Using ROC Curves to Evaluate Diagnostic Tests: An Exercise in Teaching Through Replicating Research

• Activity Number: 418
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Teaching of Statistics in the Health Sciences
• Author(s): Brianna Hitt*+
• Keywords: Research Replication ; Teaching ; Receiver Operating Characteristic Curves

Abstract

The replication of results has long been an important part of research. It is now considered an effective method for teaching statistical methods to undergraduate students. At the Summer Institute of Biostatistics at the University of South Florida, my fellow undergraduate students and I used the replication of results to learn methods of analysis in biostatistics. Specifically, I learned the steps of a complete analysis using receiver operating characteristic (ROC) curves to evaluate the Clinical Dementia Rating Scale and a modified scale. ROC curves were constructed in SAS® for six different domains of the tests and for the overall tests. The original results were successfully replicated by comparing the tests using ROC curves and their respective areas under the curve (AUCs). My partner and I went beyond the scope of the original research by learning to write SAS® macros to evaluate the tests. Additionally, the ROC curves were indexed using the minimal distance and Youden indices to predict mild cognitive impairment, which has been shown to predict dementia. We concluded our learning by presenting our replicated results and our additional work to our peers and mentors.