testing gives people the confidence to go ahead and use the drugs for other cancers. Take away that initial approval and don't think many of us would be using these drugs. (Very few drugs ever get approved for pediatric patients. They are almost always used off label. Doesn't mean I want to give some kid a drug that has never been tested or just based on the word of a drug rep.)


I think Alex misses a few key things here:

1. His evidence undercuts his agenda. If drugs are often used off label, even cancer drugs which tend to be very expensive, then what is the need for FDA changes? To update drug labels easier? Why is that important if off label use can happen now?

2. Altering the label does require a higher level of evidence, which is why off label use happens. But this does provide useful information to us. We know the label information has gone through the most rigorous levels of proof while off-label use is probably better than pure speculation but not quite as firm in its evidence.

3. Bayesian statistics here. If a particular drug is shown to be safe and effective as some agent, say killing lung cancer cells with a particular mutation. I think that increases the odds it will prove useful in other contexts (say breast cancer with the same mutation). Off label use then in a sense is trusting a drug we already know on some level, but that drug made it through 'basic training' of getting its first approval....we know most drugs don't make it through 'basic'. So which is it, have we left behind a lot of gem drugs that work great in a particular context but flamed out because we just didn't test them right OR did we leave behind a lot of duds in which case we can end up wasting a lot of time and lives playing with things that we already know don't work?

Here I think the former is less common. Drug companies invest a lot in their pipeline and they would rather bring a drug to market than not. Even if they can only demonstrate a drug works for a very rare condition or is only slightly better than what is out there, they will often do that if they can. There are many stories of drugs that started out approved for only a tiny patient population that then expanded as new indications were found to broaden their labels. If they left behind 'gems', odds are such gems only have a slight statistical signal which means it's all the more important to use a controlled study to demonstrate their effectiveness rather than rely on anecdotes On the other hand imagine it was legal to buy up 'duds' and sell them under a 'give it a try' regime? I suspect that would invite a lot of flim flam operations into the market and pull away from the hard R&D that often generates the real advances on the ground. .

There is no way a change in handling off label use will result in any real liberalization under the current regime. It will just lead to even more sclerosis, and if it becomes an approval path then doctors will start being sanctioned for any off label use.

It was hard enough getting biologics anywhere near approval for lupus and ankylosing spondylitis, anyone with anything “rare” like say vasculitis will be doomed.

Canabis Twenty-Four Seven was created to spread information about the
wonderful cannabis plant. We aim to impart information about marijuana and all the other benefits
of products derived from it like CBD oil. The site also shares as mjch information that it can that may be beneficial
for cannabnis growers and cannabis users alike. Marijuana has long been demonized but if you think about it, marijuana is a plant that is a gift from Mother Nature.
It was created so we humans can benefit from it. The cannabis plant is not only useful for its medicinal properties, but it also has other industrial uses.
Withh that said, the industrial use of marijuana is good
for the environment as it has a very low carbon footprint.
Too bad, more research cannot be conducted about it because this plant has been deemed "illegal" by the powers that control this world.
Marijuana should not be prohibited nor regulated by man-made statutes.
The state that the woeld is in right now regarding cannabis is one that iss in contrast with natural law.
Man labeling cannabis as illegal is like sayinmg that man knows more about creation than Mother Nature.
It is a shame, really.

Off label uses for pharmacuticals. Yet, Cannabis has no medical uses! Despite decades of documented use worldwide.

Comments for this post are closed