The influence of the regression model and final speed criteria on the reliability of lactate threshold determined by the Dmax method in endurance-trained runners.

Few studies verified the reliability of the lactate threshold determined by Dmax method (LT) in runners and it remains unclear the effect of the regression model and the final speed on the reliability of LT. This study aimed to examine the test-retest reliability of the speed at LT in runners, considering the effects of the regression models (exponential-plus-constant vs third-order polynomial) and final speed criteria (complete vs proportional). Seventeen male, recreational runners performed 2 identical incremental exercise tests, with increments of 1 km·h each for 3 min on treadmill to determine peak treadmill speed (V) and lactate threshold. Earlobe capillary blood samples were collected during rest between the stages. The V was defined as the speed of the last complete stage (complete final speed criterion) and as the speed of the last complete stage added to the fraction of the incomplete stage (proportional final speed criterion). Lactate threshold was determined from exponential-plus-constant and from third-order polynomial regression models with both complete and proportional final speed criteria and from fixed blood lactate level of 3.5 mmol·L (LT). The LT obtained from the exponential-plus-constant regression model presented higher reliability (coefficient of variation (CV) ≤ 3.7%) than the LT calculated from the third-order polynomial regression model (CV ≤ 5.8%) and LT (CV = 5.4%). The proportional final speed criterion is more appropriate when using the exponential-plus-constant regression model, but less appropriate when using the third-order polynomial regression model. In conclusion, exponential-plus-constant using the proportional final speed criterion is preferred over LT and over third-order polynomial regression model to determine a reliable LT.