Frances Kelsey Syndrome

Occasionally I have been told that FDA reform is something that only a few libertarian economists support. But in fact, there is strong support for reform in much of the medical community. See, for example, the survey that Dan Klein and I did on off-label prescribing and FDA reform or the many surveys of physicians done by CEI.

I mentioned Dr. Vincent DeVita in my post, Will Trump Appoint a Great FDA Commissioner? It’s worthwhile exploring DeVita’s views at greater length because he is a prominent figure in the field of oncology. Here, from Wikipedia, is some background on DeVita:

Vincent Theodore DeVita, Jr., MD is an internationally recognized pioneer physician in the field of oncology. DeVita spent the early part of his career at the National Cancer Institute (NCI). In 1980, the president of the United States appointed him as director of the NCI and the National Cancer Program, a position he held until 1988. While at the NCI, he was instrumental in developing combination chemotherapy programs that ultimately led to an effective regimen of curative chemotherapy for Hodgkin’s disease and diffuse large cell lymphomas. Along with colleagues at the NCI, he developed the four-drug combination, known by the acronym MOPP, which increased the cure rate for patients with advanced Hodgkin’s disease from nearly zero to over 70%.

DeVita was the Director of Yale Cancer Center from 1993 to 2003. He is currently the chair of the Yale Cancer Center advisory board and is professor of internal medicine and of epidemiology and public health at Yale’s medical school.

DeVita currently serves on the editorial boards of numerous scientific journals and is the author or co-author of more than 450 scientific articles. He is one of the three editors of the popular textbook (also available online) Cancer: Principles and Practice of Oncology and serves as the editor-in-chief of The Cancer Journal.

In his book, The Death of Cancer, DeVita has a chapter on the FDA. The title of that chapter? Frances Kelsey Syndrome. He writes:

The thalidomide episode sent the message to those who worked at the FDA that the way to do right by people was to say no. Saying yes would prove perilous–not only to patients, but to the careers of a reviewer. As a result, the agency tends to reward those who say no, not yes. (In fact, there’s an annual Frances Kelsey award. But there are no awards for getting a good drug quickly into the public domain.)

Exactly right. Later he discusses another problem that he sees with FDA evaluation procedures:

We are approaching what we might have considered nirvana years ago. We can design drugs that will hit a specific target, and by being so specific, they have fewer side effects. But because effective treatments almost always require hitting more than one target at the same time, some very good and relatively safe cancer drugs show no evidence of effectiveness when used alone.

What a dilemma. After spending millions of dollars developing a drug, a company may be forced to abandon it for lack of efficacy, when, if approved, it would be another exciting tool for clinical investigators who want to explore combinations of targeted therapies in post-market research trials. Compound that with the FDA’s insistence on testing them first on patients with very advanced, resistant disease, and many potentially useful drugs don’t look so good. As a result, drug companies are reluctant to invest money in new cancer drugs, because they might never make it past the FDA’s hurdles.

DeVita may be wrong but he’s certainly not uninformed.


One gets the impression from reading Tabarrok that the problem in medicine is not enough individualized medicine, but the problem in medicine is just the opposite: physicians refusing to apply standard practice in diagnosis and treatment, insisting instead that their "judgment" is somehow superior to data collected across a large population of patients with the same symptoms. Sure, it's possible that one patient would be better off with the individualized medicine Tabarrok prefers, but thousands are worse off because of it. .

What this post is describing is exactly what AIDS researchers ran into. There wasn't one simple A -> B process where a drug could fix this problem. Still isn't. One drug doesn't work, but a combination of them does. And it took experimentation over a few years to come up with a treatment protocol that was reliable, and with experience became predictable and simple and less need for individual tweaking.

The regulatory agency got in the way of stopping people from dying in that case, and in some situations do the same today. That regulatory agency in other situations prevent people from dying.

Is it impossible to have a regulatory agency that stops doing the first while continuing doing the second?

"The regulatory agency got in the way of stopping people from dying in that case, and in some situations do the same today. That regulatory agency in other situations prevent people from dying."

Can you point to a single HIV drug whose approval was delayed by the FDA? FDA worked with sponsors to figure out a way to get these drugs approved as quickly as possible. In addition, a pathway that allowed broad dissemination of experimental drugs was established (I was on that Public Health Service work group).

My understanding is that it took some pressure to change the standard procedures at the time. Having thousands of people dropping dead tends to do that to regulatory lethargy.

The problem was no one admitted HIV/AIDS existed as other than God's will punishing gay drug users, until Ryan White.

How can the FDA approve drugs that don't exist to cure a disease that doesn't exist?

More important, who will pay to do any drug research for treating a disease the government says does not exist, and that the leaders order the CDC not to investigate.

Alex never seems to explain who pays for the innovative drugs being developed other than patients who are desperate and thus willing to go deep in debt to pay anything their credit limit affords to any charlatan promising that sewing goat gonads into your belly will cure everything from impotency to balding to lack of vitality. (I grew up before FM and when Mexican superstations were where you could hear rock and roll and Wolfman Jack sponsored by the FDA banned goat gonads, radium alixers, mercury baldness cures sold mail order from Mexico, along with the Jesus prayer cures from sending $50 to the church of miracle cures..)

Which drug company CEOs in the 80s were authorizing tens of millions of dollars in researching the gay drug user disease in order to figure out what chemical compound would ward off God's wrath?

Which global drug makers are funding, out of profits from taxpayer and taxpayer subsidized Epipen profits, drug development to prevent malaria, dengue, etc which have billions of customers in nations with no FDA, not big government takeover of health care, and where medicine is free market, paid out of pocket by patients?? You know, Africa and Asia?

Given Africa and much of Asia has free market health care and virtually no regulation but 10 times the customers for patient centered and paid for health care of the US, why isn't Alex pointing to the miracle cures in Africa that he can't buy in the US?

Given Africa and much of Asia has free market health care and virtually no regulation

Sorry but what the effing heck do you think Africa is like? Have you ever heard of African socialism? What makes you think that any part of Africa has a free market in health care? You mean like Angola - ruled by a de facto Communist party? Mozambique? Zimbabwe? What part of their medical system is free precisely?

.... "The regulatory agency got in the way ..."

Perhaps the insertion of 3rd-party gatekeepers (FDA employees) into personal medical decisions for all Americans is not a good idea at all. Especially if those gatekeepers are unelected/unaccountable with vast arbitrary power over critical medical decisions and markets.

Any FDA 'reform' efforts should at least closely scrutinize how individual FDA gatekeepers are selected, trained, and supervised. No one here has more then a vague general idea of how FDA personnel actually function on a nitty gritty daily basis-- but nevertheless, most put tremendous trust in that organization.
There is no actual "FDA", only ordinary humans with FDA name tags.

Using your line of thinking, maybe we should eliminate doctors too? they are unelected and they have vast arbitrary power over critical medical decisions and markets...

Of course one can have choice in doctors, in fact one can even doctor shop.

A system where one could FDA shop would be very different than what we have, but that is getting into libertarian extacy, which has known and dangerous side effects.

This embarassingly bad argument brought to you by the left having a hard time distinguishing the use of force.

You need to argue that Africa has far better health care than the US.

No Obamacare.
No NIH picking winners and losers on medical research.
No CDC dictating what is and isn't a disease and what individuals must inject in their children.
No Medicare death panels.
No AMA preventing anyone from being a prescribing doctor.

Africa must have far superior health care than the US because it's almost entirely free market and unregulated.

Sooo weak

I know people who order drugs from countries like Vanuatu because they have not passed efficacy tests per the US FDA. Domperidone is an example of a drug found reasonably safe in testing, and used worldwide for gastroparesis symptoms, but not generally available in the US because appropriate efficacy tests were not done before it went off-patent. To the FDA's credit, they recently made domperidone an Investigational New Drug were a doctor can go through tremendous bureaucracy to get a permit to give (but not sell) it to patients. It is available over-the-counter in UK and South Africa, by prescription in Canada.

Large randomized controls are built on so many assumptions and are so expensive. You assume stable unit value treatment effects, you assume counterfactuals are well matched, you typically assume a linear treatment effect, you typically assume low power, you assume no strong conditional effects etc...

This works well for drugs with unambiguous signals that apply across a population. For low sample size, high risk, high heterogeneity, nonlinear treatments/illnesses (e.g. Cancer) that require combinations and often a multidimensional treatment plan, a physicians judgement is often stronger. After all, what is judgemsnt? The human brain can uncover nonlinear structure and patterns across various samples far better than a regression. That's why despite massive research, deep learning diagnostic algorithms still can't match clinical efficacy (or come close).

It's no secret for any methodologist that an experienced physician has insight into treatment that is developed both as a combination of both experimental design knowledge and clinical experience.

The problem is people often fetishize the experimental process, since they haven't actually worked with it long enough to vividly understand where and how it fails, and why human override is sometimes permissible.

The scenario you described (low sample size, high risk, high heterogeneity, nonlinear treatments/illnesses (e.g. Cancer) that require combinations and often a multidimensional treatment plan) is also the scenario where it becomes easy to see "patterns" when none exist.

The problem isn't that the clinical trials are a bad approach for these situations. The problem is that they're really, really hard situations. This is what is so often ignored in these situations. Criticizing the slowness of the FDA is disengenous without producing an alternative approach and identifying both Type I and Type II error rates for that approach. Most of the time, Type II errors are all you see discussed, but that's only half of the equation.

Yours is a gross oversimplification and a mashup of starkly different issues. Are there some medical issues that can be treated like a quality control issue in a car factory? Sure. Are there doctors who chronically wing it and apply "Ole Doc Jones" wisdom and thereby malpractice patients who have readily identifiable illnesses with widely agreed upon treatments? Sure. But that's not what we're talking about here. Did you know before today that there are many types of cervical cancer with striking genetic differences? ( ) Nothing has slowed progress against cancer more than the idea that cancer causes and treatments can be resolved by the Neyman-Pearson lemma.

The progressives were so heavily influenced by Henry Ford precisely because they wrongly believed that he'd solved the problem of managing complexity through the use of the sort of expert systems they loved. The problem with cancer is that it's far more complex than any project Henry Ford could have ever conceived. Those who keep insisting that there's a way to run cancer care so that "can have your cancer treatment in any color you want it so long as it's black" don't understand the extent of uncertainty surrounding cancer. That uncertainty can only be resolved through insights, flashes of brilliance, from doctors incentivized to recognize patterns in individual patients, hypothesize about the cause, derive a prediction, come up with a cocktail of hope and seeing (in properly advised and consenting adults) what happens.

Does the FDA currently apply more scrutiny to "non-essential" drugs than it does to "essential" ones? Thalidomide, for example, was supposed to treat nausea during pregnancy. That's worth treating, but very few (if any) women are going to die if that drug doesn't make it to the market soon enough. So give it more scrutiny. Cancer drugs, though, that might actually save lives, maybe get less scrutiny. First, because delay may cost lives, and second, because even if the drug turns out to be deadly, the people taking it are already facing the prospect of a greatly reduced lifespan. If you're given 6 months to live, take this drug, then die in 1 month, you've only lost 5 months.

Which cancer treatments has the FDA failed to let drug companies provide to doctors with patients at risk of dying from cancer?

Or Ebola? Is it the FDA to blame for a drug researcher failing to have the production capacity to produce candidate drugs for more than 5 or 6 patients? A few of the limited doses were administered in the US, so the FDA was not the reason more people were not given the drug that was hoped to cure ebola.

Evomela (rejectes the first time), Iressa...

I have few issues with reciprocity so I would be in favor of such a system. It could obviously go wrong, but overall it seems sensible especially since approval standard are fairly stable.

But I don't quite understand the rest. Why is it not possible to test a cocktail of drugs? I don't quite see the difference: you must propose a criteria for trying a specific cocktail and test the effect of the cocktail: why is it differet from testing an individual drug? Is the pool of potential subjects too small? Is the effect too small to detect (how do you know the effect is there if so)?

I must say that it basically sounds like a request to be able to charge patients who participate in studies: wouldn't that be the main difference? I am not totally against that, but would you also argue we should change the kind of studies we perform? If we start charging people for participation in efficacy studies, are double blind studies still doable? Would you even get people to agree to such conditions? Should we abandon double blind randomized studies? I find this last prospect very troubling. Given that most health spending is paid by thrid parties, would you mandate such coverage? Should doctors be liable for malpractice if they recommend a patient to participate in a badly designed clinical trial?

I tend to agree with Rayward above, we need more and better data. Doctors are not statisticians, they should not be free to design their own pet clinical testing framework and sell it as real science. There is different specific knowledge required to design a study which produces truly valuable data. Allowing every doctor to design his own flawed study could be worst than doctors performing no study. A false positive is something you have to spend resources to disprove. Too many of those and you halt rather than increase knowledge.

"I tend to agree with Rayward above, we need more and better data. Doctors are not statisticians, they should not be free to design their own pet clinical testing framework and sell it as real science."

Do either of you guys have any evidence that this is really the problem? Im not saying your wrong, but i have a hard time believing that doctors are uniquely allergic to "real science".

My, limited, experience is that only some doctors even try to come up with their own treatments and those that do team up with specialists to cover the holes in their expertise.

Neither of them have any idea what they are talking about. They are just trying to intuitively reason through the problem, trusting that their general rational thinking will solve any problem, so long as they reason through it.

Medical and social sciences are extremely complicated fields. They need much more rigor than hard sciences where you control and understand the experiments to a much larger degree. They are just harder than physics and engineering in some ways.
But don't read too much in what I am saying: I am not claiming doctors are uniquely allergic to "real science". It's the environment they operate that makes it so much harder. Even ignoring the complexity of the issues, you also need to consider the unique psycological circumstances: it is an incredibly stressful profession. You have real people dying. People you might know and like.
The current regulatory system is not perfect and we should consider any concrete proposal in light of whatever evidence we have. Saying there is a prize for saying no and no prize for saying yes is no evidence in my mind. Comparisons in harmful drug approval rates between different countries are evidence that stronger reciprocity is likely beneficial. Without supporting evidence, I would be in favor of trying targeted studies on lowering the burden of proof. E.g. maybe pick some areas and experiment there. Blanket changes with no evidence? I would not be in favour. I don't believe random changes from the status quo are likely to lead us to better outcomes.

so... no on the evidence then?

In general terms, would it work out better if the head of the FDA were someone whose experience is in a field where progress has been terribly slow (e.g. oncology, dementia, ...) or one where progress has been brisk?

There are two big objectives in FDA testing: efficacy and safety. Perhaps efficacy standards can be relaxed a bit, but safety standards must not.

If efficacy tests are relaxed by the FDA, the assessment of efficacy is a responsibility that falls on medicine practitioners: doctors. But, there are interesting stories around the American Medical Association. Only a few years ago, the AMA went after chiropractors for quackery and they lost. In more recent times the AMA is toothless against quacks like Dr. Oz on TV.

Assuming the FDA relaxes efficacy tests, what will replace it? Quackery is the origin of the FDA. Perhaps there could be a "new drug insurance" similar to construction insurance. If some damage is done by the doctor proposing the new treatment, damages or fraud can be "quickly" compensated without a long and complicated legal battle for consumer restitution.

I think the problem here is that the group of people pushing for the relaxation of FDA efficacy standards never talks about how to streamline the process of paying back consumers after a fraud is uncovered.

Tabarrok is a believer in the magic of markets, including the market in drugs: the market will separate the wheat from the chaff, providing the best possible result at the least cost.

It seems we forgot too soon about the "coral calcium" guy. The quack sold fraudulent cancer treatments and weight loss programs, later claimed insolvency to avoid paying restitution to consumers. A judge ordered consumer restitution on 2009 and the FTC is still collecting money from the quack

Yet Alex is not pointing to the fantastic results of free market unregulated no FDA African health care.


Perhaps because there are other relevant variables like GDP?

This is fine. Except that regulation has it's own effects.

Efficacy means that the drug has to have an effect on the condition that the piece of paper accompanying that application says that it does.

Safety means that the drug has no unknown side effects that aren't written on the piece of paper accompanying the application.

So the value added skill that a pharmaceutical company needs is to present the right piece of paper along with the application.

Those pieces of paper cost something like $500 million dollars.

And we are surprised that drugs are expensive, they are over prescribed and that there are lots of corner cases or broad swathes of necessity that aren't pursued. Antibiotics for example.

We are in a rare moment of history when all government programs are in response to the perverse side effects of another government program.

Of course those pieces of paper contain unique information, information describing an empurical understanding of how safe abd effective the drug is. All of this information is valuable to other workers in the field.

So hardly just pieces of paper

So, when you get some serious problem, say bleeding from your gums, you immediately fly to Africa where there is no FDA or AMA to prevent you from paying for the absolute best treatment for diseases like ebola? Or malaria, dengue, ....?

Yes medical tourism is a thing. No people do not go to Africa, for incredibly obvious reasons which do not include a lack of an FDA. Much easier to go to another place with much better facilities and doctors that also does not have an FDA

As a physician and oncologist, I am extremely skeptical of most calls for FDA reform. Could this large government bureaucracy made to be run more efficiently in some areas? Undoubtedly. But the issue is that “reform” so often implies a loosening of FDA standards.

Loosening of FDA approval standard is not what medicine needs. Repeat.

Consider two hypothetical categories of drugs. Class A are so effective that they will clear any bar of efficacy that the FDA might care to impose. Class B are effective enough to gain FDA approval after a loosening of standards, but not enough to do so under the current standards.

I do not want to be in the position of treating my patients with class B drugs.

FDA standards too permissive at present; they should not become more so. Under current approval standards for oncology drugs, already I cannot look my patients in the eye and tell them that I am confident that recently-approved-drug-X will benefit them. The benefits of many of the cancer drugs currently getting approved are so marginal that we are already questioning whether they are worth the side effects and the out-of-pocket expense. The drugs that go through “accelerated” (read: less rigorous) approval will be even more marginal.

Absolutely, the FDA ought to do a better job of incentivizing the development of those drugs I’ve called class A. Emphatically, lowering the bar for approval will not do that. The treatments that cancer patients need are those that will clear any bar, not those that slip in between the space between our current (already low) standards and lower ones.

I consider myself a libertarian, but my observations about how the drug industry responds to the current regulatory environment makes me very skeptical about its ability to set its own standards in ways that align with public health.

Tabarrok has no idea what he is talking about, but he is stubborn, like most libertarians.

'Tabarrok has no idea what he is talking about'

Please, the Bartley J. Madden Chair in Economics at the Mercatus Center knows precisely what is to be achieved, regardless of whether he knows what he is talking about.

Africa has the health care system Alex advocates. No FDA or AMA gatekeeper. No Obamacare. No NIH picking only loser medical researchers.

Totally patient directed and paid medical care.

Africa must have the best health care in the world.

You must be a complete idiot. Do the pharma companies, after failing to get FDA approval, nevertheless produce their drugs for sale to the African market? Or would that be an economic catastrophe for them given the inability of African patients to pay for the treatment?

Great point. I think it's important to consider the willingness of the medical community to prescribe new treatments approved under lower approval standards. My guess is they won't be as gung-ho as the Peter Thiels of the world.

That's another important part of this issue!
I wouldn't WANT to prescribe these drugs, but I would be ethically and legally obliged to do so.

You are not the only doctor in this position, but this is true for all effective drugs.

You know you don't have to be omnipotent, you can tell your patients what you just told us, don't worry plenty of doctors already do this including quite a few oncologists. My mother does this and she treats people who are occasionally psychotic. (You want to talk about side effects for uncertain gain take a look at psychiatry) It helps bring about this thing called trust.

As to the rest of it, you don't have to trust the industry in any environment, plenty of dangerous drugs got through approval, you have to keep up and pay attention to how the drugs work, they often do so very differently in different people. If I say put à well here I have to tell the customer how it can turn out to be a useless expense, if my colleague advocates a new planting system for ag he has to admit that there is no way of knowing if it will work for sure or even lead to disaster. You are trained and paid to understand you don't KNOW shit compared to what shit you know.

Maybe we don't have to trust the industry, but we are obliged to trust the data that industry produces: clinical trials.
We need that clinical trial data in order to convey to patients, as you put it, the shit we know and the shit we don't know. When we ask industry to produce less (and/or less rigorous) clinical trial data, then that hampers the degree of information we are able to provide to our patients.

Imagine how the scenario plays out. A patient's cancer has grown despite treatments A, B, C, and D. A new drug E has just gotten approved. The crappy clinical trial data I'm given shows a marginal benefit in a few, rare subgroups of patients. I say "I doubt this drug will be of much use to you." The patient asks, "are you SURE it won't help?" Being honest, and a scientist, I say, "Well, there is a small chance that it might help, but not much."

End result: the patient asks me to prescribe the drug. Based on crappy data.

Which is exactly the result the manufacturer wants.

In this instance the patient should instead be restricted to no choice at all? I don't understand why you need the FDA to increase the rules, to prevent you from being able to prescribe it, so that the patient can't ask for it. I'm also not a doctor, and I'm assuming you're a bright guy, so what am I missing?

So I appreciate the genuine question, and I'll try my best to give a respectful response.

As I stated in my initial post, I am very libertarian-leaning and very, very "pro choice" as it applies to most things in the world. My guess as to why I tend to lean a little bit "anti choice" here (I admit that while less choice is certainly not the GOAL of increased FDA regulation, it certainly is an implication of it) has to do mainly with the type of choice environment I see in the practice of medicine, which I'd guess is quite a bit different from what the more anti-FDA-regular people (including Alex, I assume) imagine it to be.

If we had a hypothetical world in which, for every drug, doctors "knew" how well it did or did not work for each different condition, and were able to discuss that with patients, then I think I would feel quite differently. I think that is the world that Alex and others imagines? But this is not the case, unfortunately. Right now the only mechanism by which we get any information at all about the drugs we use is clinical trials, which are an FDA requirement.

I have heard, and I get, the argument that in the absence of FDA requirements, the drug industry would have to organically come up with some data-generating body to compare their new drugs and "prove" they are better than their competitors. I get it because (while I am no economist!) I think I do have a basic idea of how Smith's invisible hand works to translate self-interest into general benefit in other industries. But I don't think it would here. Mainly because of how doctors think, and because drug companies know how we think.

Unfortunately, the standard we docs often apply isn't "I know is this drug will work," it is "I have a non-zero expectation that this drug might work." Especially when a patient in front of you is dying of cancer, you don't need much evidence to try something. How little? All the time we prescribe stuff (approved for other illnesses) that has shown some anti-cancer activity in, say, rats. Never had a clinical trial in humans. But, it MIGHT work. I think if we got rid of FDA clinical trial requirements, we would immediately be getting new drugs that have proven anti-cancer activity in animals, or in small, low-quality human trials. It's really expensive to test a new cancer drug in 2,000 people, and if companies could get doctors to use a drug without doing that, you bet they will.

I'm imagining two scenarios. Both start with 5 hypothetical cancer drugs that have shown great promise in rat models.
1: The FDA requires that all go through clinical trials. Only 1 of them lengthens life in humans. As a result, I strongly recommend that drug, and that drug only, to my patients.
2: No human clinical trials are conducted. Therefore, I explain to my patients that all 5 of these drugs looked great in rat models, and there is a chance that any of them might help with their cancer. They would therefore be reasonable to try any one of them.
In which case does the patient have more "choice"?

Obviously my examples are extreme and hypothetical. But hopefully that conveys my fears about a post-FDA world.
My mind is open to being swayed. But I'd need arguments other than "the drug industry will produce high-quality clinical data out of their own self-interest." It won't.

If the patient is required to pay $100,000 out of pocket because that's the profit the drug company wants but the government and insurers look at the test data and see no reasonable cost-benefit justification, the patient without $100,000 has a choice?

Why doesn't the patient just fly to Africa and pay the $100,000 to buy the drug there? No FDA. No AMA. No Obamacare insurance subsidies or regulations.

Interesting perspective Aaron. Seriously. I am genuinely asking for your thoughts so please forgive the devil's advocacy...

What about heterogeneous treatment effects?

Tx A vs Tx B can be boiled down to their respective effect sizes at a population level making Tx A look dominant. But the composition of those effects might be extremely different. What do we do to make sure that Tx B remains viable for consideration for ongoing research as well as for providing physicians more options in the future.

Moreover, if Tx B ends up as viable in the long run, then we do need to again think about incentives. How many of the possible "Tx B"s go unexplored because of low expected probabilities of FDA approval? I.e., the culture of 'why invest unless it's a homerun' has long term implications.

(Disclosure: I truly have no idea if the FDA should liberalize. I truly don't know enough to know what margin we are at. But at a conceptual level, I think these points are worth discussing.)

These are interesting points that I hadn't thought about directly before.
I'm rephrasing this in my mind as, "if we set the bar so high enough that only one drug clears, what if the 2nd place drug also would have been good in some situations?"

My sense is that, at least hypothetically, a drug company could rationally make this decision as a result of FDA regulations. If they would need to test Tx B in, say 100,000 to statistically prove an effect, even if they expect there is one, then the cost might be prohibitive. So they go with similar Tx A and continue to develop that instead, since they project having to test it in only 1,000 people. Makes sense.

However, I kind of see similar incentives in place with either a much higher or much lower FDA bar. Seems like as long as you have to go about gathering clinical data in any way, it's always cheaper to do it in the drug that looks like the surest thing. So what kind of environment would best foster us pursuing more Tx B's? I'm not sure...maybe more of a grant/prize based payout system? I feel like as long as companies are going by what they can get to market the cheapest, the Tx B's will remain on the wayside.

Thanks. It's hard to design an optimal incentive system when the probabilities are so unknown. (expected efficacy, effectiveness, approval (FDA, EMU etc.) and future prices/revenues.

To be clear, I was lamenting the general challenge society faces, not necessarily the prize option which is certainly on the table.

So then I take it you also are in favor of banning off-label prescription, which accounts for what, 50% of all prescriptions written in the U.S.? Is this medical malpractice for these doctors (no doubt including you) to be prescribing drugs that could not even clear the low bar of FDA approval for their prescribed use?

Oof. That's difficult.
I would say that in practice, its often sensible/ethical to go ahead and prescribe something off-label, where a clinical trial hasn't yet been done to "prove" benefit. But, ideally, this situation should not exist in perpetuity - if there is something that docs are prescribing something for off-label, the we need to do the dang trial!
If the evidence then showed that the off-label use actually was not helpful (my educated guess is that this would be the case in most, though certainly NOT all, off-label uses), then yes, after that point it would become malpractice to continue to prescribe them.

Why would a doctor prescribe off label?

He throws darts at a board of drug labels?

He gets a big kickback from the drug makers for every prescription he writes that results in a given drug being paid for?

His patient asks for a prescription for the drug based on an ad or something on the Internet?

Surely not from extensive double bilnd testing and research paid for by a charitable for profit NASDAQ listed drug company and published for free for a generic drug already approved by FDA and available for $10.

If a Federal grant pays for a double blind study for an off label drug, should that drug sell at cost when prescribed for such treatment, or should it be public funding of research with high private profits for successful taxpayer funded research, with the high profit drugs paid for by taxpayers?

Not to be an a$$ but your post comes with an appeal to authority which, in my experience, is potentially unwarranted. Oncologists around here (Houston-ish) mainly just oversee the administration of drugs/infusions which themselves are typically selected by a team at MD Anderson. Oncologists, save in the detection of side effects and recommendations for alternatives given patient response, rarely are the ones either doing the research or hypothesizing about what regimen/dose/sequence is the initial best bet for the individual patient (except in "routine" cancers, e.g. prostate).

As to your assertion, it's little wonder that you can't say whether a new drug will benefit an individual patient because you don't know; and you don't know because it was approved solely on the basis of nudging the survival curve of a bucket of vastly varying clinical trial subjects slightly to the right (iff p<.05). It may well be that the old drug would have put your patient in to the long tail of the right side whereas the new drug is going to put her in the long tail of the left side but because the overall population mean got shifted rightward you are incentivized to become her executioner. You can tell her it's been thoroughly tested but you know deep down inside that genetics, epigenetics, immune status, socioeconomic status, etc. mean that she'll be the first person with her type of cancer to ever receive the drug. That everybody, especially the FDA, insists on pretending otherwise, is what I'm railing about/against.

You're not being an as, but keep in mind there are older, cheaper protocols for treating cancer that are not as efficacious as more modern treatments, by "X" percent, where X is a small but statistically significant number, and what's interesting to me is that insurance companies will insist that patients use the "older, cheaper" treatment. If you had cancer, would you wish to use an older protocol that has 35% cure rate for your stage, and costs peanuts to the insurance company, or the new fangled treatment that has a 45% cure rate? Insurance company says use the old treatment, as "you're likely to die anyway", but ah, there's the rub!

I hope to be staying on the "adding some local knowledge" side rather than the "appeal to authority" side of the line. Sorry if I'm not.

Thatanos - you are absolutely correct about the difference between "average" and "individual" treatment efficacy. Not being able to view the counterfactual, we are never able to know for sure whether a given treatment will help an individual patient. In many individual cases, we will actually do harm. Chemotherapy is poison, after all.

However, at least I am able to say, on average, that treatment X moves the survival needle for people with cancer Z, even a little bit. As limited as an amount of information as that is, that is the status quo that I am trying to hold on to! Of course we can never "know" for an individual, but all of medicine is just playing the odds, and we need to know that we are at least at 50/50. The future I'm worried about is the one in which we don't even know that drugs improve survival in humans, but just that they shrink tumors in rats. When I talk about "not knowing" it is this level of truly absent data that I am talking about - not being unable to see the future for an individual patient.

Shorter AaronM: you can't cure cancer very well today, anymore than 50 years ago, so I, as a physician, hate to bring up false hope to my patients with non-FDA approved drugs, which have marginal benefits. Fair enough.

I'm not a physician (but play one on the internet) and as I said in another post, the traditional way of killing cancer is to zap it with powerful drugs and radiation that kills all cells, but kills cancer cells, which are more incomplete and fragile, faster than normal cells. This hasn't changed in a half century. What's changed is more precise delivery of the radiation dose, so depending on what stage you're in you don't have to treat the whole body.

They do have "holistic" cancer treatment centers that rely on non-FDA approved treatments, but their efficacy is so suspect you have to get a CAT scan every couple of months to make sure you are cancer free. However, cancer often being very 'fragile' they do work to put cancer into remission a certain percentage of the time. And some people will live a long life after cancer goes into remission.

What we need, says I, playing the role of a patent guru now, is reformed patent law that gives incentives to a team of engineers and scientists to come up with a cure for cancer. At present, the money goes to the employer, and you have a mere year within any public disclosure to file for a patent, which is unrealistic (not how innovation works), and, worse, only 'one person (s)' is rewarded with a patent, when in fact often, like in the invention of radio, there are many co-inventors, often working independently from one another or borrowing and improving ideas from one another. They all need compensation, even the independent inventors who simultaneously invent the same thing as another. More money for invention will increase innovation. After all, unless you are an altruistic Good Samaritan or fame seeker, why should you care if you come up with a cure for cancer? Your employer will reap all the monetary benefits. One reason I quit science to go into being a gate-keeper, and it's worked out well for me (I'm in the 1%, and made about a million dollars on my own, the rest from my parents, intestate share).

I'll stop here as I could go on with a few pages on how to reform patent law.

Ray - I am in complete agreement about the need to reform the patent system. I'm a firm believer that a system will produce the results that it sets the incentives for. Right now, the incentives we have in place a leading mostly to a lot of mediocre drugs that do little move the needle, albeit with an occasional true breakthrough (imatinib for CML is a good example, though don't get me started on its upward price trajectory over the last 15 years...).

I'm sure there are much better "from scratch" systems to go about discovering a testing drugs, as a society. If we could summon the muster to make this kind of change, I'd be all for it. But that would involve a LOT of shaking-up. If all we did, though, were to weaken FDA requirements while leaving the rest of our current system in place, I think the result would be a downward trend in the quality of drugs that get produced (not that you were necessarily arguing otherwise...but that's kind of where the whole discussion started).

AaronM January 25, 2017 at 9:50 am

Consider two hypothetical categories of drugs. Class A are so effective that they will clear any bar of efficacy that the FDA might care to impose. Class B are effective enough to gain FDA approval after a loosening of standards, but not enough to do so under the current standards. I do not want to be in the position of treating my patients with class B drugs.

Except this is a reality. There are essentially Class A and Class B drugs because the standards have been toughened. So if you give your patients a drug from *before* those standards were made harder, it is basically a Class B drug.

Two examples - aspirin and penicillin. Neither would be approved now. Penicillin kills a very small number of people who are allergic to it and aspirin kills hamsters so it wouldn't even get to a clinical trial. That is, they are Class B drugs that would not pass the modern FDA requirements to become Class A drugs.

Do you prescribe them to your patients? Should you?

Will we even have drug review in six months?

I said earlier, as we discussed Obama executive orders, that President Trump would have legal authority to gut the EPA. I wouldn't complain that it was illegal. I just said it would generate a reaction.

So, that is in full swing.

Why not the FDA? Alex, are you ready with your "wait a minute, some review is good" essay?

And the reaction seems to be strong.

Or perhaps the darker interpretation is that Alex knows Trump has gone full anti-science and STILL wants to strike while the iron is hot.

A lot of cancer drugs are already approved on weak evidence through accelerated approval. How much lower does the bar really need to go? It's no like there a real 'cures' that are being held back by burdensome FDA regulation.

How the hell do you know??

I'll back him up. A LOT of weak evidence. Not just cancer drugs, either. You can Google for the recently approved drug in Duchenne Muscular Dystrophy as a case-in-point.

Do I get to know if I have an intractable cancer before answering the question?

For some reason my early post didn't go through. I'm trying this one again and have deleted hyperlinks which may have been a problem.
To paraphrase a movie title with that starred the underrated Suzanne Pleshette, "If it's Wednesday, we must have another blog post on the FDA." Now let's start out by acknowledging that Vincent DeVita is a hugely important figure in the field of oncology and responsible for some ground breaking treatments that have saved lives. Let's also acknowledge that his book "The Death of Cancer" is a wonderful polemic, and there is nothing wrong with that. Lots of great thinkers have published polemical tracts. Yes, DeVita is correct that the FDA 'was' a highly conservative and under resourced organization for a number of years in the post-Thalidomide era (as an aside, let's also affirm that Kelsey's action or inaction depending on your view point averted a huge tragedy in this country). The resource issue changed in 1992 with the approval of the Prescription Drug User Fee Act that provided FDA with new resources in return for defined performance goals. This Act has been reauthorized every five years with increasing metrics and improvements (FYI and FWIW, I was on the industry negotiating team for the first four rounds of PDUFA). Additionally, there have been a number of changes to FDAs regulations that allow for fast tracking of new drugs for life threatening indications. We are at a point, as I made both yesterday and on Monday that important new drugs that have good clinical data get approved by the FDA in six months or less and that these are usually first in the world approvals. What's not to like about that???

To address the second set of points that Alex quotes. "We can design drugs that will hit a specific target, and by being so specific, they have fewer side effects." Maybe and maybe not. This was the mantra of the biotechnology industry at the outset. Human proteins can now be produced outside the body and they will provide safer modes of therapy. LOL, just listen to the litany of life threatening side effects at the end of TV commercials for biotech drugs. Why just today there was a report that the Hepatitis C drugs are showing some unexpected side effects that require liver transplants (something these drugs were supposed to prevent. Anyone who is versed in drug safety will be the first to tell you that a full safety profile for new drugs is not known for some years (if ever) after approval (Cox-2 inhibitors one of the more recent examples).

DeVita is of course correct that a lot of successful oncology treatments involve multiple drugs. Companies are also aware of this and various combination products do get commercialized. There is nothing in the FDA regulations that prohibits investigators form studying multiple drug therapies for various cancers. If one looks at the NLM Clinical Trial database ( one can see how vibrant this field is. Numerous scientific papers get published each years on the advances of research in this area and one only need to attend a American Society of Clinical Oncology meeting to see what is going on. I would be more worried if the number of presentations were decreasing rather than increasing.

I wonder if DeVita can point to a promising oncology drug whose development was stopped because of "FDA hurdles?" FDA approves oncology drugs that have not so good safety profiles as they treat a life threatening disease. As far as I can tell in looking at pharmaceutical industry pipelines, oncology research is alive and well.

Intrepid MR readers of this topic might want to spend some time reading Derek Lowe's blog on the pharmaceutical research pipeline. One of his recent posts on drug safety evaluation might convince some of the libertarians here to reconsider "safe drugs for all" but I doubt it.

Enough for a Wednesday morning. I'm looking forward to Thursday's chapter in this never-ending quest.

> I wonder if DeVita can point to a promising oncology drug whose development was stopped because of “FDA hurdles?” FDA approves oncology drugs that have not so good safety profiles as they treat a life threatening disease. As far as I can tell in looking at pharmaceutical industry pipelines, oncology research is alive and well.

You have to look at 'dogs that don't bark' as well. The environment for FDA approval may prevent marginal/'out there' approaches from being investigated in the first place, if they come with a billion dollar price tag.

'The environment for FDA approval may prevent marginal/’out there’ approaches from being investigated in the first place, if they come with a billion dollar price tag.'

It isn't the FDA stopping drug companies from investing the money - it is their shareholders, who only care about the company's bottom line.

FDA behavior changes drug company behavior in a way that is sub-optimal, so we should change FDA behavior.

I'm guessing your comment above is setup for 'NATIONALIZE THE EVIL PROFIT DRIVEN DRUG COMPANIES' rejoinder

The problem with the FDA approval of new drugs is that the shoe is on the wrong foot. The FDA should have a time limit in which a new drug can be proved to be safe and effective or not after the period of time without proof of the drug being either unsafe or ineffective the drug is passed for use. There could be additional safeguards as needed but not endless delays which themselves have proven to be deadly for seriously ill people.

Are you saying that if a drug maker fails to provide good test data that proves a drug safe and effective, the drug gets approved? Ie, a drug maker submits a study reporting "drug given to one test subject, subject died", and after six months the drug is automatically approved for sale?

Because we want no government takeover of drug testing, the private sector must pay for drug testing, and the FDA simply reviews the test studies and data and scientists make an assessment of the research quality, and the the FDA administrators vote to approve or not approve.

For example, the scientists reviewed an application to sell Plan B over the counter, and the safety and research data provided, recommended approval based on the research, and the FDA administrators voted to reject it because Plan B was contrary to religion, illegal abortion, violated chattel, or some other reason that ignored the research data and analysis showing it safe and effective. Political pressure changed the vote.

Seriously, do you often hurt yourself jumping to conclusions?

I think what I'm saying was clear. Set a time for the FDA to approve the drug and stick to that time unless they can prove it is harmful or ineffective OR find some kind of red flag. People die for lack of critical drugs while the FDA fiddles and often takes years and years to approve one of these drugs. So I say put on the FDA to do their job. Shit or get off the pot. To quote Yoda "No. Try not. Do... or do not. There is no try."

'Occasionally I have been told that FDA reform is something that only a few libertarian economists support. '

You really need to get out of your bubble. Lots of people support reforming the FDA. However, most people do not consider reform equal to gutting the FDA to enhance the pursuit of profit.

'The thalidomide episode sent the message to those who worked at the FDA that the way to do right by people was to say no. '

One one assume that an oncologist is also familiar with the scam that was laetrile (and since one wikipedia link is as good as another - )

'But there are no awards for getting a good drug quickly into the public domain.'

Of course not - but there are massive profits to be made by taking 'public domain' drugs (those no longer under patent protection) out of the public domain. Such as this deal, restricting distribution - 'In accordance with Shkreli's business plan, Turing acquired Daraprim (pyrimethamine), a medication approved by the FDA in 1953,[54] for US$55 million.[55][56] on August 10, 2015, from Impax Laboratories.[57] The drug's most prominent use as of late 2015 was as an anti-malarial[58] and an antiparasitic, in conjunction with leucovorin and sulfadiazine[59] to treat patients with AIDS-related and AIDS-unrelated toxoplasmosis.[60] The patent for Daraprim had expired, but no generic version was available.[61] The Turing–Impax deal included the condition that Impax remove the drug from regular wholesalers and pharmacies,[58] and so in June 2015, two months before the sale to Turing was announced, Impax switched to tightly controlled distribution.[18] In keeping with its strategy for pricing in the face of limited competition (see above), Turing maintained the closed distribution.[52] The New York Times noted that the deal "made sense only if Turing planned to raise the price of the drug substantially."'

Of course, one could blame the FDA for actually caring about GMP, but really, let us not stray into the realm of libertarian fantasy when it comes to how the pharma industry has historically cared about anything but profit unless regulated. (Laetrile being a fine example of this point.)

'After spending millions of dollars developing a drug, a company may be forced to abandon it for lack of efficacy, when, if approved, it would be another exciting tool for clinical investigators who want to explore combinations of targeted therapies in post-market research trials. '

Doesn't seem to have stopped Sovaldi (admittedly, not a cncer treatment), as noted in wikipedia - 'Sofosbuvir's U.S. launch was the fastest of any new drug in history.'

My guess is libertarians are tilting at this windmill because no one feels a need to defend it. The constituency for eliminating the FDA or limiting it in a meaningful way is the readership of Reason Magazine. So, it is a safe target for libertarians. But, it does bump up against another question I saw on a blog the other day. Would an unfettered market in health care result in better care? It would result in cheaper care for things that exist, but would we see a great leap forward in cancer cures or dementia cures?

The answer is probably no, because we know so little about human biology and the diseases that attack us. Something like the FDA, which mostly prevents medicine from killing more people than they currently do, is not a great impediment to discovery. The great leaps to come will be from genetics, not pharmaceuticals. So, the FDA playing gatekeeper is not slowing the advance of medicine.

The FDA is just as involved in regulating gene-based medicine.

"Something like the FDA, which mostly prevents medicine from killing more people than they currently do, is not a great impediment to discovery."

Begging the question?

We understand genetics.

The 'great leaps to come' will flow from our better understanding of life at the molecular level. We're almost twenty years since the elucidation of the human genome [Human Genome Project].

We're only getting started.

I just don't understand why the FDA can't be simply advisory and not a gate keeper? Just have the FDA do whatever testing they deem necessary, but let responsible grown up adults make their own decision as to whether the risks are acceptable. Why are so many busy-bodies wanting to control the decisions of other folks? Yes there are lots of dumb folks around who might make bad decisions, but we let them drive, raise kids and choose their own food.

All 3 of those things are easier than evaluating medical efficacy and safety.

And all three are regulated to most citizens' satisfaction.

Should Trump eliminate Child Protective Services?

msg - if people are not confident in their ability to determine whether something is effective or safe then they can turn to a trusted authority to tell them. What you want to do is not allow people the freedom to make their own mistakes when all the consequences are on themselves. I have no problem with insurance (either government or privately provided) requiring only drugs mandated by FDA, but, like for sex, people should be allowed to make their own decisions as to what they do to themselves.

You are arguing the government pay for non-FDA approved drugs, pay for only FDA approved drugs, or not pay for any drugs at all because the drug user with Hep C should get treated only if he has money, and everyone needs to pay to test everyone they might have contact with for Hep C to prevent getting Hep C themselves?

If free markets work better than government regulated and funded markets, the Africa must have the best health care in the world.

So here's a little story. Once upon a time researchers had this great idea for ovarian cancer. Suppose there might be something on ovarian cancer cells but not so much on normal cells. If they could find this something and make a drug that could attach a little 'flag' to that something, the immune system could be tricked into killing those cells.

Sounds simple. However they went through 125 possibilities. On the 125th try they noticed a protein that was common on most ovarian cancer cells but less common on normal ones. They called it CA-125.

However the immune therapy never worked. Instead as a consolation they created the CA-125 test, which is an ok but hardly perfect test for tracking ovarian cancer and has been around since the 80's.

The purpose of this is to illustrate that developing these treatments seems to be a huge undertaking and even if the scientific logic appears airtight, drugs often don't work just as often as do. Also only a few drugs work as clearly as antibiotics might work. A lot of the time it is only a small portion of the population (sometimes only those with a particular mutation but other times it's just a portion of a cancer population with no clear way to tell who those people are). Other times you are talking about a very marginal improvement in survival overall, which to a doctor with lots of patients doesn't look very impressive but a small median improvement may mean a huge improvement for a few patients and a modest one for a few others.

I put these stories out there because they seem to lead me too:

1. Newer drugs are likely going to require thousands of different people working to discover and bring to market.
2. How do you actually tell if these things do anything without very controlled studies? If a drug cures 99% of a type of cancer it is pretty easy to see it works. If it increases progression free survival by 10% it also works *but* it is not going to be easy to see that by just giving it randomly to handfuls of different patients.
2.1 This becomes even more difficult if you're dealing with a cancer that sometimes goes into remission using a standard of therapy. If chemo puts a cancer in remission 30% of the time if you're not doing controlled trials what does it mean if a doctor does chemo plus new drug with 4 patients and all go into remission? Does that anecdote say the drug works? But it could just as easily be he drew the lucky 30% four times in a row.
3. Combinations of drugs seem to be more common. Hep C is 'cured' by a drug combo. A drug combo seems to be able to keep HIV in check more or less for life. But again if the drugs are nearly useless absent the 'perfect combo' it seems pretty hard to tell how you're going to ever tell that without really controlled studies.

Yet the arguments here seem to be that we could spur a lot of new drug discovery if we forgo these trials. How is that going to work?

That being said I could support some reforms like reciprocity (if the EU approves a drug the US will automatically do so). But are there a huge number of drugs approved in one respectable region but not approved in the US?

Wait - I thought Joe Biden shot the moon and cured cancer! Isn't that why people kept saying there in October and November that HE should have been the Democratic candidate, because of his track record in science leadership?

Diabetes drug might fight cancer

I am a 79-year-old lady. My Hepatitis B (HBV) disease appeared at the age of 74, with no hope of a cure from the hospital i looked further for an alternative treatments, a friend of mine told me about Health Herbal Clinic in Johannesburg South Africa who sell herbal treatments for diseases including Hepatitis B disease, I contacted the herbal clinic via their website and purchased the HBV herbal remedy. I received the herbal remedy through DHL couriers within 8 days and i immediately commenced usage as prescribed, i used the herbal remedy for about a month and 1 week, my condition has greatly improved, all my symptoms including Abdominal pain, Nausea and vomiting, Loss of appetite, I am HBV free! contact Health Herbal Clinic via their email healthherbalclinic(AT)gmail(DOT)com or visit www(DOT)healthherbalclinic(DOT)weebly(DOT)com PLS HBV is not a death sentence, there is a cure!

Comments for this post are closed