Exercise intensity influences variability and test-retest reliability of pulmonary gas-exchange measurements during...

The study examined how exercise intensity and participants' characteristics influence the variability and test-retest reliability of pulmonary gas-exchange while providing domain-specific reliability benchmarks for exercise testing.

Forty-two healthy adults (V̇O_2max = 49.6 ± 8.8 ml·kg_-1·min_-1) completed a ramp incremental exercise test and six 6-min constant work-rate cycling trials distributed across moderate, heavy, and severe exercise intensities. Exercise intensities were set at 33%Δ and 66%Δ between baseline and gas exchange threshold (GET), GET and respiratory compensation point (RCP), and RCP and V̇O2max, respectively. Pulmonary oxygen uptake (V̇O₂), carbon dioxide output (V̇CO₂), and ventilation (V̇E) were recorded using two metabolic carts. Systematic differences between days and intensities were assessed with two-way repeated-measures ANOVA. Variability and participants' characteristics were examined with within-trial coefficient of variation (CV%), Pearson correlations, linear mixed-effects models (LMME), intraclass correlation coefficient (ICC), standard error of measurement (SEM), minimal detectable change (MDC), and Bland-Altman analysis.

V̇O₂, V̇CO₂, and V̇E increased with intensity (p ≤ 0.001), with no day-to-day differences (p > 0.05). CV% decreased progressively with intensity, from ∼7-10% in the moderate to <5% for V̇O₂ and V̇CO₂ in the severe domain. ICCs were excellent (>0.95) in the heavy and severe domains. Correlations showed moderate negative associations between CV% and aerobic fitness (r -0.34 and -0.58, p < 0.05), while LMME identified exercise intensity as the main factor associated with variability. MDC for V̇O₂ ranged from 166 to 415 ml·min_-1, and Bland-Altman plots confirmed minimal bias.

exercise intensity influences the variability and reliability of breath-by-breath gas-exchange responses, with relative variability decreasing at higher metabolic loads. Fitness-related differences in CV% possibly reflect physiological scaling effects. These results provide practical reliability thresholds and improve the interpretation of repeated-measures testing in exercise physiology.