Using Tolerance Limits to Evaluate Laboratory Data

Abstract

Quantitative laboratory measurements are a large part of the data collected in a clinical trial and the analysis of this type of data is often challenging. The use of tolerance limits and its variations, which describe the concentration and extremes of a distribution, provide useful methods to evaluate laboratory data. The theory of tolerance limits for the gaussian distribution and any unknown continuous distribution is presented. Tolerance limits provide adequate data reduction while still allowing appropriate interpretation of the results and are useful for evaluating individual patient responses. Examples of the utility of determining limits in which a chosen proportion of the data is likely to be concentrated for interpretation purposes are presented. Using the concept of tolerance limits to determine the size of a clinical trial designed from the safety perspective is also discussed.