Survey of librarians in the Midcontinental Chapter of the Medical Library Association (created 2010-02-10).

I have been working with Susan Sanders to develop a series of short course about Statistics that is targeted at Medical Librarians. As part of the preparation for this course, we conducted a survey. Here is the summary of the results, written up by Susan Sanders with some minor editing changes by me to fit into a webpage format. As we develop additional materials for this class, I will add them to this website.

Twenty one medical librarians responded to a survey that came out on the listserv of the Midcontinental Chapter of the Medical Library Association in November 2009. Respondents said they would be interested in a series of online webinars to increase their statistical literacy and understanding of research studies. Towards this goal, Steve Simon, Ph.D., Independent Statistical Consultant and Faculty at UMKC Dept. of Informatic Medicine and Personalized Health, is working on offering a series of webinars through the Midcontinental Chapter of the Medical Library Association. The series will be about the perils and pitfalls of research design, and how to interpret study conclusions. The course will emphasize what the studies mean to our doctors, and to medical librarians in their roles as health care educators and consumers.

Watch the MCMLA-listerv for details, and check for updates on Steve Simon's http://www.pmean.com/index.html. Steve's UMKC website is at this URL: http://www.med.umkc.edu/informatic_medicine/Faculty/simon.shtml

A summary of the survey results according to the highest percentage of content the librarians wanted to learn about is as follows:

Research Designs:

• Intention to treat analysis - 90%
• Exclusion bias - 86%
• Randomized controlled trials and selection bias - 81%
• Cohort studies, case control studies, and cross-sectional studies - 76%

Understanding the Numbers:

• Type l & type II errors - 86%
• Number needed to treat - 86%
• Hypothesis testing - 86%
• Confident intervals - 86%
• Prevalence & incidence measures- 86%
• P-value - 81%
• Odds ratio & relative risk - 81%
• Diagnostic testing /ROC curves - 81%
• Correlation - 81%

Interpretation of the Results:

• Reliability - 90%
• Clinical importance - 86%
• Validity - 86%
• Strength of association - 81%
• Retrospective measures - 76%
• Sub-group analysis - 71%
• Consistency - 71%
• Fitting the Pieces Together:
• Publication bias in meta-analysis - 90%
• Heterogeneity in meta-analysis - 86%
• Guidelines - 86%