W-04: Treatment Comparisons of Timed Tests in Neurology
David R. Bristol
Statistical Consulting Services, Inc. United States
The time to completion of several tasks is censored when some patients do not complete all tasks, but references do not provide information for between-treatment comparisons in this case.
Simulations are performed to compare various procedures for between-treatment comparisons of timed functional tests when some patients do not complete all tasks.
Some standard timed functional tests are discussed. When some patients do not complete all tasks, the times to completion are censored and standard time to event analyses can be performed, ignoring the number of tasks completed. The number of tasks can be analyzed, ignoring the times to completion. A logical ordering of the possible outcomes is equivalent to analysis of the velocity=number of tasks/time.
The analysis of the velocity=number of tasks/time is proposed as an appropriate analysis of timed functional tests.
Co-author: Nicole O. Stouffer