It’s All About Risk in Bundling Initiative

(Post Five in a series)

By Joanne Conroy, MD

One of the concerns that keep our CFOs up at night is how to minimize the financial risk of participation in our CMMI bundling initiative. There is definitely a downside risk in this project — hospitals might have to write CMS a check at the end of the year! It’s important for institutions to have “skin in the game,” so there will always be some risk. However, every CFO remembers that $2M patient who either drained the blood bank, depleted the pharmacy, or topedoed the hospital’s bottom line for the quarter. In addition to the risk associated with one catastrophic patient, there is aggregate risk to be considered: that associated with caring for a number of patients whose aggregate expenses exceed the risk tolerance of the organization.

What creates greater risk in bundling? Chris Tompkins from Brandeis University is the group’s expert on this.

  1. If you select DRGs that are low volume, you can have unstable statistics, noise, unsystematic or random risk, stochastic variation.
  2. Some DRGs, such as major chest DRGs, are very heterogeneous: The diagnoses are a mixed bag of congenital cardiac disorders and very complex cases. There is too much uncontrollable variation within this DRG.
  3. Your case mix in the bundle can change with the complexity of the patients. For example, the percentage of cases that are transfers for CABG increases over a period of time. These patients are sicker and cost more to care for than the average patient calculated in your baseline year.
  4. Your admission criteria for that patient cohort changes dramatically. Simple patients are moved to observation status and never admitted, dramatically changing your average cost of care for that DRG.
  5. You will have random catastrophic cases — outliers.

How do you mitigate risk?

  1. Make sure everyone on your team has skin in the game.
  2. Practice smart bundle design: Select high-volume DRGs as well as a large number of related DRGs to spread out the impact of stochastic variation while keeping a narrow clinical focus.
  3. Think carefully about inclusions and exclusions. Include those post-acute services and related readmissions that occur with a high prevalence in the population. Remember, you are treating patients with all their common co-morbidities.
  4. Think carefully about what factors are exogenous — outside your organization (social determinants of health, inevitable progression of disease, and so forth) — versus endogenous (things you can do something about).
  5. Reinsurance  or stop loss insurance.

Is there a role for stop loss insurance?

The catastrophic patient admission is usually a random event; you probably don’t even get them every year. Thus, they could not have been included in your historical baseline calculation for that episode. Stop loss makes sense above a certain claims trigger.

However, aggregate loss should also be a concern. You could expect that if you have a complex patient population (as all AMCs have), they would be included in your historical baseline calculation. On average, about 6 percent of teaching hospital admissions are outliers (twice the average seen at non-teaching hospitals); these outliers account for 12 percent of Medicare payments to the teaching hospital. In other words, we have more of them and they are far more complex.

But if the mix of patients changes — you start to accept more transfers or the market changes, resulting in consolidation of complex patients at academic medical centers — your baseline no longer reflects acuity and cost. So, aggregate stop loss makes sense. These patients don’t fall into the “catastrophic” category — and how many patients can you afford whose costs are $50,000 or $75,000 over the bundled price?

We have had several interviews with insurance brokers to determine the availability and cost of individual and aggregate stop loss insurance for participants in bundling. Steep learning curve!

Either catastrophic or aggregate losses could obliterate any gains you make with careful population management and care redesign — that is not really the purpose of this project.

—Joanne Conroy, MD, is Chief Health Care Officer at the Association of American Medical Colleges. She can be reached at Follow her on Twitter @joanneconroymd.