Robin Willixhofera, Marlus Karstena,b, Arianna Galottaa, Elisabetta Salvionia, Carlo Vignatia,
Alice Bonomia, Anna Apostoloa, Mauro Continia, Pietro Palermoa, Jeness Campodonicoa,
Beatrice Pezzutoa, Stefania Farinaa, Massimo Mapellia, Irene Mattavellia, Piergiusepe Agostonia,c,*

a) Centro Cardiologico Monzino IRCCS, Milan,
b) Cardiovascular Health and Exercise Research Group (GEPCardio), Department of
Physiotherapy, Graduate Program in Physiotherapy, Santa Catarina State University (UDESC),
Florianopolis, Santa Catarina, Brazil
c) Department of Clinical Sciences and Community Health, Cardiovascular Section, University of
Milan, 20122 Milan, Italy.

Paper not yet published but accepted for publication

Background

Cardiopulmonary exercise testing (CPET) is the gold standard for assessing exertional dyspnea,
providing peak oxygen uptake (VO2), and the VE/VCO2 slope. Different metabolic carts may
introduce systematic variability, confounding longitudinal assessments and multicenter trials.
We evaluated intra-device reliability and inter-device comparability of CPET parameters.

Methods
In this prospective, single-center, within-subject, repeated-measures, multi-device comparison
study 34 healthy adults (35±11 years; 59% males) completed eight CPETs: two per cart, on four
systems (MGC Ultima™ CPX, COSMED Quark CPET™, Schiller PowerCube® Ergo, Vyaire
Vyntus™ CPX). Tests used randomized order, ramp protocols on cycle ergometer, and
standardized calibration, masks, and timing (2–10 days apart; within 40 days). Breath by breath
data were 10-second averaged. Analyses included paired tests for intra-device reliability, repeated-
measures ANOVA for inter-device comparisons, and iso-workload evaluation at 25–100% of
lowest peak workload (ISO-Wpeak).

Results
Intra-device reliability was high, with no significant differences in VO2, heart rate or workload at
anaerobic threshold (AT), respiratory compensation point (RCP) or peak. Minor variances
occurred in resting VO2 (COSMED: 56ml/min, p=0.0427) and peak ventilation (Medical Graphic
9 l/min, p<0.05). Inter-device comparability showed no differences in peak VO2 and the VE/VCO2
slope calculated up to the RCP. The VE/VCO2 slope assessed as the full slope however, showed
significant differences between machines (p=0.0048). Across standardized ISO-Wpeak (25%,
50%, 75%, and 100% of ISO-Wpeak), VO2 values were largely comparable between devices, with
significant differences only for VO2 (ml/kg/min) at 25% ISO-Wpeak (p=0.022).

Conclusion
Under standardized conditions, the tested metabolic carts yield comparable values both for intra-
device reliability and inter-device comparability.