Betting on Clinical Trials Will Not Fix Bad Clinical Trials

When I first read the idea that Kalshi planned to let people bet directly on clinical trial outcomes, my immediate reaction was: Wow, this is a terrible idea.

I run cancer clinical trials at my institution, and I could immediately see how this might create some serious ethical problems. It just did not pass my sniff test.

But…gut reactions are not always right, so I spent more time thinking through the arguments in favor of it. I was also surprised when I read Vinay Prasad MD MPH’s support for the idea.

To be fair, Dr. Prasad makes some valid points. Our clinical trial system needs serious reform. I agree that some studies use weak or outdated control groups, although I see this far less often in modern trial design. And some trials do seem designed more to produce a positive result than to answer the question patients and physicians really care about.

Prediction markets might expose some of that.

Kalshi has also put some reasonable safeguards around its initial pilot. The markets are limited to late-stage trials, they do not open until enrollment has closed, employment verification is required, and people with material nonpublic information are prohibited from trading.

But I still think this is not the way to fix it.

Why would we rely on strangers placing bets to pressure companies into designing better and more ethical trials? That responsibility belongs to regulators, ethics committees, investigators, sponsors, journals, and health systems that open these studies. If a control group is inappropriate, it should be challenged long before the trial opens. It should not require a betting market to expose the problem after patients have already enrolled. The market might embarrass a sponsor or influence a future trial. It cannot fix the study those patients are already in.

A market can estimate whether a trial will meet its stated endpoint, but it cannot tell us whether a control arm is ethical, whether an endpoint is clinically meaningful, or whether the outcome will actually improve our patients’ lives. Strong early data may make us fairly confident that a treatment works while important questions remain about the size and durability of the benefit, toxicity, quality of life, overall survival, or which patients are most likely to benefit.

Clinical equipoise is much more complicated than 50/50 betting odds.

I also think Prasad is partly right about the stock market. A small biotech company with one lead product can function almost like a wager on one clinical trial. But the fact that people can already profit indirectly from clinical trial results is not a great reason to create an easier way to do it. Expanding that financial incentive does not somehow make the system more fair or ethical.

Clinical trials involve a huge network of people: investigators, research nurses, coordinators, pharmacists, statisticians, vendors, sponsors, and referring physicians. Many of us notice small pieces of information along the way—more safety/toxicity letters, unusual protocol amendments, patients remaining on treatment longer than expected, or a lack of communication from the sponsor. Being in the clinical trial arena, I see this every day. While no one person may know the final outcome, once money is attached, this creates a new incentive to gather those fragments, or share them, or exaggerate them, or bet on them. Employment verification and insider-trading rules may lower that risk. I do not think they remove it.

Of course, we hope most people will behave responsibly. But Dr. Prasad has argued for years that our healthcare system is distorted by financial incentives and that people often respond predictably to them. That is exactly why his support for this surprises me.

Systems should be judged, at least in part, by the behavior they encourage.

Creating a direct financial incentive to collect, share, exaggerate, or trade on information from an ongoing clinical trial seems likely to create more of the behavior he has spent years warning us about

One of the biggest risks of all is confusing patients. A market showing a 78% chance that a trial will “succeed” does not mean an individual patient has a 78% chance of benefiting. It only means the study is likely to meet a statistical endpoint. The actual benefit could be small, based on a surrogate measure, or outweighed by toxicity. Whether the result is clinically meaningful—or improves overall survival, something Dr. Prasad has appropriately emphasized for years—may not be known for quite some time. This distinction is already difficult to explain, and I am not convinced adding betting odds makes informed consent any easier.

Patients join clinical trials because they trust us. They accept uncertainty, inconvenience, and sometimes serious risk because they hope the treatment may help them, but also because the knowledge gained may help people who come after them (altruism). We should be extraordinarily cautious about building a new financial market around what happens to these patients.

I agree that we need to address weak trials, misleading endpoints, and inappropriate control groups. But the answer is to do the hard work to fix the system responsible for approving and running them, not outsource part of our ethical oversight to a betting market.

After thinking it through some more, I still land back where I started: this still feels like a terrible idea.