Xu F; Department of Functional Testing Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, Beijing, China.
Sun XG; Liu F; Zhai WX; Song Y; Tai WQ; Wang JN; Zhang YF;Zhou QQ; Shi C; Xie B; Chen JH; Huang J; Zhang ZF; Xiang MJ; Ye SD; Li L;

Journal of thoracic disease [J Thorac Dis] 2025 Jun 30; Vol. 17 (6), pp. 3863-3872.
Date of Electronic Publication: 2025 Jun 26.

Background: Gas exchange measurements, such as oxygen uptake ( INLINEMATH ) and carbon dioxide output ( INLINEMATH ), of cardiopulmonary exercise testing (CPET), are the key and gold standard for human cardiopulmonary functional evaluation. However, in terms of quality control, they are unstable and inaccurate. We used a metabolic simulator (MS) to detect measurement errors and enhance quality control.
Methods: In the Fuwai CPET laboratory, we performed CPET after systems had: (I) passed all the steps of regular system calibrations for flow and the partial pressure of O ₂ and CO ₂ ; and (II) passed the MS validation of INLINEMATH and INLINEMATH at low, medium, and high metabolic rates (MRs) daily from 2014 to 2023 for eight different CPET carts/systems. The absolute percentage difference of the 1 ^st validation of both INLINEMATH and INLINEMATH was calculated as follows: |[(measured – ideal) / ideal] × 100%|. A difference of <10% was set as the 1 ^st validation pass standard to run the laboratory, while a difference of ≥10% was classified as a 1 ^st validation failure. The absolute percentage difference of the 1 ^st validation among the eight carts/systems was compared using the Kruskal-Wallis H test. The rate of the 1 ^st validation failure, the number of validation days, and the median absolute percentage difference of the 1 ^st validation among the different CPET carts/systems were clustered using the hierarchical clustering method.
Results: In total, we completed 1,810 validation days for the eight CPET carts/systems, and found a 10,860 absolute percentage difference of the 1 ^st validation of INLINEMATH and INLINEMATH . The number of validation days completed by each cart/system and the 1 ^st validation failure rates were as follows: 8 (87.50%), 10 (90.00%), 54 (48.15%), 349 (43.27%), 20 (45.00%), 759 (21.21%), 525 (29.52%), and 85 (22.35%), respectively. The overall absolute percentage difference of the 1 ^st validation of each cart/system was 7.32% (P ₂₅ , P ₇₅ : 3.67%, 13.82%), 9.12% (P ₂₅ , P ₇₅ : 3.33%, 30.4%), 6.82% (P ₂₅ , P ₇₅ : 4.31%, 9.06%), 5.40% (P ₂₅ , P ₇₅ : 2.60%, 8.26%), 4.90% (P ₂₅ , P ₇₅ : 2.21%, 9.68%), 4.32% (P ₂₅ , P ₇₅ : 2.17%, 6.78%), 5.62% (P ₂₅ , P ₇₅ : 2.96%, 8.19%), and 5.35% (P ₂₅ , P ₇₅ : 2.55%, 7.81%), respectively. The Kruskal-Wallis H test results revealed significant differences among the eight carts/systems (H=274.86, P<0.001), and the pairwise comparisons showed that cart/system F had the lowest absolute percentage difference of 4.32% (P ₂₅ , P ₇₅ : 2.17%, 6.78%). The hierarchical cluster classified carts/systems A and B as one cluster, carts/systems C, E, and H as another cluster, and carts/systems D, F, and G as yet another cluster.
Conclusions: Using an MS can decrease measurement errors and variability for CPET. It can also improve the quality control of CPET.