TO THE EDITOR:
We recently had an opportunity to study carefully the paper by Dufek and coworkers (Dufek J, Mercer JA, Sherwood Kosta C, et al. Metabolic cost during submaximal walking with a rigid rotational-control ankle–foot orthosis: A preliminary investigation. J Prosthet Orthot 1997;9:152–156), which examined the effects of rotational-control orthoses on energy expenditure and perceived exertion. Although the article has not been widely cited, we have heard speakers refer to this article as providing evidence for the effectiveness of rotational-control orthoses. The article appears to have received an uncritical acceptance in the professional literature that we feel is unwarranted. For this reason, we are taking the unusual step of writing to you so long after the original publication of the research.
Briefly, the investigators studied oxygen consumption (VO2), heart rate (HR), and ratings of perceived exertion (RPE) in treadmill walking with and without an orthosis in six individuals who were using rigid rotational-control ankle–foot orthoses (AFOs) to manage gait abnormalities and lower limb joint deviations. The authors reported significant decreases in VO2 and RPE with orthosis use and noted high positive correlations between VO2 and both HR and RPE. In explaining these results, the authors drew heavily on the ability of the orthosis to control out-of-plane movements.
Because the raw data for the study were replicated in the article, we were able to repeat the statistical analyses reported. The results of our reanalyses differed from those reported in the article in a number of important respects.
The authors reported that there was a significant difference in VO2 between the no-orthosis condition and the orthosis condition. We duplicated the repeated-measures analysis of variance, including data for four subjects (data for two subjects were excluded from the analysis by the authors for various reasons). Our results indicated that although the results were close to being statistically significant, the results were not significant at the p = .05 level (F = 9.832, df = 1,3, p = .052). It could seem pedantic to take issue with such a small difference, but we also believe that the choice of statistical test was incorrect. When subject numbers are small, statisticians usually recommend use of nonparametric statistical tests because these tests are more robust to violations of the assumption of homogeneity of variance that are likely to occur in small samples. The equivalent nonparametric test in this case is Wilcoxon’s signed rank test. The result of this test was not statistically significant (z = 1.826, p = .068).
The authors reported a significant difference in RPE values between the orthosis and no-orthosis condition, but our reanalysis indicated that the difference was not significant (F = 8.0, df = 1,3, p = .066). The more appropriate nonparametric analysis (Wilcoxon’s signed rank test) was also not significant (z = 1.633, p = .102).
The authors noted a strong relationship between VO2 and heart rate (r = 0.884). Our reanalysis indicated that the correlation was somewhat lower (r = 0.848) and not statistically significant (p = .07). Although there is little doubt that VO2 and heart rate are related within individuals, differences in fitness mean that the heart rate responses of individuals to similar workloads are different. It makes little sense to examine this relationship between individuals, as was done by Dufek and coworkers.
The authors also noted a strong relationship between VO2 and RPE (r = 0.930) and suggested that RPE could therefore be used as a clinical measure of metabolic efficiency. Our reanalysis showed that the magnitude of the correlation was correct as reported but the correlation was actually negative, indicating that higher VO2 was associated with lower RPE. (Indeed, this was the most highly statistically significant result [p = .023] in the study.) Borg’s1 RPE scale was developed as an intraindividual scale. Numerous investigations have demonstrated high correlations between physiological measures of exertion and the 15-point Borg scale, and there is little doubt that it is an appropriate measure of exertion within subjects.2 Comparisons between individuals of relative exertion can be made if we assume that each individual subjectively rates exertion relative to their own perceived maximum exertion.1 Because aerobic capacity varies widely among individuals depending on fitness and activity levels, it is not surprising that a positive correlation was not observed.
Overall, the RPE and oxygen consumption results indicate a trend for lower energy expenditure with orthosis use. In their discussion, the authors suggest that the explanation for the reduction in VO2 is that unwanted and energy-wasting out-of-plane rotations were controlled by the orthosis. Unfortunately, the comparison of a no-orthosis condition with an orthosis condition does not enable conclusions to be drawn about any particular feature of the orthosis. The only way that such a conclusion could be drawn would be to test two orthosis conditions: one with rotational-control features and another, identical in every other respect (mass, trimlines, flexibility, and so on), but without rotational control features. If a significant difference between the two orthoses could then be shown, these differences might be attributed to rotational control features.
It is important to establish an evidence base for orthotic practice. The authors are to be commended on their effort to contribute to this base. The article indicates a strong trend for more efficient gait with orthosis use. Unfortunately, like many other papers in the prosthetics and orthotics literature, it suffers from a small sample size. The statistical analysis does not bear careful scrutiny. Many of the conclusions made by the authors are not supported by the evidence presented. Importantly, the article provides no evidence that orthoses with rotational-control features are more effective than those without such features.
Margaret C. Hodge, BP&O
Timothy M. Bach, PhD
National Centre for Prosthetics and Orthotics
La Trobe University
1. Borg G. Borg’s Perceived Exertion and Plain Scales.
Champaign, IL: Human Kinetics; 1998.
2. Chen MJ, Fan X, Moe ST. Criterion-related validity of the Borg ratings of perceived exertion scale in healthy individuals: A meta-analysis. J Sports Sci