Risk adjustment in medicine and healthcare is a story of convergent evolution. On one hand, risk adjustment has a history that emerged with the creation of health insurance in the United States in the 20th century. The “risk” is the risk assumed by the insurance company to insure each individual person. In this setting, it represented an estimate of how chronically ill a particular person is, and therefore, how costly that person is to insure.1 On the other hand, similar efforts were underway at the hospital level to create methods to compare one hospital with another; risk adjustment allows comparisons by understanding the variation in the probability of adverse or favorable outcomes based on patient characteristics versus hospital-driven outcomes.
Why do we need to understand risk adjustment as healthcare delivery scientists? We will spend this chapter discussing the origins of risk adjustment and several current examples in healthcare financing. Then, armed with the tools discussed in Chapters 16 and 17, we will discuss methods of evaluating risk adjustment strategies. We will return to an example of using risk adjustment work in local healthcare delivery science at a single institution and how you can employ risk adjustment techniques. Finally, we will discuss what can go wrong with risk adjustment.
WHAT IS RISK ADJUSTMENT, AND WHY WE SHOULD CARE?
Risk Adjustment Designed by Insurers
Risk adjustment in healthcare financing began as a method for health insurance companies to estimate the cost of insuring a person. A 25-year old man who does not smoke and exercises regularly is unlikely to use much healthcare. But a 64-year old man who formerly smoked, is overweight, and already suffers from hypertension, chronic obstructive pulmonary disease (COPD) on home oxygen therapy, and type 2 diabetes is likely to require multiple healthcare interactions in the coming year. If given a choice, insurance companies would only enroll the first man, charge him health premiums, and never have to pay any of those costs. But if it has to insure both, the insurance company wants to identify which patient is which and then charge more to provide insurance to the second man because he is likely to cost the company more. Joe Newhouse described the insurance company in a similar scenario as walking through the produce aisle of the grocery store, squeezing each apple, and trying to pick only the best ones.1 Apples have a single cost whether you grab a bad or a good apple, so you will want to pay for the good apple only if you can figure out which one that is. Insurers want to try to figure out which patients are the good apples and only enroll them in their plans, if they can. If they can’t choose the healthy from the sick up front, they will want to try to hedge their bets by getting the ...