MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Statistical Failings: The Shadow Still Hiding Clinical Truth

D Schlesinger1*, J Chang2*, W Sensakovic3*, M Altman4*, (1) University of Virginia Health Systems, Charlottesville, VA, (2) Northwell Health, Lake Success, NY, (3) Florida Hospital, Orlando, FL, (4) Washington University in St. Louis, St. Louis, MO





Presentations

(Tuesday, 7/31/2018) 7:30 AM - 9:30 AM

Room: Karl Dean Ballroom A2

Many aspects of the medical physics profession are unavoidably dependent on statistical analyses. Common clinical physics tasks as diverse as evaluating process risk, determining treatment uncertainties, and interpreting quantitative imaging studies involve statistical analysis. Increasingly there is a de-facto requirement that to be published, studies must base their conclusions on evidence provided by formal statistical tests. However, recent evidence demonstrates that a large proportion of clinical and technical studies cannot be independently replicated. Medical physicists therefore have a critical need to understand and be able to interpret statistical methods and results. However many medical physicists have minimal or no training in the sort of practical statistical methods commonly found in clinical studies or even in day-to-day tasks in the clinic. They may therefore have an inadequate ability to detect statistical errors and limitations that are an unfortunately frequent.

In this session we will demonstrate common statistical errors frequently encountered in peer-reviewed literature as well as in common tasks performed by clinical medical physicists, distinctive symptoms that can be identified to detect these errors, and explanations for how they might have been corrected. In the process, we will explain some of the basic concepts of inferential statistics. Some specific case studies we will cover will include misunderstanding the meaning of p-values, misinterpreting statistical power, conflating statistical vs clinical significance, and understanding the use and limits of common statistical tests.

Learning Objectives:
1. Learn about the frequency of statistical problems in published studies
2. Identify common signs and symptoms of potential problems in various types of statistical tests
3. Learn methods for correctly implementing statistical analyses of the type commonly found in medical physics publications and in routine clinical physics activities.

Handouts

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email