The Voices section is a place for physicians, staff and community leaders to share their perspectives on all things healthcare. Dr. Michael Scahill writes about the intersection of technology and healthcare and offers his perspective on a constantly shifting landscape.
MADERA, Calif. – Seems everyone talks about healthcare costs, especially in the United States, being too high and ever rising. And yet, no one from the federal government to giant powerful corporations to hardworking safety net providers seems to be able to solve this problem. Here’s why and how artificial intelligence (AI) might finally be the key.
Spoiler alert: Healthcare costs rise so fast… because of advances in non-healthcare technology.
Healthcare is Expensive
As a younger man in health economics seminars, I heard that technology was the main driver of rising healthcare costs. That’s true! But not in the way it’s usually presented. The usual story is that it’s the invention of ever fancier and more expensive drugs, MRI machines and maybe even giant, clunky electronic health records that drives up costs. That usual story isn’t “wrong,” but it’s definitely not the main driver of rising costs.
Healthcare (Cost) is Always and Everywhere a Human Resources Phenomenon
A top-of-the-line MRI scanner will set you back something like $3M, so it’s no small investment. But a scanner should last 10 years in accounting terms, hopefully considerably more in real life, so it should amortize out to $300k/yr. Google tells me the average nurse in California earns $133k/yr, so even that super fancy MRI machine is only a little over 2 nurses. With some 5-6 thousand staff (fewer FTEs) of various sorts, Valley Children’s, like every healthcare provider in the US and probably the world, spends more on wages than on everything else combined. If that’s true of high complexity referral hospitals like us with beautiful facilities and expensive MRI machines, you know it’s all the more so in small clinics with rented office space and just a few computers. Wages are often 90% of total costs at such clinics! [1]
To be clear, I am not complaining about this. Healthcare workers, and nurses in particular, have HARD jobs. If anything, I wish they were paid better. But it remains true that healthcare is a service industry. Labor is by far the biggest driver of costs and therefore of cost inflation. Basic economics teaches we’d be foolish to expect staff to accept a massive pay cut, and thus, the only way to cut healthcare costs meaningfully is to make labor way more efficient.
Of Doctors and Barbers
Before we get to an efficient labor solution, all this begs the question: why do labor costs in healthcare keep rising? Perhaps I should start with why I think this is such an interesting question by talking about my haircuts.
A very friendly guy named Daniel at Off the Top Barbering, conveniently located between the hospital and the Scahill Ranch, cuts my hair these days. He’s a bargain, but he charges way, way more than my dad’s barber when he was a young man in the 1960s. I think that’s remarkable because we’re getting the same haircut performed over the same amount of time. At least for the unfashionable, conservative style I rock, the technology and procedure of haircuts hasn’t changed since Wahl invented electric clippers in 1919. So why do I pay Daniel, reasonably priced though he is, so much more than my dad paid his barber?
Most people, to the extent they think about this question at all, probably shrug and say “inflation”. But as you’re reading my blog, surely you’re not most people and demand an answer. To boot, as the excellent graphic tragically since taken down by the New Zealand’s statistics agency points out, haircut prices have, like those in healthcare, risen much faster than inflation could explain. Presumably it’s not because I could listen to Spotify’s Taylor Swift station on a Bluetooth speaker from Daniel’s iPhone when all my dad had was Frank Sinatra on a scratchy radio? [2]
No… but sort of yes. The reason prices rise even in industries where the labor itself doesn’t change for decades is that labor in OTHER industries changes dramatically over the same time. To get a track on the radio, Sinatra needed a full orchestra in an elaborate recording studio with a cadre of sound engineers, a giant vinyl record factory and a small army of agents to distribute his records to radio stations across the country.
While I know nothing about Taylor Swift except that she is (or was?) dating a football player, I know there are plenty of artists today who can produce and distribute their music with little more than a Macbook and a Spotify account. We see this pattern repeated in countless industries from agriculture to satellites – any industry amenable to technological innovation gets way more productive. Meanwhile, barbers and healthcare workers are stuck at the same productivity level as their grandparents. [3]
Cost Disease and Its Discontents
The key point to grasp this problem (and the opportunity it presents) is to realize that healthcare, for all its distracting modern technology, works at the same speed it did in the time of our grandparents. Indeed, in many ways, it’s the same speed since the days of Hippocrates or Galen. At its core, it’s a human conversing with and examining another human. It’s beautiful & poetic, but it’s very expensive in our modern world.
This poetic human experience keeps getting more expensive over time because of a phenomenon coined “cost disease” by the great economist William Baumol who first described it. The intuition is exactly that of the Sinatra to Swift evolution versus industries less impacted by technology. As technology makes many jobs much more productive, all the other jobs that technology does NOT change get more expensive per unit time in order to keep up. Agriculture, big business here around Valley Children’s, is a great example. Since 1948, US agriculture has tripled in productivity! 15 minutes of work on a strawberry patch produces triple the strawberries today – amazing! But 15 minutes with a patient in 1948 is… still 15 minutes with a patient in 2025. Doctors, barbers, bartenders and the like who don’t see productivity gains still “demand” higher wages over time just to keep enticing people to work in those industries. If my high school guidance counselor had told me I would only make 1/3 my salary as a doctor versus a farmer, meds schools would be lonely places. Thus, salaries rise in service industries even when the work does not change. Baumol pointedly called it a “disease” because he recognized what a problem this was then and that it would worsen over time. Without technology that really makes their work more efficient – in the way a combine harvester upgraded productivity over harvesting by hand – services like healthcare get harder and harder to afford.
Will AI, Finally, Cure Cost Disease?
Is there possibly a new technology that could make conversing with a human more efficient? Why, yes there is! I’m hardly the first to notice the potential of artificial intelligence large language models (LLM), of which ChatGPT is the most famous example, finally to change the fundamental economics of seeing a patient that go back to Hippocrates. At Valley Children’s, we’re already using an LLM in this vein. Instead of the provider having to type out a summary of their conversation with a patient, the LLM listens in (with patient consent) and prepares a summary for the provider to finalize, nifty timesaver! But as anyone who’s played with ChatGPT knows, it’s already pretty good at conducting a human-like conversation. It’s not hard to imagine a near future where the LLM can at least start the conversation on the provider’s behalf, and indeed we’re talking to several startups innovating on exactly this approach. We expect human providers to be in the loop on crucial medical decision making and adding the ineffable human touch for a long time, but if these technologies can make care more accessible for our patients and more meaningful for our providers, we’re all for it.
Footnotes
[1] Hospitals & Hotels: This lesson was driven home for me in my grad school accounting class. Our professor gave us a handful financial statements without identifying the firms and asked us to match them to their industry. I figured the hospital on the list would have lots of inventory and hardware (PP&E in accounting jargon), and, boy, was I embarrassed when I, the class doctor, was wrong. The hospital’s financial statements looked just like the hotel’s – a big empty building where the vast majority of ongoing costs went on staff. High acuity referral centers like Valley Children’s definitely have more inventory & PP&E than the standard community hospital, but I still learned my lesson – the economics of hospitals are all about the people working in them.
[2] Inflation: Inflation is an ephemeral topic, but one way to think about it over the long run is as the accumulation of the productivity gains discussed here across the entire economy. Innovation generates productivity gains which generate wealth. More people can afford more stuff. When that stuff is scarce relative to demand, prices rise. This is way beyond my expertise, but I suspect this is part of the reason there was near zero long term inflation for centuries before the modern era. Discounting brief price spikes around bad harvests or coin debasement by broke monarchs, prices were remarkably stable. “You could sell a house, bury the money, dig it out in 40 years, and buy the same house back.” Inconceivable today, even if the house itself had collapsed leaving just a pile of rubble and the plot of land to buy. It’s probably no coincidence prices started to rise dramatically right around the invention of the printing press. Innovation was really hard when you needed to hand copy all the knowledge in the world just to preserve it from rotting parchment. We’re lucky Euclid’s “Elements” and Homer’s “Iliad” made it through the centuries when many texts did not. Just so, perhaps also not a coincidence inflation in the chart above got steeper right around the invention of the computer.
[3] Nordhaus: Economist William Nordhaus won a Nobel Prize in large part for his study of these productivity gains over the sweep of human history. His classic example is the cost of light, which went from extremely expensive (gathering wood by hand and rubbing sticks together) to so cheap we barely think about it. He estimated more than a 300X increase in productivity/decrease in cost just since tallow candles in 1800. Doctors, barbers & bartenders are certainly not 300X more productive than they were in Napoleon’s time. Nordhaus’ work was collected nicely by NBER. Obviously, many technologies in my career alone have had impacts in healthcare, most notably on the scope & quality of care. Catheter clot busting for heart attacks, monoclonal antibodies and now gene and CAR-T therapies have all revolutionized certain treatments in wonderful ways. But none of them change the fundamental limit that a doctor can only see so many patients in a working day.
