Healthcare, Moore’s Law and Joints: A Statistical Appendix and Discussion of Case Costing – Part II of II
Prior period cost-based models have serious limits and are grossly overused and under-comprehended. Normative rather than empirical modelling is probably a better way to go to set prices. Or better still, leave the costs up to the provider and pay based on auctions with appropriate incentives for localness, teaching and research, and quality of care.
But paying costs is “fair”! That is it is “fair” to providers and “fair” to citizens. Because costs are fair. They are what it costs! And that is, well, fair.
Except it doesn’t work out that way, prior period cost reporting ends up being the black box model that we spoke of in the first part of this essay with flaws around outlier exclusion and allocation methods and other challenges. And, to make a long story short, we do not have the time to fix it. That would take years of diligent work (pace JPPC) to manage methods and allocations. I have talked to a number of folks who are much smarter on this than I am and have convinced myself that we simply cannot work through the details of a prior period empirical model in the time available for us to implement activity-based pricing. And if we make an activity-based payment system wait for this hard work to be done we will delay meaningful health reform for at least one four-year mandate and probably more. [Note: I am not going into more detail here but I am happy to discuss why I believe this).
Even were we successful in “torturing the data until it confesses” a price for activity-based payment, we would still be subject to all of the constraints of prior period empiricism, including:
- Delays in updating (often of two plus years)
- An inherent upward bias as cost-plus thinking infects reporting
- Active gaming by rogue participants
- Locking in of poor production models and a bias against innovation (because poor players can recover their too high costs)
There are several other options that should be considered and combined with prior period empiricism in setting prices
Normative models based on best practice pathways provide a reasoned way of examining a well-defined course of treatment and establishing the likely direct costs for the care. This can be combined with an appropriate overhead burden and thoughtful risk based payments for complications, nosocomial infections, and readmission. This should work well if analysts remember not to extend it beyond well understood care pathways. In Christensen language, this is a payment system for a value added process NOT for a solution shop. Coming back to joints, this means simple hips and simple knees without complications and no second tries. Frail elderly, diabetics, and others should be excluded a priori, and a “stop-loss” provision should be set up for cases that go beyond 5x or 10x the average cost/LOS. We want these VAP shops to be like Kensington for Cataracts. Doing the easy cases, cheaply and well and not doing the complex stuff. Remember that Christensen says that the big savings is from straightening out the process and reducing the overhead allocation. THIS IS NOT A PAYMENT METHODOLOGY FOR HIGH END ERRATIC COMPLEX PROCEDURES. Those are paid on a time and materials basis (aka Global Budgets).
Auction-based pricing is a much simpler way to get to activity-based pricing quickly.
There are several types of auctions that we could use and which might be appropriate:
- Set a price (using both normative and empirical costing) and see if the MOHLTC gets enough takers to cover the needed volumes during LHIN negotiations of annual agreements. Economically, this leaves a lot of money on the table but it does not destabilize the system
- Does a shared winner auction (like a bond auction) in which bidders share in the volumes based upon their bids but there is no winner take all. Less disruptive but also less aggressive.
- True spot market auctions in which a volume of services is put to bid and the winner takes all. Very aggressive pricing with volume swings.
All of these pricing systems can and should be combined with bonuses for teaching, perhaps a bonus for research (although one could argue this is paid elsewhere), a bonus for “localness” (that would allow Hamilton to compete in Mississauga but at a disadvantage), and a bonus for quality [my preference is 1/(SQRT of (HSMR/100))]
My former colleagues (Sanjay Cherian, Nadir Hirji and others) and I worked up a model which we called “C-Bay” which would allocate among the three systems above 75% annual agreements, 20% bond auctions, and 5% true spot markets. The clear benefit here is that the MOHLTC and LHINs would gain real pricing information from the 25% of volumes allocated through true auctions and this would supplement the black box models whether they are empirical or normative/pathway based. And if wait times moved up policy makers could add some volumes to the spot market.
The Ministry will still have plenty of modelling left to do! Anyone who thinks that pricing does not involve a lot of supplier cost modelling has never worked in a complex private sector supply chain. You can’t negotiate a good price without a good model of your suppliers cost. But you give up pricing power if you do not see multiple bids and allow the market to help you in setting prices. A true internal market system is needed as part of activity-based pricing.
A final note on the perennial red herring: privatization. Do not allow private sector players to bid in the first three years of implementation. It will just derail the sensible discussion that needs to take place. I have nothing against the private sector but including them early will make this a holy war and that is unneeded. We need to get our payment methods fixed first
About the Author(s)
Will Falk (@willfalk) is an Executive Fellow in Residence, Mowat Centre for Policy Innovation, School of Public Policy & Governance and Adjunct Professor, Rotman School of Management, University of Toronto.
For recent articles and speeches click here.
Be the first to comment on this!
Personal Subscriber? Sign In
Note: Please enter a display name. Your email address will not be publically displayed