February, 2013

posted Feb 13, 2013, 7:05 AM by PC/SM Chief Resident   [ updated Jul 10, 2013, 6:44 PM by Angela Jeffers ]
Pay for Performance


As physicians, we recognize that quality improvement is our professional duty. Supporters of pay-for-performance argue that it incentivizes high quality care. Detractors say that it is unproven and may potentially do more harm than good.

Questions for discussion:

1. In a general sense, what is "risk adjustment" and why is it necessary when comparing hospital performance? When we talk about "risk adjustment models" what do we mean?

2. What factors are generally included in current risk adjustment models? Which are not? (see Table 4, Kansagara et al.) 

3. Think about your patients at the hospital or in clinic. Which factors do you think put them at risk for hospital (re)admission? Why might some of these not be included in existing risk adjustment models?

4. The c-statistic is a measure of how well a model performs. More specifically, it is the ability of a model to distinguish between high-risk and low-risk. For example, if we compare two patients where patient A was readmitted and patient B was not, we would hope that our model predicts a higher risk for patient A. How do the models from Kasangara et al. perform in predicting readmission? (See 'Model Discrimination' in Table 3)

5. Imagine you are an administrator at a hospital with high risk-adjusted CHF readmission rates. What could you do to lower them?

6. Bhalla and Kalkut argue that failure to include sociodemographic factors in risk adjustment penalizes providers caring for disadvantaged and vulnerable patients. Others (including the US government) argue that risk adjustment for sociodemographic factors sets a "lower bar" or "double standard" for certain communities. What do you think?

7. Beyond the limitations of risk adjustment, Woolhandler and colleagues argue that pay for performance and health care may be fundamentally incompatible. Do you agree or disagree?