Within Subject Nested Clinical Trial Laboratory Data:Sample Size, Components of Variance and Cost

• Activity Number: 357
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Consulting
• Author(s): Borko Jovanovic*+ and Hariharan Subramanian and Irene Helenowski and Alfred Rademaker and Angela Fought and Hemant Roy and Vadim Backman
• Companies: Northwestern Univ-Feinberg School of Medicine and Northwestern University and Northwestern University and Northwestern University and Northwestern University and Northwestern University and Northwestern University
• Keywords: sample size ; nesting ; spectoscropic microscopy ; lung cancer ; cells ; pixels

Abstract

Nesting of experimental factors is well established in statistical literature relating to agricultural and engineering experiments. It seems less often discussed in biological and laboratory experiments stemming from clinical trials and the use of using human bio-specimens. In a typical such study several levels of measurement are nested in the sampling process: subject, tissue(subject), cell(tissue(subject)),sub-cell-elements(cell(tissue(subject))). Sample size considerations are most often provided in very general hypotheses testing scenarios on the subject level and do not provide advice regarding number of sampling units at lower levels. In this paper we revisit the experimental nesting framework, and discuss how variability, effect size and cost at various levels may be thought of and coherently utilized.