Wednesday, December 18, 2013
Present: Marianne Burke, Kat Cheung, Abby Crocker, Kairn Kelley, Amanda Kennedy, Rodger Kessler, Ben Littenberg, Connie van Eeghen
Start Up: The value of a “D” degree (PharmD, DPT, DrPH, PhD), whether in 3 years or 6 after baccalaureate; mostly positive experiences, but it depends.
1. Discussion: CROW’s schedule for Spring Semester is set for every Thursday. We’ll gather at 11:30, topic discussion from 11:45 – 12:45.
2. Discussion: Development of an analytic plan for medical student evaluation data
a. Connie is working with Alan Rubin and Cate Nicholas on an article about introducing an EHR curriculum in a pre-clinical doctoring skills course. Medical students are evaluated by Standardized Patients (SPs) during Clinical Skills Exams (CSEs) on a variety of skills. Among these, six questions evaluate their PRISM skills and 1 evaluates their patient-centered skills while using PRISM. Note that this is not a research area that falls inside Connie’s FINER goals, but it provides great opportunities for networking, skill building, and development of future opportunities.
b. The group discussion identified many key questions/issues for Connie to clarify. These included:
i. Are the co-authors willing to publish, regardless of results?
ii. Have they submitted an IRB protocol yet? Can Connie be included as a "key personnel?" Can the rest of CROW be included, to participate in data analysis?
iii. Understand the 7 questions (6 PRISM and 1 patient-centered) on which the students are evaluated. Do the SPs first complete a check list, which they then use to score the questions? Or, at the end of the CSE, do they just score the 7 items from memory? What is the process used to create the data? How are scores of "yes," "unsatisfactory," and "no" determined? Will some of the data be missing?
iv. It's customary to describe the population being studied in a general way. Are demographic data about the students available (age at time of test (or year of birth) and gender)?
v. It's possible that these 7 questions are related to the score received for each CSE as a whole. In other words, if a student is having a bad day, test-wise, the score for the entire CSE will reflect this. Consider adding to the final score for each CSE to the data set.
vi. Make sure the medical student identification is coded, to prevent identification. Consider whether demographic data are, by themselves, identifiers.
vii. Find out if SPs score for "patient-centered" characteristics on any CSEs last year when PRISM was not being used. This might be a way to see how they scored on patient-centeredness when NOT distracted by PRISM.
c. Analytical approach
i. Descriptive: look at (graph) the medians by time period
ii. Look at a segmented bar graph, in which the segments are the three score categories
iii. Put ALL the dots on the graph; do a low S curve (non-parametric)
iv. Identify how many students passed each question for each test (pareto diagram)
v. Consider looking at within-subject variation (Kairn willing to help with this)
d. Thank you, everyone!
a. December 20: POTLUCK! Along with a presentation by Ben on Depression and social networks on the web, with Chris Danforth and Peter Dobbs.
Posted by Connie at 12/18/2013 08:04:00 AM