Protocol changes can improve the reliability of net oxygen cost data

There are concerns that experimental imprecision of measured oxygen cost reduces its clinical utility. These concerns are exacerbated when net energy (walking − resting) is assessed. Therefore, the variability of net oxygen cost should be carefully evaluated, including an exploration of the potential for protocol changes to improve the measureʹs reliability. In this study a retrospective analysis was used to compare the variability in net oxygen cost between two protocols at a single center. Two groups of subjects (protocols) were analyzed. The subjects in each group met the following criteria: (i) diagnosis of cerebral palsy, (ii) no prior surgery, and (iii) oxygen cost data collected while walking barefoot. The protocols differed in terms of subject preparation, duration of the resting period, and the method used for determining both resting and walking steady state. The net non-dimensional oxygen cost versus dimensionless speed showed essentially identical quadratic regression fits for the two groups, suggesting that the samples were well matched, and the protocols were unbiased. The variability, as measured by mean square error, showed an approximately five-fold reduction with respect to the regression (MSEold = .262, MSEnew = .050), indicating a significant decrease in experimental errors. Further analysis is warranted to determine the primary sources of error reduction, and to further optimize the testing protocol.