CBO Report on Value Based Payment Demonstration Projects

Today’s Managing Health Care Costs Indicator is 10%

I blogged on Tuesday about the Congressional Budget Office report on disease management and care coordination. Today, I’d like to talk about the CBO reporton four demonstration projects on value based payment.

The headline is that these demonstration projects were not very successful.  That’s no surprise – the CMS payment demonstration projects violated basic fundamentals of effective extrinsic incentives.

-        The incentive system should be transparent and easy to understand
-        The goals would be clear and achievable
-        The incentive should be available soon after the desired behavior
-        The target of the incentive should clearly be able to influence the outcome
-        The incentive should be presented independently from other payments

CMS wasn’t able to build incentives that fulfilled any of these criteria.  The demonstration projects were long, there was little feedback along the way.  None of the surgeons or hospital administrators felt abiding confidence that they could influence the outcomes. Payments were made years after savings were realized. These programs were inadvertently designed to fail.

The real surprise is that not all of them failed!

The big news is that one of these projects actually saved money!  The Medicare Heart Bypass Bundled Payment project saved 10% of the cost of bypass surgery without any sacrifice in quality.  (David Cutler’s 2010 review says 15%).  Two of the other demonstration projects showed small improvements in quality-based process measures, and one of the projects showed no significant change in either cost or quality.

Here is a description of this project from Health Affairs in 2008:

…under Medicare’s Participating Heart Bypass Center Demonstration, four hospitals in the 1990s were paid a single amount covering both hospital and physician services for CABG surgery. An evaluation showed that Medicare paid 10–37 percent less, physicians identified ways to reduce length-of-stay and unnecessary hospital costs, and patients preferred the single copayment, with no cost shifting to outpatient care.

Gail Wilensky, a former CMS Administrator, asserts that further projects of bundling payment were stymied by regulatory findings that prohibited hospitals from gain-sharing with their physicians.  That’s possible. Clearly, an incentive for the hospital that cannot be transmitted to the cardiac surgeon making decisions isn’t very promising.  Also, the demonstration project was small – and it’s possible that it wouldn’t scale.

Still, it’s a surprise that CMS has not tried to replicate this!  I suspect that hospitals weren’t especially enthusiastic for expansions of this demonstration project. Through the late 2000s cardiac surgery was a reliable profit center, and lowering revenue from this service line looked very unattractive to hospitals.  .

The CBO conclusion is that we need to move away from fee for service. The writer concludes that it’s hard to have an impact with payment reforms that leave the underlying fee for service system untouched. I agree that fee for service is highly inflationary, and bundled or capitated payment systems can help bring us more value.  See a series of posts from 2009 on this topic: Part One Part Two, Part Three

But transitioning from the fee for service will require many changes in the provider system, and is unlikely to be successful in rural areas and medical communities with little competition.  Furthermore, fee for service is likely the best way to pay for some rare or unusual conditions.   Therefore, we need to develop payment reform that is compatible with the existing fee for service system. Here’s a link to a Catalyst for Payment Reform issue brief on this topic.  

Medicare showed us through this early 1990s demonstration project how to effectively implement bundled payment for selected services in the context of overall fee for service payment.  It’s time to put that knowledge to use.
Click image to enlarge.  Source