When healthcare is decided by algorithms, who wins?

I’m starting this with a personal story. My mother was a New York City public school teacher for many years. When she joined the school system, part of the deal was that, when she retired, many of the costs of her traditional Medicare plans would be subsidized by her union and by the city. So far, so good. However, now the city, in order to save money, is moving all its retirees, including the public school teachers, to a Medicare Advantage plan. 

(If you don’t know what Medicare is, or the difference between traditional Medicare and Medicare Advantage, don’t worry about it — a lot of people don’t. I’ll explain in a moment.)

A lot of city retirees are not happy about this switch — and, in fact, have been fighting this in court for the last couple of years. Why? Because, among other things, Advantage plans give health insurance companies much more power to deny coverage — and those denials are being based on predictive algorithmic tools rather than medical personnel.

To understand Medicare Advantage, it might help to know a bit of background. (Stick with me; this is stuff you hopefully will need to know about eventually.) Medicare started out as a government-run health insurance program that was established in 1965 to help fund care for people over 65 who were no longer being covered by employers and were usually considered too much of a risk by private insurers. It was funded by taxes pulled from employees’ paychecks and matched by employers. And with today’s atmospheric increases in the price of healthcare, it has become a necessity for most of today’s older people.

Like all government programs, Medicare is, to say the least, complicated. You think you have trouble doing your annual taxes? Try figuring out how to deal with Medicare. There’s Medicare A, which handles hospitalization, Medicare B, which handles payments to doctors (and which has an annual fee), Medicare D, which handles medications (and is also not free) and several other for-pay alphabetical Medicare programs. But even with all the complications and extra fees, Medicare means that, if you’re lucky enough to last past 65, you should be able to afford to go to a doctor and get care.

Eyes glazing over yet? Wait — now we’re going to get into Medicare Advantage and its algorithms.

As you might imagine, health insurance companies are not fond of traditional Medicare. Although they handle Medicare B and other for-pay aspects of the program, they are limited by governmental regulations and rules as to how much they can charge for services and how much power they have over doctors’ recommendations for care. So in 1997, Medicare Advantage (also known as Medicare Part C) was created.

Medicare Advantage means that a private insurance company is handed control of all parts of your Medicare benefits — the hospital part, the doctor part, the drug part, and all the other parts. Advantage has, well, advantages, at least at first — it costs less to you than the government programs, is easier to deal with (because it’s a single entity), and there are all those really cool advertisements showing active, carefree gray-haired people golfing, vacationing, hugging their grandkids — oh so happy that they handed over the responsibility for their healthcare to Big Health Insurance, Inc.

However, because an insurance company running your Medicare Advantage program has more control over payments, it can demand that you — and your doctors — get advance approval for pretty much anything short of a checkup. And since the fewer procedures it has to pay for, the more profit it can keep, there’s a strong incentive to deny as many procedures as possible.

According to a recent report by the medical journal Stat, insurance companies have been using these algorithmic tools — rather than doctors or other medically trained people — to determine whether patients who are enrolled in their Medicare Advantage programs are worthy of care. These tools are being used, according to the report, “to pinpoint the precise moment when they can shut off payment for a patient’s treatment. The denials that follow are setting off heated disputes between doctors and insurers, often delaying treatment of seriously ill patients who are neither aware of the algorithms, nor able to question their calculations.” Since appeals challenging those denials can take months or even years to wind their way through the various steps needed, some of those appeals can last longer than the patient. Which certainly saves money.

And the advertisements for these tools are not shy about what they are meant to do. A pamphlet for nH Predict, one of the products used for this purpose, states, “Using the nH Predict tool, case managers now determine first if lower levels of care can meet patients’ clinical needs, and then recommend higher levels of care on an individual basis when medically necessary.” 

However, it now seems that the case managers mentioned in that quote are apparently saving time by passing on the “individual basis” part and taking the tool’s recommendations at face value. Last year, the Center for Medicare Advocacy issued a special report that stated, “Although most AI-powered decision-making tools claim to offer only recommendations that are not intended to substitute for clinical or medical judgment or for Medicare law, in the Center’s experience, users often implement the tools’ recommendations without any critical examination of their impact on patients.”

In other words, if your doctor feels you need a procedure to, say, prevent you from having a stroke, and your insurance company’s algorithmic tool pops up saying that you don’t need that procedure (and so the insurance company can deny payment for it), you have a couple of choices. You can pay for the procedure, go into serious debt, and hope that you and your doctor can successfully challenge those denials. Or if you can’t afford to pay on your own, you can wait and hope that your appeals will come through before you have that stroke.

Or there’s always GoFundMe.

It’s not easy to be the caretaker for an older adult or parent. It’s going to get a lot worse if, when my mother needs medical care in the future, I (and her doctor) will have to battle an algorithm as to whether she deserves the prescribed care. I’m not looking forward to it. And I’m not the only one dealing with this issue. There are many, many people whose parents are in Advantage plans and who may one day be informed that, according to the judgment of an unnamed statistical tool, the treatment that their parents’ doctor says is necessary — is not.