Can we trust self-reported walking distance when determining EDSS scores in patients with multiple sclerosis? The Danish MS hospitals rehabilitation study

Study design

The current study is a retrospective analysis using data from the Danish MS hospitals rehabilitation study. Between March 2012 and May 2014, n = 427 participants completed 4 weeks of specialized MS rehabilitation at either the Danish MS hospital in Ry or in Haslev. The main study investigating long-term effectiveness of in-patient multidisciplinary rehabilitation on health-related quality of life have previously been published ¹³ along with a detailed description of the full study protocol. ¹⁴ The Danish MS hospitals rehabilitation study was approved by the research Ethics Committee, Region of Zealand, Denmark (ref. no. 1-01-83-0002-07) and the Danish Data Protection Agency has granted permission to collect and store the required project information (ref. no. 2011-41-6751).

Data collection

At baseline, patients were instructed to classify their self-perceived maximum walking distance (based on their “usual best function” in every daily life) according to one of seven standardized categories contained in the EDSS (>500; 300–499; 200–299; 100–199; 20–99; 5–19; 0–4 m) including any use of unilateral or bilateral assistance (i.e. walker, cane(s), crutch(es), ankle-foot orthotic device(s)) while walking. To determine the level of ambulation, patients performed an indoor maximum walking distance test on a 32-meter marked lane. An experienced assessor supervised and observed the walking test that was terminated when the patient was unable to walk any further or when 500 m was clearly passed (by >50 m). After the walking test, all patients underwent a full neurological examination by a trained neurologist to determine EDSS scores based on the patient’s actual walking performance (EDSS_actual). ⁷ Retrospectively, the EDSS from self-reported walking distance (EDSS_self-report) was also determined in order to detect potential differences from EDSS_actual. As an inclusion criteria of the present analysis, only MS patients with an EDSS_actual score ⩾4.0–7.5 were included (from the total of n = 427 participating in the Danish MS hospital study, Figure 1), since EDSS scores in this range are heavily influenced by walking distance and the use of assistive devices.^3,7 This comprised n = 303 MS patients, yet due to missing data (n = 30) complete datasets were available from n = 273 MS patients to be used in the present study (Figure 1).

OPEN IN VIEWER

Figure 1.

Flow diagram describing the enrollment procedure of the Danish MS Hospitals Rehabilitation study and the final inclusion procedure of the current analysis.

Data analysis

The data were analyzed using STATA version 14.2. Demographics were calculated as mean ± SD, unless otherwise stated. As a first step, patients were categorized as either “correct classifiers” or “mis-classifiers.” This categorization was based on the correspondence between actual maximal walking distance and the correspondence to the category obtained from self-report. No margin of discrepancy was allowed in order to be categorized as “correct classifier.” Kappa-statistics was applied to establish the agreement between actual maximal walking distance and the category obtained from self-report. A kappa score (K) of “zero” reflects the level of agreement that can be expected by chance alone, whereas a score of “one” is when a perfect agreement is observed. For intermediate values, the interpretation suggested by Landis and Koch was followed, stating that a K-score below 0.0 is poor, 0.00–0.20 is slight, 0.21–0.40 is fair, 0.41–0.60 is moderate, 0.61–0.80 is substantial, and 0.91–1.00 is almost perfect. ¹⁵ Next, the proportion of patients that underestimated or overestimated their actual maximal walking distance when self-reporting were determined among the “mis-classifiers.” To further identify the influence of patient characteristics, subgroup analyses of the total sample were made for MS phenotype (relapsing remitting, secondary progressive, primary progressive), level of EDSS, the use of walking aids, and the type of walking aid using a two sample test of proportions (2 groups) or a chi-square test (>2 groups). Finally, EDSS_actual was compared to EDSS_self-report. The statistical significance level was set at p < 0.05.

Actual versus self-reported walking distance

A total of 146 patients (53%) misclassified their maximal walking distance, whereas 127 patients (47%) made correct classifications, corresponding to their actual maximal walking distance. Kappa statistics showed that 18% would expectedly be correctly classified by chance, as compared to the observed 47%, giving a K-score of 0.36 (“fair-agreement,” p < 0.05). Of those that misclassified, significantly more were underestimating (78%) than overestimating (22%) their actual maximal walking distance (p < 0.05, Figure 2(a)).

OPEN IN VIEWER

Figure 2.

Actual versus self-reported maximal walking distance in MS patients (n = 273). Proportion of MS patients that misclassify or make correct classification of their actual maximal walking distance, including levels of over- and underestimation in the group that misclassify. (a) Total sample, (b) MS phenotypes, (c) EDSS levels, (d) Walking aid (yes/no).

Misclassification according to subgroup

A significant difference in the proportion of misclassifiers was observed when dividing patients into groups based on MS phenotype (primary progressive MS differed from relapsing remitting and secondary progressive MS, p < 0.05, Figure 2(b)) and walking aid usage (p < 0.05, Figure 2(d)). The type of aid (uni-, bilateral cane, or walker) did not influence the results.

Impact of misclassification on EDSS scoring

Comparison of EDSS_actual and EDSS_self-report revealed that EDSS scores differed by ⩾0.5 point in 24% (n = 65) of the patients (18% under (n = 49), 6% over (n = 16)). In 9% (n = 24) of the patients, the EDSS score was affected by ⩾1.0 point (8% under (n = 21), 1% over (n = 3)) (Figure 3). When dividing these participants into subgroups of EDSS 4.0–5.5 and EDSS 6.0–7.5 (uni- or bilateral assistance/wheelchair users), respectively, the EDSS_actual and EDSS_self-report scores differed in 38% (n = 44) and 13% (n = 20) of the subjects, respectively (Figure 3).

OPEN IN VIEWER

Figure 3.

Proportion of MS patients with EDSS scores based on self-reported maximal walking distance that is either similar or deviate from EDSS scores based on actual maximal walking distance (i.e. EDSS_self-report vs EDSS_actual) (n = 273).

Misclassification of walking performance

Several aspects may explain the observed misclassification of maximal walking distance. First, self-reported walking distance is challenging and depends on different parameters such as the patients’ overall ability to judge distance and their ability to rate their own current physical condition. Second, patients may be somewhat “modest” when rating their own abilities. In the present study, patients had to self-report on their usual walking performance just before performing the actual walking test, whereas in daily clinical practice, a dialogue about the patients walking ability (i.e. self-reporting) is probably the most common procedure. While an actual walking test is superior to self-reported walking capacity, a dialogue that takes different aspects of everyday living into account should not be neglected, especially when considering the between- and within-day variations in walking performance previously reported in MS patients.^16,17 Albrecht et al. ¹⁶ showed that between-day variability in unaided walking in certain cases could lead to EDSS variations up to ±1.5 points. On this basis, the authors concluded that assessment of walking performance in MS patients should not be based on a single measurement, and that gait speed might be more consistent compared to walking distance. ¹⁶ In agreement, our study underlines the importance of an accurate measure of walking distance to ensure reliable EDSS scoring. Despite between-day variability in walking distance, it does, however seem impractical to conduct several measurements on walking distance as proposed by Albrecht et al. ¹⁶ Third, a reason for patients underestimating their walking performance might be that the actual walking test was carried out at a safe indoor facility, without disturbances related to the surface of the track, the surrounding traffic, or the weather conditions. This may generally have allowed patients to walk further than during real life settings. Of note, a generally accepted standardization of how to assess maximal walking distance during EDSS assessment is however warranted if the EDSS should become more responsive to changes in walking performance. Such standardizations would optimally include standardization of preceding physical activities as well as the patients’ medication routines. Time of day seems of less importance as Feys et al. ¹⁸ have shown consistent walking performance during the day.

Similar to our findings, a recent pilot study by Berger et al. ⁹ reported that 44% misclassify their maximal walking capacity leading to a change in EDSS. However, they found that only 24% of those misclassifying underestimated their actual maximal walking capacity as opposed to 78% in the present study. A plausible explanation for this discrepancy is likely due to the relatively small sample size (n = 66) in the study by Berger et al. ⁹ compared to the present study (n = 273). Moreover, participants in the study by Berger et al. ⁹ viewed the walking course and subsequently self-reported (estimated) how far they could walk on this course, whereas in the present study participants self-reported walking at their “usual best function” in every daily life according to specific categories. Also, due to the much larger sample size, we believe that our in-depth analysis evaluating subgroups at particular risk of misclassifying their walking ability is more robust than the one carried out by Berger et al. ⁹ Berger et al. ⁹ did not identify any subgroups at particular risk, whereas we identified MS phenotype as a factor affecting classification of walking ability (primary progressive misclassified to a greater extent than relapsing remitting and secondary progressive, Figure 2(b)). This may be explained by the fact that the primary progressive MS patients were older and had more marked impairments in walking ability (and plausible in cognitive function, although not assessed in the present study participants) compared to relapsing remitting and secondary progressive MS patients, despite comparable mean EDSS scores (relapsing remitting 5.0, secondary progressive 5.8, primary progressive 5.7). EDSS level did not appear to predict the ability to estimate walking distance in MS patients. However, the use of assistive devices was shown to increase the risk of misclassifying maximal walking distance in the present sample, although being independent of the type of aid. One explanation could be that multiple factors influence walking when using aids, such as upper body strength and coordination as well as fatigue/fatigability and motivation.

Limitations

This study has some limitations that must be kept in mind when interpreting the results. First, the influence of cognitive status on the ability to estimate walking performance was not evaluated. It seems plausible that the ability to estimate walking performance is affected in patients with marked cognitive deficits. However, this was likely a minor issue as the inclusion criteria of the Danish MS Hospitals Rehabilitation Study only allowed participation of patients with sufficient cognitive status (functional system score ⩽ 2) to complete a number of self-reported questionnaires. ¹⁴ To establish the extend that cognitive impairment impacts the ability to classify walking distance, future studies could add the Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS) or Brief Repeatable Battery of Neuropsychological tests (BRB-N) to a study design similar to the present study. Furthermore, exclusion of patients based on the cerebellar functional system (FSS ⩽ 2) should be avoided as inclusion criteria. Second, the maximal walking distance test was performed on an indoor 32-meter lane, where patients were instructed to turn around cones at each end. Although most patients were unaffected by this, it is possible that those with balance issues or aid users with mobility problems were limited by these turns.

Clinical implications and perspectives

Since EDSS is an important tool when making clinical decisions as well as a measure of disease progression, any change in EDSS must be regarded as relevant. The present study demonstrates that inability to correctly classify walking ability by self-reporting – which is particularly high in patients using assistive devices and in those having primary progressive MS – leads to incorrect EDSS scoring by ⩾0.5 point in 24% of MS patients and by ⩾1.0 point in 9% of MS patients. Hence, our results emphasize that walking ability should optimally be based on actual (objectively measured) walking distance to keep EDSS scoring valid, thus in agreement with the official recommendation of the Neurostatus EDSS scoring. Even an error difference of 0.5 in EDSS (and of 1.0 in particular) may be a concern to address since this might impact clinical decision-making. ¹¹ Due to the practical challenges of carrying out long walking distance tests (i.e. time and facilities), one suggestion could be that future studies investigate whether a short walking distance test (e.g. the 25-foot walk) could provide a valid estimate of walking distance. In support hereof, previous studies have shown strong correlations between the short distance walking tests and the 6-minute walking test,^19,20 so it might be possible to establish valid EDSS scoring based on T25FW performance. However, an important point presented by Cohen et al. ² states that, the EDSS should not be fundamentally changed to keep its acceptance by regulators.

Another interesting observation was that the concordance between EDSS_self-report and EDSS_actual was greater in MS patients having an EDSS_actual of 6.0–7.5 compared to MS patients having an EDSS_actual of 4.0–5.5 (Figure 3). As EDSS scoring did not affect the extent of misclassifying walking ability (Figure 2(c)), this discrepancy is likely explained by the fact that when MS patients use assistive devices (which corresponded to 84% in MS patients having EDSS_actual 6.0–7.5), the actual walking performance generally overrules the walking range when determining the EDSS step, although the use of aids does not predict the final EDSS alone. Of note, the EDSS scale does not provide detailed distinction of walking ability and disease progression in EDSS scores 6.0–6.5. One suggestion could therefore be that EDSS scores 6.0–6.5 are always reported along with the actual walking distance. By reporting this, more information on MS progression will be identified.