Sample Size for Repeated Measures Designs of Method Comparisons

• Activity Number: 35
• Type: Topic Contributed
• Date/Time: Sunday, August 11, 2002 : 4:00 PM to 5:50 PM
• Sponsor: Biopharmaceutical Section*
• Author(s): Gene Pennello*+
• Affiliation(s): Food and Drug Administration
• Keywords: method comparison ; repeated measures ; sample size ; root mean square ; linear model ; gold standard

Abstract

Devices that measure health quantities of interest include blood glucose monitors, pulse oximeters, and bone densitometers. When comparing the measurements of one device with another, a common agreement index is the root mean square (RMS) of the difference between paired measurements from the two devices. Often, the sampling design is to make repeated paired measurements on a number of subjects. In this talk, I determine the subject sample size for a given number of repeated measures per subject. Because of between-subject variability, the subject sample size does not decrease to zero as the number of repeated measures per patient increases to infinity. Sample size is based on an approximate normal distribution for the square of RMS. The square of RMS is approximately normal if the differences are assumed to be normal, because it can then be written as a sum of independent and identically distributed random variables. The mean and variance of the normal distribution on the square of RMS are easily obtained from formulas for quadratic forms.