The O2uptake (V O2) data at the onset of an exercise are usually fitted with a mono-exponential function, after removal of the data pertaining to a conventional initial time period (DTr) lasting _20 s. We performed a thorough quantitative analysis on the effects of removing data pertaining to different DTr, aiming at identifying an objective method to establish the appropriate DTr. Breath-by-breath O2uptake responses, acquired from 25 healthy adults performing a step moderate-intensity exercise, and 104 simulated biexponential responses, were analyzed. For all the responses, the kinetic parameters of a mono-exponential function and the corresponding asymptotic standard errors (ASEs) were estimated by nonlinear regression, removing the data pertaining to progressively longer initial periods (1 s each) up to 60 s. Four methods to establish objectively DTr were compared. The minimum estimated s was obtained for DTr % 35 s in both the V O2and simulated data, that was about 30% lower compared with that obtained for DTr % 0s. The average ASE values remained quite constant up to DTr % 35 s, thereafter they increased remarkably. The s used to generate the simulated response fell within the confidence intervals of the estimated s in _85% of cases for DTr = 20 s ("20 s-w" method); this percentage increased to _92% of cases when DTr was established according to both the minimum s and its narrowest confidence interval ("Mixed" method). In conclusion, the effects of removing V O2data pertaining to different DTr are remarkable. The "Mixed" method provided estimated parameters close to those used to generate the simulated responses and is thus endorsed.
