Universal Coverage Mighht Be Nice, but an AI Tax Is Necessary

I was amused – oh, I should be polite and say “interested” -- to see a new study, led by researchers from Yale School of Public Medicine, about the benefits of a universal single payor health system. It concluded that we could save 100,000 lives annually and save some 1.04 trillion each year – some 20% of our health care spending. What’s not to like? I’m sure Bernie Sanders is already drafting the bill.

AI needs to give back, especially for health. Credit: Microsoft Designer

The savings come from five sources: using Medicare payment rates for all providers, using “international reference pricing” for pharmaceuticals, reducing administrative costs to Medicare’s levels, reducing fraudulent billing (“consistent with the experience of other single-payer transitions”), and reducing emergency room visits and hospitalizations due to improved access to primary care.  Good goals, all.

Credit: Pandey, et. al. 

Steffie Woolhandler and David Himmelstein, among others, have been making these or similar arguments for decades, and they are not without merit. It is shameful that we don’t have universal coverage. It is distressing how much money we spend on healthcare. It is embarrassing that we spend so much money on administration.  It is maddening that so many people don’t get the care they need, get the wrong care, or get their care in the wrong places/at the wrong times.

We could do better, we should do better, but, if anything, we’re doing worse: more people are losing coverage, more providers are going out of business, our rates of chronic (and some infectious diseases) are going up, and we’re dying sooner.

I want to quickly point out some of the problems with the proposed sources of savings, then discuss other courses of action that might lead to these or even better outcomes.

• Medicare payment rates: yes, a lot of money could be saved by using Medicare payment rates, but I doubt you would find many providers who would say they could survive. They make their money on private insurance rates, are lucky to break even on Medicare rates, and lose money on Medicaid. This one is not going to happen.
• International pharmaceutical reference pricing: first, I’m not sure such a thing exists. It is true that drug prices are typically lower in other countries. Both President Biden and President Trump seized upon this, with some signs of modest success. But, as with the Medicare pricing, it would be a shock to the pharmaceutical industry to have prices slashed across the board, wiping out trillions of dollars of value and, oh-by-the-way, eventually reducing investments on new and better prescriptions.
• Administrative costs: as a percentage of spending, Medicare’s administrative costs are lower than private insurance, but that is partly due to Medicare spending per capita being so much higher. Also, costs incurred by other agencies – e.g., Social Security or the IRS – are not always counted. But certainly the complexities of so many plan designs by so many health insurers while tracking the current eligibility of everyone is a cost that is much higher than it should be.
• Reducing fraudulent billing: I mean, really: do people really think that Medicare does a better job of reducing fraudulent billing than United Healthcare or Anthem, much less than other countries?
• More primary care: reducing emergency room visits and hospitalizations has been the goal of countless private health insurance efforts, such as disease management or chronic health programs, and the track record has generally been underwhelming. But the real problem is – where are we going to get all the primary care physicians to handle all the underserved people?  

So, much as I agree with the goals, count me a skeptic that single payor is going to magically make everything better.

Here’s where I inevitably turn to AI. An article by Alex Janin in The Wall Street Journal marveled at how “AI Is Helping Patients Solve Medical Mysteries.”  Ms. Janin writes: “AI can be especially adept at flagging potential rare and hard-to-diagnose diseases, which may otherwise go undetected for years because doctors don’t often see them.”

That’s the kind of use AI advocates have been promising for years, and it is exciting to see this use finally bearing some fruit. For the small percent of patients with these kinds of diseases, AI can literally be a lifesaver, but let’s remember that they are a small percent. When I read the article, I keep thinking about bigger problems I want AI focused on. E.g.,

·       Flagging fraudulent and/or duplicative billing;

·       Identifying both unnecessary tests and procedures and the providers who most commonly perform them;

·       Identifying providers who deliver sub-standard care.

Want a more efficient/effective healthcare system? Let’s start there. The savings potential may not be as gaudy as Yale’s $1.04t, but these would not require as massive an upheaval.

While I’m at it, I want to bring up another AI-related area of healthcare. The not-so-hidden but too-little discussed secret of U.S. healthcare is that we have a lot of third world outcomes, largely in lower socioeconomic households and disproportionately impacting people of color.

Sure, we can put in single payor, but will that solve the problem of rural Mississippi or south side Chicago? Too many people don’t have access to clean air, clean water, enough food, adequate shelter, or accessible/affordable healthcare. The great lesson of 20^th century U.S. healthcare was not the gains from new medicines or more hospitals/physicians, but in public health efforts like improved sanitation and more immunizations.

So where are our investments in 21^st century public health? Do we want to make marginal improvements in the health of the middle/upper income households, or dramatic improvements in lower income households?  I suspect I know what this Administration would say, and they’re wrong.

By every measure of income inequality or social mobility, we’re in a have/have not society, and there is every reason to believe AI will make that so, so much worse. It’s going to be NAFTA but much worse. But I always remember: NAFTA didn’t cause all those jobs to go abroad. Those jobs went because U.S. CEOs chose to send them abroad, in order to make them and their stockholders richer. Think they won’t do the same with AI?

That’s why I firmly, fiercely believe we need some sort of AI tax to help make the adjustment to the new AI world. The financial gains from AI need to be broadly distributed, and one of those distributions has to be for addressing our third world health outcomes. That could be through 21^st century public health investments, and/or through some sort of universal basic income (UBI).  

Universal coverage might be nice, but an AI tax for public health and universal basic income might be necessary.