Walking tests are frequently used to examine ambulation in MS patients [1, 2]. However, analytic procedures applied to data derived from walking tests, with few exceptions [8–10], remained relatively superficial, leaving temporal walking dynamics almost unexplored. We examined three characteristics of walking behavior in two walking tests, comparing MS patients of mild and moderate disability to healthy controls. Besides the common parameter of mean walking speed, velocity profiles included the linear and the quadratic trend of walking speed during a 6 min Walk (6MW) and a 12 Minute Walk (12MW). The linear trend reflected a measure of deceleration over time, while the quadratic trend estimated the degree to which the walking profile of each group approximated a U-shape. With this analysis, we intended to confirm and extend observations made by Goldman et al.  who reported distinct patterns in the walking behavior of MS patients and healthy controls during the 6MW.
Mean walking speed
Firstly, our results are consistent with findings of Goldman et al. , confirming that MS patients walked slower than controls in both tests. In our sample, both, mildly and moderately disabled MS patients displayed attenuated walking speed, relative to controls. As MS commonly affects ambulation, this observation is in line with the extant literature [4, 6, 9]. The fact that even MS patients with mild disability (Mild MS group) differed significantly from controls in their mean walking speed is noteworthy. This finding hints at an early influence of MS on walking ability in the examined sample and supports previous reports [7, 10, 16–18].
Linear trend in walking speed: mean decline
More importantly, results of the current study indicate that differences in walking behavior between MS patients and controls do not only manifest in mean walking speed, but also in altered dynamic walking parameters.
MS patients with moderate disability slowed down more rapidly than controls on both tests. Complementary, MS patients with mild disability displayed a pronounced deceleration in case of the 12MW. These novel findings indicate that continuous deceleration, as reflected by the linear trend in walking speed, represents an additional, clinically relevant feature of impaired ambulation in MS patients. While for patients with moderate disability, the 6MW represents a test which is sufficient to detect this clinical feature, for patients with mild disability, a considerably longer walking duration, as in the 12MW, seems necessary.
The linear decline in walking speed was significantly pronounced in patients, relative to healthy controls, which indicates that deceleration represents a clinically relevant feature. However, the current study provides further original information on the utility of this parameter. In particular, the linear decline in walking speed was highly correlated with subjective somatic fatigue. This provides convergent evidence, indicating that the pronounced deceleration also represents a good estimate of patients' subjective constraints. It is especially noteworthy that linear deceleration showed highly significant correlations with subjective somatic fatigue (all p-values < 0.001), whereas the commonly used parameter of mean walking speed showed only minor associations with somatic fatigue (all p-values < 0.05). Since these correlations differed significantly from each other in the 6MWb, linear deceleration may in fact represent a more suitable parameter to assess somatic fatigue than mean walking speed. This appears plausible, since the dynamic notion of deceleration seems more congruent with the phenomenon of motor fatigue, than the mean walking speed. In sum, our results indicate that deceleration reflects a potentially useful parameter, which is suitable for the assessment of somatic fatigue. As such, our findings provide strong support for the notion of Goldman et al. , who suggested to interpret the decline in walking speed as an indicator of fatigue in MS patients.
Quadratic trend in walking speed: degree of observed u-shape
We also found a U-shaped velocity profile (quadratic trend) across both tests in all groups combined. A U-shaped pacing strategy is a known phenomenon in healthy individuals . As hypothesized, the degree of the U-shape was attenuated in MS patients, relative to controls in the 6MW. Even MS patients with mild disability showed an attenuated U-shape relative to controls on this test. This indicates, that the U-shaped profile can also be regarded as a clinically relevant parameter in MS. However, compared to the parameter of linear deceleration, it does not appear to be as informative, since no association with subjective fatigue was obtained and differences in the U-shaped pattern between the studied groups were not found in the 12MW. Nevertheless, since studies in which this dynamic characteristic is utilized are sparse, the current work provides original results on a new, promising clinical feature, worthy to be explored in more detail in the future.
Which walking test to use?
Recently, it has been suggested that a brief test including only two minutes of walking (2MW) represents a sufficient measure of walking ability in MS patients . This suggestion was based on a high correlation between the distances covered during the 2MW and the 6MW . Given this correlation, such a suggestion appears feasible when considering the total distance walked. Nevertheless, results of the current study suggest that the choice of which test to use depends on the walking parameter which is supposed to be assessed. For practical reasons, the parameter of mean walking speed, or total distance covered during a given amount of time, has received most attention in research on ambulation in MS. The current study provides novel findings, according to which further parameters can be derived from standard walking tests, which are of high clinical relevance. Particularly the linear trend, as a measure of deceleration, appears to be promising as an estimate of impaired walking ability, since it can be easily derived from a standard walking test. However, since it is not necessarily warranted to derive this measure from a test of only two minutes duration, the current findings do not provide direct support for the suggestion to drastically shorten common tests. In line with this conclusion, Motl et al.  have reported that a further putatively relevant parameter, i.e. oxygen consumption, does not reach a steady state within the first two minutes of walking, but remains unaltered only after the third minute. The latter authors suggest that the 2MW functions as a measure of primarily anaerobic and the 6MW as a measure of primarily aerobic performance. Based on our results obtained in case of the 6MW, particularly this test appears to be a feasible measure to capture both, well established parameters, such as the total distance walked, as well as parameters which vary during the walking test. Our findings indicate that an abbreviation of the test duration might result in the loss of potentially important information on dynamic walking behavior, which to date has remained relatively unexplored. Dynamic parameters could serve as a means to increase sensitivity of walking tests to abnormal walking behavior within MS patients.
Limitations and future directions
In an innovative study Phan-Ba et al.  have recently utilized the combination of a timed 500 meter walk (T500MW) and a timed 25 foot walk (T25FW) to assess deceleration in MS. The authors quantified deceleration by computing a combined deceleration index, consisting of the ratio of walking speed during the last 100 meters of the T500MW and throughout the T25FW. The authors suggest that such a ratio may be particularly useful to assess ambulation impairment during late stages of MS. Hence, future studies which distinguish between MS subtypes and which explore methodological means to derive further parameters of dynamic walking characteristics may be warranted.
While the current results are in accord with this assumption, they are to be interpreted in the context of few limitations. In particular, it should be noted that the quantification of walking behavior was achieved through visual inspection by an examiner equipped with a stop-watch. This setting was chosen over the one in which a patient is accompanied by an examiner equipped with a precise measurement wheel, to reduce a putatively confounding influence of the setting on walking performance. It is also the approach commonly used in the clinic where the work was carried out. On the other hand, it can be argued that visual inspection may also have an effect on walking performance, and the use of a measurement wheel would have provided a technically more precise measure of walking speed. Hence, the examiner could either have accompanied the patient with the measurement wheel, or the patient could have been equipped with the measurement wheel. Either option may represent a useful alternative to the method implemented in the current study. This fact may be regarded as a caveat to the interpretation of the current results.
Finally, it needs to be noted that the walking parameters explored in the current study imply an alteration of the common nomenclature used in the literature. Performance on walking tests is usually referred to in terms of the total distance covered during a given amount of time. For the use of walking behaviour which changes throughout the test (e.g. deceleration) as a test parameter, a nomenclature involving walking speed appeared more feasible. Nevertheless, when it comes to the interpretation of the current results, it needs to be considered that this nomenclature does not match the one of the extant literature on common walking parameters.