Sources of error and variability in the determination of anaerobic threshold in healthy humans

The derivation of anaerobic threshold (AT) from the ventilatory responses to incremental exercise is associated with several sources of variability including true biological variability and the error attributable to the observer interpretation of data. To define and quantitate the sources of variability in AT determination we exercised 6 healthy volunteers 6 times and submitted plots of ventilation (Ve), CO2 production (VCO2), respiratory exchange ratio (R) and the ventilatory equivalent for oxygen (Ve/VO2) in random order to 4 independent observers. Within-subject variability in AT ranged from 7 to 55% depending on the subject, ventilatory parameter and observer with an overall mean coefficient of variability of 24%. Significant day-to-day variability was demonstrated in 4 of the 6 subjects using AT values derived from at least one of the ventilatory parameters (Anova, p less than 0.05-p less than 0.001). Mean values for AT obtained with the Ve and VCO2 plots (1.91 and 1.69 liter/min VO2) were significantly lower than those obtained from R and Ve/VO2 plots (2.28 and 2.6 liters/min VO2; p less than 0.001, Anova). Anova showed significant differences in AT values derived by one of the observers compared to the other three (p less than 0.001). A significant observer/ventilation parameter interaction was also found (p less than 0.001) due to one observer consistently estimating higher values of AT from the R plots. The observer error in deriving AT from each exercise test using the Ve plots = 24%, for VCO2 = 19% for R = 29% and for Ve/VO2 = 15%.(ABSTRACT TRUNCATED AT 250 WORDS)