The dark side of the treadmill walking test


Dear Editor,

We read with interest the review by Robinson et al. [1]. They found that treadmill training improved gait speed by about 21 cm/s in people with Parkinson's disease (PwPD). This value exceeded the minimal clinically important difference (MCID). However, in our opinion, this change in gait speed is overestimated and affected by the methods of one of the studies [2], included in Fig. 2 of the review [1].

Gait speed is commonly used to track progression through rehabilitation [3] and has been considered a primary outcome in systematic reviews on PwPD [14567]. However, the choice of some of these reviews to lump together in a single forest plot gait speed data collected from different tests is, in our opinion, questionable. In fact, data from heterogeneous tests could make the interpretation of results misleading and invalidates the application of the MCID. Gait speed changes from 13 to 18 cm/s obtained with a walkway system have been reported to be clinically important in PwPD [8]; but how appropriate is it to use this MCID to interpret gait speed change obtained from the Timed Up and Go (TUG) test [457], 6-Min Walk Test [45], or even from a treadmill walking test (TWT), as done by Robinson et al. [1]?

To try to answer this question, we calculated gait speed changes in PwPD who underwent 15 days of treadmill training. A secondary analysis of data from our recent trial on treadmill training [9] showed that the mean change on the treadmill was double that on the walkway system and almost three times that on the TUG test (Fig. 1A). These findings cast doubts on the feasibility of interpreting results from different tests with a single MCID value and raise a note of caution about the transfer to daily life of rehabilitative gains with the treadmill. Another critical aspect linked to using the treadmill as an outcome measure is familiarization. In healthy subjects 7 minutes could be sufficient to demonstrate adaptation to the treadmill [10], but in our experience PwPD require a longer learning time of at least 3 treadmill training days (Fig. 1B). Therefore, we believe the inclusion in the review [1] of the study by Ganesan et al. [2] that assessed gait speed with TWT in naïve PwPD overestimates the effectiveness of treadmill training.