In-Field Validity and Inter-Unit Variability of Metabolic Carts During Simulated Exercise

Bas Van Hooren;  Institute of Nutrition and Translational Research in Metabolism (NUTRIM),Maastricht, the Netherland
Tjeu Souren; Bart C. Bongers

Scandinavian Journal of Medicine & Science in Sports, 2026; 36:e70297
Headings by Dr Older

Aims The present study had three main objectives: (a) to evaluate the in-field validity of different commercially available cardiopul-
monary exercise testing (CPET) systems when used by end-users following typical calibration procedures, (b) to measure the
variability in accuracy among identical CPET units, and (c) to explore the relationship between the age of the units, as well as
the maintenance practices, and their measurement accuracy.
Methods Fifty-seven CPET systems, calibrated and operated by end-users inclinical practice, research, or sports settings,
were assessed against a metabolic simulator that simulates breath-by-breath gas ex-
change. The values measured by each system [minute ventilation (V̇E), oxygen uptake (V̇O2), carbon dioxide production (V̇CO2),
and respiratory exchange ratio (RER)] were compared to the simulated values to evaluate the accuracy. Absolute percentage
errors during the simulations ranged from 1.41% to 24.6% for V̇E, 3.29%–10.6% for V̇O2, 2.86%–13.3% for V̇CO2, and 1.90%–10.0%
for RER. Inter-unit variability (%) ranged from 1.98% to 12.7% for V̇O2, 1.49%–8.10% for V̇CO2, and 1.93%–4.24% for RER.
Results No consistent relationship between system age and accuracy was observed, nor between annual maintenance and accuracy. The validity
of metabolic carts for measuring respiratory gas variables varied significantly even between identical systems, despite passing
manufacturers’ calibration checks. Furthermore, inter-unit variability of most systems exceeded intra-unit test-retest variability,
thus necessitating caution when using devices interchangeably, as this may increase measurement noise, even within the same
laboratory. Most inaccuracies seemed related to technological errors, although some user errors were also identified, indicating
the need for a holistic approach to identify errors.