Activity Number:
|
557
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 3, 2016 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistical Consulting
|
Abstract #320819
|
|
Title:
|
Correlations for Repeated Measures Can Be Calculated, but May Not Add Value
|
Author(s):
|
Ke Yan* and Dudley Benson and John Scott and Robert Niebler and Debra Newman and Pippa Simpson
|
Companies:
|
Medical College of Wisconsin and Children's Hospital of Wisconsin and Medical College of Wisconsin and Medical College of Wisconsin and Blood Center of Wisconsin and Medical College of Wisconsin
|
Keywords:
|
Correlation ;
Repeated measure
|
Abstract:
|
Introduction: 25 patients' blood was sampled at four defined time points and various measures of the blood were made. Of interest was whether a ROTEM measure could be used to predict excessive bleeding. ROTEM is an instrument which yield multiple summary measures for blood function. Methods: ROTEM and laboratory values such as platelet count and fibrinogen were collected. Plots were generated and correlations at each time point were examined. Mixed models were used to obtain overall correlations and predictive ability. Results: Although the overall correlation appeared significant and of reasonable magnitude, it could be misleading possibly due to a "floor" and "ceiling" effect of the ROTEM measures. For prediction purposes, it was debatable whether this mattered. Conclusion: Care should be taken when examining repeated measures correlations.
|
Authors who are presenting talks have a * after their name.