Objectives: To assess reproducibility of core laboratory performance and impact on sample size calculations.
Background: Little information exists about overall reproducibility of core laboratories in contradistinction to performance of individual technicians. Also, qualitative parameters are being adjudicated increasingly as either primary or secondary end-points. The comparative impact of using diverse indexes on sample sizes has not been previously reported.
Methods: We compared initial and repeat assessments of five quantitative parameters [e.g., minimum lumen diameter (MLD), ejection fraction (EF), etc.] and six qualitative parameters [e.g., TIMI myocardial perfusion grade (TMPG) or thrombus grade (TTG), etc.], as performed by differing technicians and separated by a year or more. Sample sizes were calculated from these results. TMPG and TTG were also adjudicated by a second core laboratory.
Results: MLD and EF were the most reproducible, yielding the smallest sample size calculations, whereas percent diameter stenosis and centerline wall motion require substantially larger trials. Of the qualitative parameters, all except TIMI flow grade gave reproducibility characteristics yielding sample sizes of many 100's of patients. Reproducibility of TMPG and TTG was only moderately good both within and between core laboratories, underscoring an intrinsic difficulty in assessing these.
Conclusions: Core laboratories can be shown to provide reproducibility performance that is comparable to performance commonly ascribed to individual technicians. The differences in reproducibility yield huge differences in sample size when comparing quantitative and qualitative parameters. TMPG and TTG are intrinsically difficult to assess and conclusions based on these parameters should arise only from very large trials.