Tag Archives: Risk adjustment

PMean: How do you select variables for a risk adjusted model?

I was helping a colleague write a response to a reviewer who asked about a risk adjusted model. How did you select the variables for adjustment? He/she speculated that we had used some type of stepwise selection. I used to do this, but stopped doing it in favor of adjusting for any or all variables that were known or suspected to be important. There are serious problems with screening using stepwise approaches to select variables for risk adjustment. But the literature is quite complex and there is no apparent consensus on what is best. Here are some quotes from a few publications about this issue. Continue reading

PMean: Forget confounding, and think of things in terms of covariate imbalance

Someone noted in a passing comment in their email that they found the term “confounding” to be difficult and confusing. I’ve been doing this stuff for over thirty years, but to be quite honest, I get a little nervous about this as well. But I took the time to explain a simpler concept, “covariate imbalance.” Continue reading

Recommended: Comparison of Logistic Regression versus Propensity Score When the Number of Events Is Low and There Are Multiple Confounders

Propensity scores represent an intriguing alternative method to reduce the impact of confounding variables, especially when there are multiple potential confounding variables. This paper considers a range of models comparing the propensity score approach to the more traditional approaches of adjusting for confounders. I think the conclusions are overly simplistic, but the paper is still worth reading. Continue reading