Healthcare and Data Science Brings Smart Pricing and the anecdotal evidence that we have gathered suggests that the price-setting process for most hospitals is done using an elaborate set of spreadsheets with a strong dose of gut instinct and conventional wisdom. Moreover, tradeoffs with payers are done frequently (“I’ll give you $1000 back on a hip replacement if you up the reimbursement of a C-section by $1500”). The starting point of the discussion is often a forward forecast of how many procedures are likely to be done by the system within each DRG. Often these forecasts are just as weak, relying on a rather simple this-is-what-we-did-last-year basis. Out of the other end of this sloppy process comes a reimbursement schedule for that provider for that hospital, and the process starts all over again for the next provider and the next. On top of pricing, the contract may also stipulate a range of clauses pertaining to the provider: allowed networks and credentialing requirements to name a few.

We see several problems with the current system of price setting:

1. The tools used in the process are primitive relative to the sophistication that could be brought to bear on such a complex problem.
2. Hospitals are clearly outmatched by payers on analytical firepower.
3. Scenario planning across a wide variety of medical futures is rarely practiced.
4. Prices are not integrated well with strategy, marketing, physician groups, and operations.
5. Historical data going back many years is not mined for important patterns that might be useful in price-setting; additionally, some historical data is faulty due to miscoding DRGs to actual procedures.
6. The amount of executive time consumed by price-setting at every negotiation interval is extensive.
7. Very large spreadsheets are common causes of errors and lack transparency and explainability.
8. Contract clauses and options are not valued using a Real Options framework.
9. Price setting processes are highly tilted toward “art” and away from science. We see a balance of the two as being optimal.
10. The emphasis is on a single reimbursement number allows little room for cooperative pricing schemes like gainsharing (sometimes referred to as value-based pricing).

So what is the alternative?

We would propose solving many of the problems cited above in the same way that many companies across different industries are solving complex problems: through models. A properly constructed model with all of the data necessary to forecast system profitability based upon negotiated prices allows users to conduct a near-infinite number of “what if” scenarios for pricing effects, even in real-time as the negotiation with payers is taking place [2]. In principle, this model works just like a spreadsheet but uses more powerful calculation capabilities, an easy User Interface (UI) and by nature is much less error-prone. This is a separate problem from the additional challenge of detecting the miscoding of DRGs, but the same model could be harnessed for this purpose as well.

We anticipate that users will “experiment” over and over with the model all the while stress testing the fee schedule against many demand scenarios, changes to fees, changes to terms and conditions, response to competitor’s pricing, etc. Moreover, we expect the model to remain of value after the negotiation to update the forecast against actual facts on the ground routinely across the contract period(s) [3].

The Problem of Optionality…and a Solution

Having built models of union negotiations, yield management pricing, property prices, and even professional athlete contracts, I’ve seen the complexity that can arise from what appears to be a rather simple transaction. The complexity often comes in the form of “terms and conditions” — all of the non-numeric clauses in the contract that we define here as options.

Let’s take two rental car contracts A and B. These contracts are identical except that A allows me to return the car at any affiliated location but B requires me to return the car to the pickup point. B is clearly more restrictive than A, and therefore less valuable, right? In my experience, many companies see A and B as the same because of the face value of the contract. That’s wrong. Enter Real Options. This is a technique that actually puts a numeric value on “optionality” in contingent contracts, allowing the real differences between two agreements to emerge.

Here are examples of provider-to-payer contract items that could be valued using Real Options:

• Definition of medical necessity
• Termination
• Network requirements
• Unilateral agreement to contract changes
• Dispute resolution
• Definition of a “clean” claim
• Allowed amounts
• Provider credentialing

Summary

Coming up with a price for a medical procedure is an intensely complex math problem. Why not avail yourself of the best Data Science that can be brought to bear on the problem? Why not institutionalize pricing in a way that allows it to get better and better each round? This is precisely what top-performing organizations do.

__________________________________________________________________________

[1] Hospital systems sued to block the law but lost in Federal court.

[2] This is a separate problem from the additional challenge of detecting the miscoding of DRGs, but the same model could be harnessed for this purpose as well.

[3] This same type of model can be used to negotiate with physician groups as well.

References:

“How Much Does a C-Section Cost? At One Hospital, Anywhere From $6,241 to $60,584”. Wall Street Journal, February 11, 2021