Prediction of normal values for lactate threshold estimated by gas exchange in men and women

• Published in: European journal of applied physiology Vol. 76; no. 2; pp. 157 - 164
• Main Authors: Davis, J A, Storer, T W, Caiozzo, V J
• Format: Journal Article
• Language: English
• Published: Germany 01-08-1997
• Subjects: Adult; Aged; Carbon Dioxide - blood; Ergonomics; Female; Humans; Lactates - blood; Linear Models; Male; Middle Aged; Models, Biological; Oxygen - blood; Pulmonary Gas Exchange - physiology; Reproducibility of Results
• ISSN: 0301-5548; 1439-6319; 1439-6327
• DOI: 10.1007/s004210050228

Lactate threshold (LT) is an index of exercise capacity and can be estimated from the gas exchange consequences of a metabolic acidosis (LT(GE)). In recent years, it has emerged as a diagnostic tool in the evaluation of subjects with exercise limitation. The purpose of this study was to develop LT(GE) prediction equations on a relatively large sample of adults and to cross-validate each equation. A total of 204 healthy, sedentary, nonsmoking subjects (103 men and 101 women), aged 20-70 years, underwent graded exercise testing on a cycle ergometer. The V-slope technique was used to detect LTGE as the oxygen uptake (VO2) at the breakpoint of the carbon dioxide output versus VO2 relationship. Multiple linear regression was used to develop 12 equations with combinations of the following predictor variables: age, height, body mass, and fat-free mass. Eight of the equations are gender-specific and four are generalized with gender as a dummy variable. The equations were cross-validated using the predicted residual sum of squares (PRESS) method. The results demonstrate that the equations had relatively high multiple correlations (0.577-0.863) and low standard errors of the estimate (0.123-0.228 1 x min(-1)). The PRESS method demonstrated that the equations are generalizable, i.e., can be used in future studies without a significant loss of accuracy. Since we tested only healthy, sedentary subjects, our equations can be used to predict the lower limit of normal for a given subject. Using individual data for healthy and diseased subjects from the literature, we found that our gender-specific equations rarely miscategorized subjects unless they were obese and mass was a predictor variable. We conclude that our equations provide accurate predictions of normal values for LT(GE) and that they are generalizable to other subject populations.