Bayesian Investor Blog

Problem

The US, and to a lesser extent much of the developed world, has concentrated interest groups (e.g. big pharma), which have incentives to increase medical spending. The main check on pro-expense interest groups used to be patients’ desire to spend less of their money. We’ve carefully eliminated that incentive for most patients. That leaves us with a situation in which spending increases to absorb much of the increase in disposable income.

I originally started writing this post in reaction to Aduhelm’s conspicuously bloated price. But it now seems that the system has enough sanity to avoid major waste there.

I’m also interested in the situation with statins. There’s reasonably good evidence that they saves the lives of a small fraction of people who take statins, but also some reason to doubt that cholesterol best describes what problem they fix (I don’t have a good link for these doubts. Here are some mediocre ones: 1, 2, 3).

I suspect that statins treat a problem which could be targeted much more narrowly if researchers did a better job of identifying why statins work. Not just to save money, but to save many patients from hassles and side effects.

That would likely require expensive research. Statin producers have a strong incentive to avoid that research, and hardly anybody has much of an incentive to devote many resources to it.

A Simple Fix

Here’s a simple approach that we could start with: if a study convinces Medicare / Medicaid to stop covering treatment X for condition Y (presumably via altering FDA approval), the government should pay whoever is responsible for the study the amount that Medicare and Medicaid save.

Patents

My main proposal is a more comprehensive version of that.

I suggest expanding the patent rules to allow a patent that covers a medical professional’s decision to not prescribe a specific treatment.

This approach would include the ability to patent the claim that all currently approved uses of a treatment are worthless. E.g. I could patent the process of always saying no when someone asks whether a patient should be given Aduhelm to treat Alzheimer’s. If someone convinces the FDA that that process of saying no is safe and effective, then the patentholder should get the money that the US government would otherwise have expected to pay for Aduhelm.

For something that’s sometimes beneficial, such as a statin, I expect the patent would describe new tests, or new combinations/interpretations of existing tests. I.e. each time a medical professional uses the test to decide not to treat a patient, they would owe the patentholder payment that is presumably a fraction of the money that the treatment would have cost.

The patent approach looks better when most of the spending is via private insurance or out-of-pocket payments. It might also have some value when it involves testing for a condition that the FDA hasn’t yet decided to classify as a disease.

Drawbacks to the Patent Approach

Enforcement will be poor.

Many of these patents would be unenforceable in practice, which most likely means that patentholders wouldn’t put in the effort needed for FDA approval, leaving us with the same result as we have today.

In a normal industry, it would be crazy to expect patentholders to collect much money. 50 years ago, the medical industry was decentralized enough that I would only have made this proposal on April 1. But we’ve got a centralized and ossified set of bureaucracies. A patentholder only needs to investigate a few organizations to get rich. (Is there any risk hope that patenting non-treatment would cause patients to route around the bureaucracies? I suspect not.)

The organizations presumably have some ability to hide info from patentholders. But doing so likely requires some innovation, which isn’t one of their strengths. They likely have many employees with access to the relevant info, and have had below-average success at cultivating employee loyalty. I expect that the bureaucracies would initially mostly pay the appropriate amounts, and find work-arounds over periods of many years, and end up paying patentholders 10 to 40% of the appropriate amounts.

Those incentives for patentholders would be a large improvement over what we have now. They’re not close to optimal, but that’s no excuse for sticking with the current mess.

The need for proposals such as these is, of course, a sign of serious systemic flaws. My proposals here are quick fixes that have some hope of being adopted this decade. They likely bear little resemblance to what dath ilan would implement.