Examination of gas exchange and blood lactate thresholds in Paralympic athletes during upper-body poling

The primary aim was to compare physiological and perceptual outcome parameters identified at common gas exchange and blood lactate (BLa) thresholds in Paralympic athletes while upper-body poling. The secondary aim was to compare the fit of the breakpoint models used to identify thresholds in the gas exchange thresholds data versus continuous linear and curvilinear (no-breakpoint) models. Fifteen elite Para ice hockey players performed seven to eight 5-min stages at increasing workload until exhaustion during upper-body poling. Two regression lines were fitted to the oxygen uptake (VO2)-carbon dioxide (VCO2) and minute ventilation (VE)/VO2 data to determine the ventilatory threshold (VT), and to the VCO2-VE and VE/VCO2 data to determine the respiratory compensation threshold (RCT). The first lactate threshold (LT1) was determined by the first rise in BLa (+0.4mmol·L-1 and +1.0mmol·L-1) and a breakpoint in the log-log transformed VO2-BLa data, and the second lactate threshold (LT2) by a fixed rise in BLa above 4mmol·L-1 and by employing the modified Dmax method. Paired-samples t-tests were used to compare the outcome parameters within and between the different threshold methods. The fit of the two regression lines (breakpoint model) used to identify thresholds in the gas exchange data was compared to that of a single regression line, an exponential and a 3rd order polynomial curve (no-breakpoint models) by Akaike weights. All outcome parameters identified with the VT (i.e., breakpoints in the VO2-VCO2 or VE/VO2 data) were significantly higher than the ones identified with a fixed rise in BLa (+0.4 or +1.0mmol·L-1) at the LT1 (e.g. BLa: 5.1±2.2 or 4.9±1.8 vs 1.9±0.6 or 2.3±0.5mmol·L-1,p0.06). The outcome parameters identified with breakpoints in the VCO2-VE data to determine the RCT (e.g. BLa: 5.5±1.4mmol·L-1) were not different from the ones identified with the modified Dmax method at the LT2 (5.5±1.1mmol·L-1) (all p>0.53), but were higher compared to parameters identified with VE/VCO2 method (4.9±1.5mmol·L-1) and a fixed BLa value of 4mmol·L-1 (all p0.931), the continuous no-breakpoint models had the highest probability (>68%) of being the best models for the VO2-VCO2 and the VCO2-VE data. In Paralympic at