Tuesday, March 16, 2010

Ben needs input...

On April 2, I have agreed to present some new material for CROW (Fellows Workshop). After we do the last discussion of Gawande's Checklist Manifesto, I can either:

Demonstrate the use of NHANES data concentrating on the newly-released long-term mortality follow-up of ~20,000 respondants over 15 years. This would involve investigating a hypothesis posed by Amanda that aspirin and ibuprofen may interact negatively to reduce the apparent mortality effects of either one alone. The new additions to NHANES allow us to do some pretty powerful pharmacoepidemiology right from the comfort of our laptops.

Talk about how to make predictions using data. What is the role of prediction in clinical care? In testing hypotheses? What are the methods available? How should we judge predictions? This talk would be more theoretical without a data set to demonstrate on. 

Which talk would you prefer? Why? (Please use the comments to answer. Thanks.)

1 comment:

  1. Ben, I would much prefer to dive into using the NHANES data w/mortality and Amanda's hypothesis as an example, esp. w/the CROW crew in attendance to add insight and value. That said, the second option is something I'd like to participate in as a later CROW discussion/review topic.


Note: Only a member of this blog may post a comment.