Standardizing lung function laboratories for multicenter trials.

Multi-center studies provide advantages in clinical research but differences between centers can introduce bias. Three specialist pediatric respiratory laboratories standardized their methodology and examined differences between centers. The specific aims were to (i) assess the variability of measurements on adults within and between centers and (ii) to exchange and cross-analyze data from children to assess the extent of agreement between centers. Each laboratory used identical equipment and software. Inter-laboratory visits were used to (i) standardize protocols for data collection and analysis and (ii) make spirometric and plethysmographic measurements on participating staff at each location. Staff also had repeat measurements in their home laboratories. Measurements from children in each laboratory were exchanged on disk, cross-analyzed, and data compared by ANOVA. There were no significant within-subject, between-center differences in FVC, FEV1, FEF50, FRCpleth, or VC. There was a slight trend for TLC and RV (P=0.07) to be higher at one center. The 95% limits of agreement within and between centers were similar for all parameters. There were no differences between centers in cross-analyzed data from 10 children. By standardizing hardware, software, and protocol, potential inter-laboratory differences can be minimized. We recommend that this approach be adopted prior to multi-center studies. (c) 2006 Wiley-Liss, Inc.