Quantity and Quality of Mobility in People with Multiple Sclerosis, Parkinson’s Disease and Matched Controls During Daily Living

Introduction

Mobility (gait and turning) is crucial for maintaining independence and quality of life. Gait and turning impairments are very common in patients with neurological disorders, leading to increased risk of falls and reduced quality of life [1]–[3]. Specific types of mobility impairments differ depending upon the neurological disorder. For example, weakness, spasticity, and ataxia impairments characterize mobility impairments in people with multiple sclerosis (MS), whereas bradykinesia, shuffling, freezing, and difficulties in turning characterize mobility impairments in people with Parkinson’s Disease (PD). Gait speed slows in individuals with any neurological disorder or aging [4]. However, slow gait is a general, universal characteristic of impaired mobility and may not be the most specific nor discriminative mobility impairment for each neurological disorder.

The standard neurological examination of mobility in the clinic (such as the 25-foot walk test for MS or the ‘postural instability and gait disability’ (PIGD) subscore of four items in the Motor Part III of the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS)) relies primarily on very short duration gait while being observed by a clinician. The key challenge in the clinic is that mobility impairments may not be observable in a small examination room nor during short walks when all attention is focused on walking [5]. Patients perform better in the clinic, while being observed (the white coat effect [6], [7]), than at home. Also, turning, a complex coupling of balance and gait critical for functional mobility is usually not even evaluated [8]. Further, the clinic observation at one-time point misses the daily fluctuations (variability) of mobility in response to medications, fatigue, and other conditions, and as a result, may not reflect actual, functional performance in the natural environment. Hence, there is an unmet need to determine the specific measures of mobility (gait and turning) that can objectively and continuously monitor mobility during daily life that are representative of different disorders.

Recently, the use of wearable sensors has made it possible to quantify digital mobility outside the clinic and during real-life [8]–[27]. Various researchers have shown that wearable sensors can be used to augment the standard clinical assessment based on active monitoring, such as performing prescribed/predefined tests at home or based on passive monitoring, during routine, daily life tasks in the community [28]. In this study, we derived digital outcome measures of mobility from passive monitoring from natural walking and turning during unrestricted daily activities over one week.

Clinical trials or clinical practices evaluating response to interventions need to specify which particular characteristics of mobility to monitor as digital-mobility outcome measures for each neurological disorder. Key challenges in using wearable sensors to monitor mobility in daily life are an excessive number of gait and turning measures and a lack of consensus on the most useful measures for each neurological disorder [11]. Mobility measures from passive continuous monitoring during daily living can capture both quantity characteristics (e.g. number of strides per hour, number of turns per hour, number of gait bouts, etc.) as well as quality characteristics (e.g., spatio-temporal gait or turning characteristics such as stride length, stride time variability, turning velocity, etc.) [29].

To reduce mobility characteristics, previous studies have used factor analysis or principle component analysis to identify separate domains of mobility. For example, lower body, upper body, and turning have been previously shown to be relatively independent during clinical gait testing [30]. In addition, activity and variability domains have been used to characterize mobility in the laboratory and during daily life [11], [19], [31], [32]. In this paper, we grouped our mobility measures into five domains: Upper Body, Lower Body, Turning, Activity and Variability. The measures incorporating all the domains of mobility, except activity, are referred as “quality of mobility”, and measures within the activity domain is referred to as “quantity of mobility”. Determining which mobility measures best discriminate mobility impairments related to severity of MS and PD compared to corresponding age-matched healthy control cohorts is needed to monitor the response to therapy [11], [13], [33]–[35].

The objective of the present study was to investigate which specific measures best discriminate mobility, during real-life monitoring, in people with MS and PD from their age-matched control cohorts and the extent to which these measures show face validity by relating to clinical evaluation of disease severity. We hypothesized that the most discriminative domains of mobility will differ for MS and PD, and the worse these mobility measures, the more severe the disease.

Methods

Results

Discussion

Our findings demonstrate that different digital measures over a week of daily monitoring discriminated mobility in people with MS than mobility in people with PD. Specifically, smaller median number of strides in the bout was the most discriminative measure in people with MS whereas smaller turn angle was the most discriminative in people with PD. Both quantity and quality of the mobility discriminated well mobility of people with MS from MS-Ctl, however, only quality of mobility discriminated well people with PD from PD-Ctl, and not quantity.

We found that measures of mobility discriminating the older (PD-Ctl) from the younger (MS-Ctl) groups (see supplementary Figure S1), specifically, slow turns, slow cadence and slow gait speed with accompanying changes in gait temporal characteristics) may be related to slowness due to aging and/or various comorbidities (such as arthritis, heart disease) [46], [47]. The mobility measures distinguishing the MS-Ctl versus PD-Ctl control groups were distinct from the top mobility measures characterizing the mobility impairments in MS and PD. Hence, we used age-matched, control cohorts for the two different neurological disorders.

Activity measures, and specifically median number of valid strides in a bout and number of valid strides per hour were smaller in MS, and were the most discriminative measures characterizing mobility impairment in MS. In fact, lower levels of physical activity, and fewer daily step counts have previously been observed in individuals with MS compared to healthy control subjects (see refs. in Block et al. 2016; Giggins, Clay, and Walsh 2017). Furthermore, lower physical activity levels, specifically, fewer steps per day have been shown to be strongly related to EDSS, MSWS-12,Timed 25-Foot Walk test, and 6- Minute Walk test [49], [50]. Thus, it is not surprising that gait speed during passive monitoring in daily life and temporal measures of gait associated with gait speed (double support and swing time as percent of the gait cycle, cadence, step duration and stride length) were also very discriminative measures characterizing gait impairments in MS.

In contrast to people with MS, turning, and specifically smaller turn angles, best characterized mobility disability in people with PD during unsupervised daily mobility. In fact, turning performance has been observed to be greatly compromised in people with PD leading to a reduced quality of life and injurious falls [12], [51]. Specifically, turning characteristics are even more sensitive to early, unmedicated PD than are characteristics of straight-ahead, linear gait in a clinical Timed Up and Go Test [30]. Turning in individuals with PD is characterized by long turning durations, more steps to complete the turn, and slower peak and average speed compared to age-matched, healthy controls in daily life [8], [10]. Turning, specially turning in place 360 degrees is also a trigger for freezing of gait so it is possible that people with PD avoid large turns to avoid freezing. Indeed, a significantly smaller turn angle in freezers was observed in a study of 94 non-freezer and freezer PD subjects passively observed during 3 days of daily life [19]. Also, a recent longitudinal study in community dwelling old adults found that fewer turns in daily life differentiates future recurrent fallers from non-fallers [51]. After turn angle, the swing (percentage of gait cycle) variability measure was the most discriminative to PD. Specifically, higher swing variability was observed in people with PD compared to PD-Ctl. In fact, higher variability in various gait measures was observed previously in patients with PD compared to healthy control subjects in the laboratory [52] and in free living conditions [10], [11], [22]. Further, variability of mobility measures (such as stride-time variability) have been shown to be related to fall risk in PD [26]. Thus, it is not surprising that turn angle, swing CV during passive monitoring in daily life and measures characterizing freezing or shuffling (such as pitch angle of foot at initial contact) were also very discriminative measures characterizing mobility impairments in PD.

Although previous studies have found that people with PD had fewer steps/day than age-matched controls [24], [53], we did not find a significant difference (p=0.35) in number of strides/hour between PD (82.11 ± 43.75) and PD-Ctl groups (98.25 ± 66.83). Further, our total number of steps per day is far less than what has been reported in the literature. Several possibilities may explain this discrepancy: 1) Our number of steps/day did not include turns while turns were included in previous studies. Considering an average ~ 80–100 turns/hour and ~ 3–5 steps per turn, if we add turning, our total number of steps will increase drastically. 2) Lord et al. [24] considered at least one step to qualify as a stepping bout, whereas we considered at least 3 steps to qualify as a bout. 3) We only considered valid strides in each bout for our analysis to avoid the effects due to potential false positive steps, 4) Algorithm differences defining a valid bout and a valid step could result in very different number of steps (or strides) per day. Hence, for activity measures, we recommend caution while comparing our results with consumer grade activity trackers (such as activPAL™, Fitbit).

Mobility measures may be useful to track progression of neurological disease severity. We used the relationship between mobility measures with patient-reported outcomes and clinician-reported outcomes representing disease severity as a type of face (clinical) validity or meaningfulness of the measures. The median number of strides per bout over the week was correlated with the self-reported fatigue scale, MFIS for people with MS. The turning and gait shuffling mobility measures over the week were correlated with the UPDRS PIGD sub score that represents clinician’s evaluation of balance and gait for PD. However, future studies need to determine the test-retest reliability and sensitivity to disease progression of passive mobility monitoring in daily life compared to prescribed tests of gait and turning in the clinic.

There are several limitations of the current study. First, the turn algorithm used in this study was defined by a threshold on vertical rotational rate of the pelvis (15 degree/s) [8]. This can be challenging when turns are nearly as slow and small as the angular velocities and angles of trunk motion during gait. Too high a threshold to detect onset and offset of turns could result in smaller turn angles rather than longer turn durations with the initial and final rate of change of trunk rotation is very slow. Second, we performed all mobility analysis by taking the mean of each measure for all the strides over a week for each subject and thus gave equal weight to each stride. But in reality, gait speed and other measures vary for gait bouts of different lengths [11] and people with MS might have fewer long gait bouts because of the fatigue. Hence, future work will focus on analyzing the effect of bout length on each mobility measure and how gait bout length affects the discriminatory power of each mobility measure. Thirdly, we ranked mobility measures characterizing mobility impairments in MS and PD based on AUC, but we cannot assume this ranking would be identical across all cohorts. Future work with larger cohorts are needed to investigate if these findings would generalize to other individuals. Lastly, in this paper, we report the most discriminative measures of mobility for a specific group of people with MS and a group with PD, but to apply this knowledge in clinical practice, each individual’s gait characteristics are needed to tailor interventions specific for each individual’s specific mobility disability. Future studies need to develop age-related normative values for passive monitoring of mobility. Future work would also explore how to characterize diurnal fluctuations of mobility throughout the day.

Conclusion

Quantity and quality of the mobility both discriminated well people with MS from MS-Ctl, however, only quality of mobility discriminated well people with PD from PD-Ctl, and not quantity. Several of these mobility measures from a week of passive monitoring in daily life were also related to clinical and patient-reported outcomes of neurological severity, providing face (clinical) validity. The most sensitive objective measures of daily life mobility for clinical trials and clinical practice need to be specific for each neurological disease.

Supplementary Material