Present: Levi Bonnell, Justine Dee, Emily Houston, Ben Littenberg, Adam Sprouse-Blum, Connie van Eeghen (6)
Guest: Brendan Thoms (Rheum MD, PhD in treatment of osteoarthritis, Research interests)
1. Warm Up: Introductions
2. Brendan and Levi Paper: Predictors of Inflammatory Arthritis Among New Patient Referrals – the audience is Rheumatologists, although it involves PCPs as well
a. Title: is it specific enough? Patient population? Type of study?
b. Abstract
i. Unstated problem: rheumatologists are overwhelmed by volume of patients and trying to get to the sicker ones quicker. Also, some of that volume is inappropriate and may be better managed at the PCP level.
ii. Objective: could be clearer
iii. What is a referral letter? At UVM MC: PCP writes an order. What is the best way to describe the data source?
iv. Not accurate: distinguish inflammatory vs non-inflammatory arthritis? This is already known. Clarify the issue around predictive variables.
v. Not clear that the variables used were available to the PCP. Is the focus on the pre-specialty visit? Then organize the message around their work process.
c. Introduction
i. What’s the hook and how does it logically flow?
ii. The message is about timely dx and treatment of an inflammatory disease; this gets muddy
d. Methods
i. Referral letter not described
ii. How does the list of exclusions relate to the message of the paper: things that the PCP can do?
iii. Prediction models: Ben now prefers a CART model to handle a variety of variables, including “missingness”, better than a multi-regression logistics model
e. Results
i. There is an unspoken issue: there are no standard expectations about when to refer to Rheum and what tests to do ahead of time
ii. Focus on the message and the audience. Is the goal to make a PCP tool or a Rheum tool?
f. Next steps: testing the goodness of the variables that may predict inflammatory disease.
3. Next week: TBD
Recorded by: CvE
Posts
