Dealing with and Defining Abnormalities in Laboratory Data

The measurement and assessment of laboratory data pose some special difficulties for those involved in clinical trials, particularly with respect to the definition and use of normal ranges. This paper describes a number of these problems. One important problem relates to multicenter studies in which each center sends its samples to a different laboratory for analysis; another concerns changes in normal ranges within a single laboratory during a study; yet another concerns the use of different units of measurement at different laboratories.

A transformation used by Hoffman-La Roche is proposed as one possible solution to the problems that occur in connection with normal ranges. The advantages and disadvantages of this method are discussed with examples, and it is concluded that the transformation provides a useful option if used carefully.