Background: Precision and accuracy assurance in cardiopulmonary exercise testing (CPET) facilitates multicenter clinical trials by maximizing statistical power and minimizing participant risk. Current guidelines recommend quality control that is largely based on precision at individual testing centers (minimizing test-retest variability). The aim of this study was to establish a multicenter biological quality control (BioQC) method that considers both precision and accuracy in CPET.
Methods: BioQC testing was 6-min treadmill walking at 20 W and 70 W (below the lactate threshold) with healthy non-smoking laboratory staff (15 centers; ~16 months). Measurements were made twice within the initial 4 weeks and quarterly thereafter. Quality control was based on: 1) within-center precision (coefficient of variation [CV] for oxygen uptake [V̇O2], carbon dioxide output [V̇CO2], and minute ventilation [V̇E] within ±10%); and 2) a criterion that V̇O2 at 20 W and 70 W, and ∆V̇O2/∆WR were each within ±10 % predicted. "Failed" BioQC tests (i.e., those outside the predetermined criterion) prompted troubleshooting and repeated measurements. An additional retrospective analysis, using a composite z-score combining both BioQC precision and accuracy of V̇O2 at 70 W and ∆V̇O2/∆WR, was compared with the other methods.
Results: Of 129 tests (5 to 8 per center), 98 (76%) were accepted by within-center precision alone. Within-center CV was <9%, but between-center CV remained high (9.6 to 12.5%). Only 43 (33%) tests had all V̇O2 measurements within the ±10% predicted criterion. However, a composite z-score of 0.67 identified 67 (52%) non-normal outlying tests, exclusion of which coincided with the minimum CV for CPET variables.
Conclusions: Study-wide BioQC using a composite z-score can increase study-wide precision and accuracy, and optimize the design and conduct of multicenter clinical trials involving CPET.
Trial registration: ClinicalTrials.gov identifier: NCT01072396; February 19, 2010.