Latest / The Clinical Realist / The Economics Nobody Taught You in Medical School

Transcript

The Economics Nobody Taught You in Medical School · The Clinical Realist

• 0:01
  SPEAKER_00: There's a question I get from health tech founders more than almost any other question.
• 0:05
  SPEAKER_00: They have a product and it works.
• 0:07
  SPEAKER_00: And they have peer-reviewed evidence.
• 0:09
  SPEAKER_00: They have a pilot that proved ROI and they can't get a health system to buy it.
• 0:14
  SPEAKER_00: And they always phrase it the same way.
• 0:16
  SPEAKER_00: They say, we don't understand why this is so hard.
• 0:19
  SPEAKER_00: And I do.
• 0:20
  SPEAKER_00: And today I'm going to explain it.
• 0:22
  SPEAKER_00: The answer is not about your product or solution.
• 0:24
  SPEAKER_00: It's not about your evidence.
• 0:26
  SPEAKER_00: It's not even about the decision maker you're pitching to.
• 0:29
  SPEAKER_00: The answer is about the economic structure of the American healthcare system.
• 0:33
  SPEAKER_00: And specifically about a problem that has no equivalent in any other industry.
• 0:37
  SPEAKER_00: The complete separation of the person who uses a product, the person who pays for it, and the person who decides to buy it.
• 0:43
  SPEAKER_00: So once you understand this, you will understand everything.
• 0:47
  SPEAKER_00: Why AI fails to scale, why procurement once you understand this, you're going to understand everything.
• 0:54
  SPEAKER_00: Why AI fails to scale, why procurement takes two years, why a physician can love their tool and a CFO can still kill the deal, and why value-based care is harder than anyone planned for.
• 1:04
  SPEAKER_00: Why RBUs exist and what they do to clinical behavior.
• 1:08
  SPEAKER_00: That's an ugly one too.
• 1:09
  SPEAKER_00: So this is medical economics, and nobody taught it to you in medical school.
• 1:13
  SPEAKER_00: So let me start with a principle that governs every market on earth except healthcare.
• 1:18
  SPEAKER_00: In a normal market, the person who wants something, the person who buys it, and the person who pays for it are all the same person.
• 1:23
  SPEAKER_00: You want a coffee, you buy it, you pay for it.
• 1:26
  SPEAKER_00: And boom, the transaction's clean.
• 1:28
  SPEAKER_00: The feedback loop is immediate.
• 1:30
  SPEAKER_00: So if the coffee's bad, you stop buying it and the market corrects.
• 1:33
  SPEAKER_00: Healthcare doesn't work that way.
• 1:35
  SPEAKER_00: In healthcare, especially in the US, you have three separate actors.
• 1:39
  SPEAKER_00: First, the user.
• 1:40
  SPEAKER_00: That could be the patient, that could be the doctor, it could be the person who received the care.
• 1:44
  SPEAKER_00: Depends on what you're selling, right?
• 1:46
  SPEAKER_00: Second, the decision maker.
• 1:48
  SPEAKER_00: And that's the physician or the committee or the hospital administrator who decides what care gets delivered, what technology gets used, what AI tool gets piloted.
• 1:58
  SPEAKER_00: And third, the payer.
• 2:00
  SPEAKER_00: That's the insurance company, the Medicare, Medicaid, or the employer who's self-insured.
• 2:04
  SPEAKER_00: The payer is almost never in the room when the decision is made.
• 2:07
  SPEAKER_00: So these three actors have very different incentives, different information, different time horizons, and they're frequently in direct conflict with each other, without naming that conflict out loud.
• 2:19
  SPEAKER_00: So let me give you a concrete example.
• 2:21
  SPEAKER_00: Say a health system is evaluating an AI tool that can identify sepsis, say six hours earlier than current protocols.
• 2:28
  SPEAKER_00: The evidence is pretty clear.
• 2:30
  SPEAKER_00: Earlier identification saves lives, great.
• 2:32
  SPEAKER_00: It also reduces ICU length of stay by 1.4 days.
• 2:36
  SPEAKER_00: All right.
• 2:37
  SPEAKER_00: So the physician loves it.
• 2:39
  SPEAKER_00: And earlier identification means that she could intervene before the patient even deteriorates.
• 2:43
  SPEAKER_00: So that's a clinical win.
• 2:45
  SPEAKER_00: But the CFO is worried.
• 2:47
  SPEAKER_00: Reducing ICU length of stay by 1.4 days means 1.4 fewer days of ICU billing per patient.
• 2:55
  SPEAKER_00: So depending on the payer mix in the hospital's margin structure, that might be a revenue reduction, not a revenue gain.
• 3:01
  SPEAKER_00: And it's not because anyone wants patients in the ICU longer, but because the financial model was built around the existing length of stay.
• 3:09
  SPEAKER_00: So the patient who's the user has no vote in this conversation.
• 3:12
  SPEAKER_00: Their insurance company, who's paying, has every incentive to want shorter stays, but it's not in the room either.
• 3:17
  SPEAKER_00: So this is the economic structure of American healthcare and explains more about why health technology fails to scale than any other single factor.
• 3:27
  SPEAKER_00: So I want to talk about something that shapes every clinical decision a physician makes, whether they know it or whether they don't.
• 3:32
  SPEAKER_00: And it's called the Relative Value Unit, the RVU.
• 3:35
  SPEAKER_00: The RVU system was created by Medicare in 1992 to standardize physician payment.
• 3:40
  SPEAKER_00: And the idea was really straightforward back then.
• 3:42
  SPEAKER_00: Assign a numerical value to every clinical service based on the time, skill, and overhead required to actually perform it.
• 3:48
  SPEAKER_00: And this would make sure that you pay physicians based on the RVUs they generate.
• 3:52
  SPEAKER_00: Sounds simple.
• 3:53
  SPEAKER_00: But what it actually created was a production incentive system that's driven physician behavior.
• 3:58
  SPEAKER_00: So health system staffing decisions are also decided by this, and technology adoption patterns have also been this way for over 30 years.
• 4:07
  SPEAKER_00: So here's how it works in practice.
• 4:09
  SPEAKER_00: A surgical procedure might be worth 20 RVUs.
• 4:12
  SPEAKER_00: A 20-minute office visit might be worth 1.5 RVUs.
• 4:16
  SPEAKER_00: And a 45-minute conversation with a patient about managing a complex chronic condition might only be worth two RVUs.
• 4:24
  SPEAKER_00: So the procedure pays 10 times more per hour than the conversation, even though the conversation might prevent the procedure and produce substantially better long-term outcomes.
• 4:33
  SPEAKER_00: And this is not a moral failing of physicians, no.
• 4:35
  SPEAKER_00: It's a payment system designing behavior.
• 4:37
  SPEAKER_00: So when your income depends on RVU production, you produce RVUs.
• 4:40
  SPEAKER_00: And when the system rewards procedures over prevention, you get more procedures and less prevention.
• 4:45
  SPEAKER_00: And it's not because anyone's trying to do it, it's just how the system's working right now.
• 4:49
  SPEAKER_00: So when coordination, communication, and care management are valued at a fraction of technical intervention, you get a system that's structurally based toward doing things rather than preventing things in a fee-for-service world.
• 5:02
  SPEAKER_00: Now, say you bring AI into this.
• 5:04
  SPEAKER_00: An AI tool that reduces redundant imaging, streamlines documentation, or catches duplicate medication orders is reducing work.
• 5:12
  SPEAKER_00: And in a straight RVU system, some of that work was generating revenue.
• 5:16
  SPEAKER_00: An AI tool that reduces time per patient by 20% sounds like efficiency.
• 5:22
  SPEAKER_00: But in that same RVU model, it might actually mean each physician can see more patients in the same hours.
• 5:27
  SPEAKER_00: But it might also mean the health system needs fewer physicians to generate the same RVU volume.
• 5:32
  SPEAKER_00: Nobody says this out loud, but every CFO is running this calculation.
• 5:36
  SPEAKER_00: And this is why physician adoption of AI is not a technology problem, it's an economics problem.
• 5:41
  SPEAKER_00: So the physicians who are supposed to use these tools are operating inside an incentive system that wasn't designed with AI in mind, and nobody is reconciling that conflict before the pilot starts.
• 5:52
  SPEAKER_00: So for the last 15 plus years, the solution to the RVU problem has been called value-based care.
• 5:58
  SPEAKER_00: I know there's lots of opinions on this, and again, people will pick their sides.
• 6:02
  SPEAKER_00: But the idea is pretty elegant.
• 6:04
  SPEAKER_00: Stop paying for volume and start paying for outcomes.
• 6:07
  SPEAKER_00: So if a physician keeps a patient healthy, it pays them for health.
• 6:12
  SPEAKER_00: It stops paying for procedures and starts paying for prevention.
• 6:15
  SPEAKER_00: And I'm not here to tell you value-based care is a bad idea.
• 6:18
  SPEAKER_00: I think the direction is actually right.
• 6:19
  SPEAKER_00: I'm here to tell you why it is so much harder to execute than the architects plan for.
• 6:24
  SPEAKER_00: So first, problem one, outcomes take time and payment cycles do not.
• 6:28
  SPEAKER_00: A health system that invests in a diabetes management program today might see the benefit in reduced amputations, hospitalizations, and dialysis over five to ten years.
• 6:38
  SPEAKER_00: But if you're the CFO writing quarterly projections, that's tough.
• 6:42
  SPEAKER_00: And the physicians are being evaluated sometimes not necessarily on a full value-based contract, but still kind of RBU productivity.
• 6:51
  SPEAKER_00: So the AI vendor will need a 12-month ROI story to close a contract.
• 6:55
  SPEAKER_00: So the time horizons aren't aligning.
• 6:58
  SPEAKER_00: Problem two, attribution is generally quite difficult.
• 7:03
  SPEAKER_00: If a patient stays healthy for three years, who gets the credit?
• 7:07
  SPEAKER_00: The primary care physician who managed the diabetes?
• 7:09
  SPEAKER_00: The health system that deployed a remote monitoring tool?
• 7:12
  SPEAKER_00: The insurance company that covered the medication?
• 7:14
  SPEAKER_00: The social worker who helped them navigate housing instability?
• 7:17
  SPEAKER_00: So value-based care models require you to answer this question precisely because money follows attribution, and that answer is almost never very clean.
• 7:26
  SPEAKER_00: And then problem three, the risk is not equally distributed.
• 7:30
  SPEAKER_00: In a fee-for-service model, the insurer bears the financial risk of a high cost patient.
• 7:34
  SPEAKER_00: In many value-based care models, the risk shifts to the provider in the system.
• 7:39
  SPEAKER_00: And a physician group that takes on a high-risk patient population can be financially penalized if that population has high utilization, even if the utilization was medically appropriate and the care was excellent.
• 7:50
  SPEAKER_00: So this creates incentives to avoid complex patients, which is the opposite of what value-based care was supposed to do.
• 7:56
  SPEAKER_00: So I'm telling you all this, not because the healthcare economics are insolvable.
• 8:00
  SPEAKER_00: I'm telling you is because if you're buying health technology, selling it, deploying it, or advising on it, you need to understand the financial water your buyers are swimming in.
• 8:10
  SPEAKER_00: So the deal that makes clinical sense may face an economics objection that has absolutely nothing to do with your clinical evidence.
• 8:17
  SPEAKER_00: The pilot that proves ROI may still not get purchased because the ROI lands in a budget category that's not the one your champion controls.
• 8:25
  SPEAKER_00: So let me bring this to the practical question.
• 8:27
  SPEAKER_00: If you understand the medical economics, how does it change the way you think about clinical AI adoption?
• 8:34
  SPEAKER_00: For health system buyers, the first question may not be, does this AI work?
• 8:38
  SPEAKER_00: The first question is where does the economic benefit land and who controls that budget?
• 8:42
  SPEAKER_00: An AI tool that reduces nursing documentation saves money in the nursing budget.
• 8:47
  SPEAKER_00: But if nursing is a cost center and the budget savings don't flow to the department that champion the tech, you're gonna have a hard time sustaining the initiative after the pilot ends.
• 8:55
  SPEAKER_00: Map the money before you map the workflow.
• 8:59
  SPEAKER_00: If you're a health tech company, your ROI story needs to account for the payer structure of your target customer.
• 9:04
  SPEAKER_00: A tool that generates savings under value-based care contracts is a very different pitch to the system that is still predominantly fever service.
• 9:12
  SPEAKER_00: A tool that reduces emissions is a win for a captain plan and potentially a revenue loss for a health system that depends on inpatient volume.
• 9:19
  SPEAKER_00: So know your customer's economic model before you build your value prop.
• 9:23
  SPEAKER_00: And for docs, understanding that you are operating inside an incentive.
• 9:29
  SPEAKER_00: And for physicians, understanding that you're operating inside an incentive structure is the first step to advocate for changing it.
• 9:35
  SPEAKER_00: The physicians who have the most influence over tech adoption at their institutions are the ones who can translate clinical evidence into economic terms.
• 9:42
  SPEAKER_00: And that's not because clinical outcomes are not enough, but because the people who control capital allocation are making decisions in economic language.
• 9:49
  SPEAKER_00: So if you want to accelerate the adoption of tools that genuinely improve care, you have to be able to speak that language.
• 9:56
  SPEAKER_00: And for everybody, the misalignment between user, payer, and decision maker is not going to go away.
• 10:01
  SPEAKER_00: It's the structural reality of American healthcare until the incentive architecture changes at a large policy level.
• 10:08
  SPEAKER_00: But we can't wait for that.
• 10:09
  SPEAKER_00: In the meantime, the most effective strategy is just not to ignore the misalignment, but to build around it.
• 10:14
  SPEAKER_00: Find the deals where the incentives align.
• 10:17
  SPEAKER_00: Build the coalitions that bridge the budget silos.
• 10:20
  SPEAKER_00: Design pilots that produce evidence in the language the CFO needs to hear.
• 10:24
  SPEAKER_00: So this is what clinical AI strategy actually looks like in practice.
• 10:28
  SPEAKER_00: Not just deploying tools, navigating economics.
• 10:32
  SPEAKER_00: So I started this episode with a question that health tech founders ask me all the time.
• 10:37
  SPEAKER_00: They have a product, it works, and they can't get it bought.
• 10:41
  SPEAKER_00: The answer is not your product.
• 10:42
  SPEAKER_00: The answer is that you're trying to sell something in a market where the person who actually benefits is not the person who pays.
• 10:48
  SPEAKER_00: The person who decides is not the person who uses.
• 10:51
  SPEAKER_00: And the system that governs the whole thing was built around a payment model that actively resists the efficiency you're probably trying to create.
• 10:58
  SPEAKER_00: And that's not a reason to give up.
• 10:59
  SPEAKER_00: It's a reason to get more precise.
• 11:01
  SPEAKER_00: So know whose budget you're saving, know whose incentives you're disrupting, and know where the risk lands and who's burying it.
• 11:08
  SPEAKER_00: Build your value proposition for the economic reality of your buyer, the one they're living in now, not the clean ROI model in your pitch deck.
• 11:16
  SPEAKER_00: So medical economics is not a side course.
• 11:19
  SPEAKER_00: It's the operating environment for every clinical decision, every technology purchase, and every conversation about what it means to improve care.
• 11:26
  SPEAKER_00: And now you know it.
• 11:27
  SPEAKER_00: If this episode gave you a framework you needed, share it with someone in your organization who's making technology decisions without it.
• 11:33
  SPEAKER_00: If you're working through the economics of a specific AI deployment and want a thought partner who's been on both sides of the table, my link is in the show notes.
• 11:40
  SPEAKER_00: And I will see you next Wednesday.
• 11:41
  SPEAKER_00: Thanks for being with The Clinical Realist.