The influence of multiple scorers on cytogenetics study results

Cytogenetics data resulting from one laboratory in a multiple-laboratory study were analyzed to determine if 5 well-trained scorers produced significantly different results using metaphase scoring procedures. Although the scorers reached the same general conclusion, results show that scorer differences exists (p less than 0.01). Consequently, all participating scorers in a laboratory should be used equally in all treatment groups and the results should be analyzed accordingly to account for scorer variations. This is easily accomplished in controlled prospective experiments; however, it is often difficult in retrospective studies using data which exists. In such studies, every effort should be made to analyze and interpret the data so that scorer differences are taken into account. For severely damaged cells not only were there scorer differences but the difference were greater at higher doses. This phenomenon may be related to the operational definition of a severely damaged cell, since scorers who identify more damage than other scorers would logically tend to classify more cells as severely damaged both overall and at lower doses.