This is the first in what will be a series of posts discussing how new medicines are introduced and regulated in the United States, and how the status quo could be improved. As will be established over the course of the series, the current system is slow and leads to poor outcomes for patients.
Why Medicines Are Different
The choice to use a product ideally should be left with the individual consumer. But for some products, it can be difficult for a consumer to know if the product will do what is expected or, as in the case of some medicines, the consumer may not even fully understand what the product will do.
It thus makes sense to test invasive and potentially risky products like medicines thoroughly before they reach the market; harms from new products can and do occur. But by nature, testing products before they reach the market delays their introduction. The question therefore becomes: how much delay is warranted, and how much is overkill? Are the health gains from regulating these substances more or less than the health costs of the regulation?
Risk of a New Product vs Risk of No New Product
Rather than focus solely on the possible hazard of the product itself, we must look at its place within the market of existing products or, if none exists, the space within which a new product might fall. There are risks to new technologies, but they often displace older and greater risks. At a basic level, we must be able to try new things, so that those that improve our lives can sustain. To prevent or slow the introduction of new products denies or delays development.
The U.S. Food and Drug Administration (FDA) regulates medicines in the United States. The agency demands vast troves of data from animal studies and clinical trials before it will allow physicians to prescribe a medicine to patients. These data are meant to ensure the safety of products. But as the economist and risk expert Sam Peltzman explains, regulators tend to focus on the risks posed by new products (Type One errors) so much so that they often ignore, or at least downplay, the risk of not using the new product (Type Two errors).
Where there is uncertainty, should we rely on experts, or should we leave the decision with the individual? This is a crucial question. The expert opinion of, e.g., FDA staff scientists, physicians, scientific researchers at universities and companies, and journal editors all should undoubtedly inform our decision making, but where possible, individuals should make the decision.
Currently, researchers and physicians are important in providing data, but they are secondary to FDA committees, because the FDA regulates what companies are required to do by law. Without that legal threat, the notion goes, companies might not provide all of the necessary data and might cut corners to get drugs to market. But relying on the FDA slows everything down, and that comes with its own costs. In future posts in this series, I will spend a lot of time detailing the delays, which can amount to many years (and, in some cases, a decade) for new medicines.
Advances in almost any scientific field are usually a result of incrementalism, or trial and error. Not every conceivable trial is attempted, but physicians and scientists are more likely to undertake the predictably safest ones with the greatest potential returns. This is the process of muddling through. Small moves facilitate action, while reducing to small amounts how much is unknown.
By contrast, comprehensive decision making (such as normal FDA approval) requires the abrogation of uncertainty by demanding vast knowledge of consequences before action is taken. As Bartley Madden explains:
[I]ncrementalism has weaknesses – it may not be possible, for instance, in the midst of vast numbers of interactions, to observe the consequences of small moves. Harm may result from even the smallest of trials. But thanks to the work of Herbert Simon on the cognitive limits of rationality in the 1960s, incrementalism is both a description of how choices are made and a prescription of how they ought to be made.
Incrementalism Is Largely Outlawed from Drug Taking
Decisions about whether to take speculative drugs—that is, those that have not been through a decade of trials—are largely taken away from the individual by both the FDA and drug manufacturers, who do not want to be sued in case of failure. Companies are also concerned that uses of products outside of approved clinical trials may skew data and limit the chances of FDA approval. FDA staff also want to protect their turf and control the determination of what is safe and effective.
Mary Douglas, Aaron Wildavsky, and other social anthropologists have developed methods of analyzing how people view the world and the risks within it, as well as how people use information to further their own aims. Douglas and Wildavsky could have been writing about FDA approval processes when they asserted that groups like “to be panic stricken about dangers from technology…because they serve their own moral purposes.”
Pro-regulatory voices point to occasional lapses in oversight in service of demanding the use of existing regulatory structures to slow and ultimately deny the approval of new products. Proponents of command-and-control regulation understand that, by focusing on a new product’s absolute (rather than relative) risks, they can deny its approval. While existing products may be more harmful than new ones, regulators will nonetheless ignore new products’ relative risk.
The experience of the COVID-19 pandemic helps to illustrate how regulators can prioritize control, rather than risk aversion. If the FDA had followed the usual rulebook on medicines, we would still be waiting for COVID vaccines. Yet, vaccines were introduced within a year to almost everyone in the United States. Indeed, while authorities were alarmist about the disease’s dangers, possibly to ensure that everyone took the vaccine, they arguably downplayed—and, in some instances, ignored—any risk posed by the vaccine.
While U.S. authorities approved the COVID vaccines rapidly, this was, on balance, a huge success. In other words, the COVID vaccine approvals vindicate a far faster approval process. Products could be available in months, not in multiple years or decades.
As was clear with COVID vaccines, while there are risks with new technologies, they often displace older or greater risks. Myocarditis is a serious risk for a few of the younger people who were vaccinated, but societally, the benefits of approval massively outweighed the risks. Wildavsky says it’s “better to have trial and error than trial without error.”
The COVID vaccine processes were a massive example of trial and error that worked. Yet, in most other cases, moderate risk aversion means some breakthroughs will never reach patients. Trial and error will not be allowed. And the result is what Bartley Madden calls the “invisible graveyard”—those who die quietly from disease while not receiving a treatment that might have saved them.
Precaution Is Warranted, but Can Be Dangerous
Large double-blind clinical trials that demonstrate significant benefit from a drug are the surest way to know that it is, indeed, safe. But the cost and delays of that undertaking are significant. Prior to major regulatory changes in the 1970s, it took seven months to approve a new medicine; by 1998, it took more than seven years.
In addition to the data and risk assessment that trials yield, the primary reason they are undertaken is to provide political and legal cover if something is missed and harms are later found.
Part of the issue with the gold standard of trials is that it is a formula that participants learn to follow close to the letter. Sometimes, the formula diverts attention from the best available knowledge—following the formula is given a higher priority than anything else. As Bartley Madden explains: “Before you know it, the formula itself predetermines the result.” After all, if it is too onerous and costly to follow the formula, the result is no new breakthrough.
It is here where COVID provides the best example of violating the formula to huge advantage. Almost every normal rule, every part of the formula, was broken.
Perhaps we should apply this approach to most medical breakthroughs. At every step, we must evaluate what available data show. If that means changing the status of products (abandoning them, approving them, continuing to study them), then that is what should happen. If we don’t, the message we send is that the process is more important than the data, or the circumstances in which patients find themselves. It is true that, in most trials, data are continually evaluated by the scientists involved, but it is not common for the trial to be changed entirely given interim results; the formula dictates the outcome, rather than up-to-date information.
Challenge Trials: COVID Shows a Way Forward
Rapid small clinical trials could provide a way for faster drug approvals. In principle, testing a medicine on 100 people over a few months might provide very useful information. In cases where there is no existing treatment or where a patient might die while waiting, promising results in a small trial could lead to rapid but tentative approvals. Of course, the physician and patient must understand the risks involved, but their burden of understanding and risk-taking is increased under this scenario.
With the COVID vaccines, a small number of healthy individuals with a very low chance of mortality from COVID, and no knowledge of any downside, took an experimental vaccine that ended up harming them. Surely, very sick patients with no treatment options should be allowed to try a medicine with some chance of success, even if the risks of treatment are significant. The feedback of data from use by these sick patients would then add to the small trial data and increase knowledge rapidly. This should encourage other researchers and their financial backers to act similarly.
Scarred by History, or History Avoided
It is said that Germany has an institutional, societal, and even emotional hatred of inflation. The economic destruction caused by the hyperinflation of the 1920s meant that every subsequent chancellor of Germany has fought inflation with a passion, and has been supported in this endeavor by the nation’s citizens. A century since hyperinflation, it’s still a part of societal memory.
While the United States never really experienced thalidomide babies, the fear of them certainly hangs over American drug regulators, much like inflation does more broadly in Germany. The FDA’s rejection of thalidomide is folklore, and the agency’s greatest safety selling point. Yet, as I will discuss more thoroughly in my next post, thalidomide’s problems were not found in clinical trials. They were found when it was used for purposes for which it was not originally tested. Ironically, thalidomide shows the other risk of following a formula and having the government decide what is important. Not only does it slow approvals of new drugs by specifying vast amounts of trial data, but it also misses actual problems found outside of the formula.
The FDA focuses primarily on new medicines. It sets the rules on clinical trials, and ensuring that companies follow those rules leads to the introduction, if slowly, of new medicines. But once a medicine is approved, the job of oversight is not done. Clinical trials may not find the problems at a population level, simply due to a lack of original participants. Also, certain groups might have problems with a new product (patients with comorbidities, or older or very young patients, or members certain racial groups who can and do react differently to some medicines). If such individuals were not represented or were underrepresented in trials, the problems may not be seen at all until there is more widespread use.
Hence, it is important for post-market surveillance to be undertaken. This can find problems, as well as benefits. If a physician and patient decide to try a medicine for a disease or condition for which it hasn’t been approved, problems may result, but so may benefits. This feedback is extremely important, whether the news is positive or negative.
The lesson from thalidomide should be that we must place greater value on post-market surveillance, and not on more extensive clinical trials. Real world feedback is important, but it is not among the FDA’s priorities. As future posts will detail, I undertook surveys of oncologists and primary-care physicians to understand their use of new medicines. One question pertained to FDA post-market surveillance (PMS). Most (57%) found the process slow, difficult to use, and irrelevant.
Part of the importance of PMS is to ensure that companies continue to make products consistently well. Much of the reason for extensive clinical trials is distrust of the companies that introduce medicines. One reason Congress gives the FDA a large budget is because many believe that, without FDA oversight, drug companies might allow nonperforming medicines to be sold to patients.
But the same skepticism on corporate morality is missing when it comes to continued manufacturing standards. After all, it is easy to cut corners on production quality. If oversight is missing, poor quality might well result. This is especially the case where production takes place in locations with limited regulatory capacity, and where the products are mainly for export (notably, from India and China to the United States). While the FDA does undertake plant assessments in these nations, it often visits each plant only once a decade. Its oversight is therefore weak, at best, and useless, at worst. As Dinesh Thakur explains, many Indian drug firms have routine production problems that result in substandard medicine.
Yelp for Medicines?
One issue that came up fairly consistently in my interactions with physicians was that feedback loops for existing medicines must be improved and sped up. If the FDA isn’t up to the job, maybe we should bypass it.
One physician suggested a “Yelp for Medicines,” which would allow for fast, real-time feedback. There might be vindictive reviews, or reviews from those with commercial interests (either pro or con), but a method could be found for physicians to respond rapidly. This could help to find the next thalidomide or reveal previously unknown benefits. After all, it was feedback like this that led emergency-room physicians to know what mix of medicines and procedures to use with critical COVID patients.
There are obviously other reasons why drug approvals are so slow, and one of them is litigation. The threat to companies and physicians of acting faster and against the status quo discourages the rapid use of new medicines. Overhauling the liability rules would be an important part of improving drug approvals.
As medicines become more specialized via emerging techniques like gene therapy, there will be fewer patients for each new medicine, making large clinical trials both harder to undertake and too expensive to justify for the market for that drug.
Right now, it is hard to imagine wholesale changes to current practices, but patients and patient groups will be critical to any progress. If they demand change, it is possible that it will come. In the meantime, the best approach is to try to improve FDA regulations and oversight. That will be the subject matter of my next few posts.