Power Analysis for Reproducibility Precision Studies of Multilevel Designs

• Activity Number: 657
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Beimar Iriarte*+
• Keywords: Reproducibility, Variance Components, Multilevel, Precision, Power

Abstract

Precision is an important analytical performance characteristic of a measurement procedure. Good reproducibility of results in a clinical laboratory is of vital importance for the proper management of the condition of the patient. Reproducibility studies under expected operating conditions or factors are required by the reviewing, regulatory and approving bodies of the measurement device. These reproducibility studies are inherently multilevel or hierarchical in design, and variance components associated with such factors are the natural outcome of these studies, as is the existence of intra-class correlations greater than zero. Power analysis on sample size that takes into account the various levels of the designs should be an important step in the design of such studies, both for the analytical and financial cost targets. Methodologies and approaches for sample size determination for the various levels in the design will be illustrated with the 2 and 3 level designs of common use in reproducibility studies in a clinical laboratory.