Normative data on spontaneous stride velocity, stride length, and walking activity in a non-controlled environment

Duchenne muscular dystrophy (DMD) is a severe, rapidly progressive neuromuscular disorder characterized by muscle weakening. It has an estimated incidence of 1 in 5000 males [1]. DMD is caused by out-of-frame mutations in the dystrophin gene, which leads to an absence or deficiency of the protein dystrophin and the degeneration of muscles fibres [2]. Therefore, loss of ambulation occurs generally around the age of 12.

The 6-min walk distance test (6MWT) has been used as a primary outcome in most published pivotal phase 2 and 3 trials in ambulant patients with DMD [3‐6]. Other outcomes used include the 4-stair climb (4SC) and functional scales such as the North Star Ambulatory Assessments (NSAA) in which participants are rated on their ability to perform standardized motor function tasks [3, 5‐10]. In DMD, the low Standardized Response Mean of these different outcome measures, which illustrates the power of these measures to demonstrate a change over a certain period of time, meant that pivotal trials required over 100 patients per group and trial durations of 18 months and 2 years (NCT02851797 and NCT02500381, respectively). To accelerate clinical development and investigate in parallel several approaches without being limited by the number of patients available, it is crucial to validate more powerful outcome measures. This is the case for DMD and for numerous other rare diseases, within or outside the neuromuscular field.

The last decade has seen an increase in the availability of wearable technology for continuous monitoring of health and wellbeing [11]; for example, wearable devices that can be used to assess ambulation range from crude step counters to sophisticated multisensory systems [12]. Unlike consumer devices, medical devices must demonstrate validated measurement accuracy, sensitivity, and specificity [13]. Wearable devices have the potential to provide a complete view of a patient’s condition over a long time period by band-pass filtering day-to-day variation. Therefore, wearable devices provide a complementary approach with a major advantage over hospital-based assessments, which only provide ‘snapshots’ of a patient’s condition that can be affected by fatigue, illness, or lack of motivation.

In this context, Sysnav, a medium size private company, along with academic partners specifically developed a CE-marked class 1 wearable medical device, called Actimyo®, that records passively, in a precise and sensitive way, upper and lower limb movements in everyday life [14, 15]. From the capture of any single movement, several outcomes may be extracted. In ambulant patients, the identification and quantification of every individual stride allows for the calculation of the distribution of stride length and stride speed as well as for the analyses of different centiles such as the 95th centile stride velocity (SV95C). These variables are measured in a home-based environment over a 180-h period and are reliable and highly sensitive to changes in ambulant patients with DMD [16].

Recently, the EMA qualified SV95C as a valid secondary endpoint in clinical trials on ambulant patients with DMD [17]. Additional data are needed to qualify the measure as a primary endpoint [16]. Wearable technology is or has been used in clinical trials of therapies for spinal muscular atrophy [18, 19], facioscapulohumeral muscular dystrophy (NCT02579239, NCT04004000), limb girdle muscular dystrophy type 2E (NCT02579239), centronuclear myopathy (NCT02057705), and Angelman syndrome (NCT04259281).

To properly interpret the longitudinal evolution of the SV95C over time in patient populations, it is necessary to understand the longitudinal evolution in subjects without muscle conditions within the same age range, in particular between 5 and 18 years old, which is the age range mostly targeted in clinical trials of ambulant DMD patients and during which growth or maturational factors may significantly interfere with measures [20, 21]. Therefore, we conducted a longitudinal study in ambulant healthy subjects between 5 and 85 years of age to evaluate the changes in SV95C, as measured with the Actimyo® device, after a 12-months follow-up.

As a result of this study, we provide normative data for spontaneous stride velocity, stride length and walking distance per hour obtained passively with a wearable device during a one-month recording period in a non-controlled setting and the one-year evolution thereafter. Most previous studies using wearable devices provide us with quantitative parameters [24]. Qualitative gait analysis as stride length and velocity were mostly obtained based on evaluations inside clinical facilities [25, 26]. Recently, several efforts were made to collect, in a home-based setting, qualitative data from subjects with various diseases and control populations [18, 19, 22, 27]. However, to the best of our knowledge, we found no other published data regarding normative longitudinal evaluations of spontaneous stride velocity, stride length, and walking activity in an everyday setting.

As suggested in previous findings [28], acquiring normative data was feasible. The overall compliance with device use was good. More than 90% of subjects wore the device more than the minimum requirement of 50 h, which indicates that wearing devices at the ankle and at the wrist for one month was acceptable for most participants. The main reason given for data collection below the 50-h threshold was ‘technical issues’, and compliance was not influenced by gender or age. In addition, the recording duration did not influence the variables that were studied.

As expected, results of timed tests were growth dependent [22]. We also observed the influence of height (in adults) and height and growth (in children) on stride length. These results support previous findings [29, 30]. Height did not influence SV95C, for which only a very weak inverse correlation was found with height in the adult subgroup only. In children, all correlations between age, height and stride length or speed were much weaker with maximum performance as measured by the 95th centile than with median values. Independence of growth is an advantage in clinical trials that are conducted in children over 1 or 2 years, especially in DMD where growth is altered by steroid use.

In comparison to ambulant DMD patients of the same age, healthy children present a 67.3% higher SV95C [17]. The difference for other variables was not as extreme, and there is some overlap between DMD and control participants (SL50C 31.39%, SL95C 52.41%, SV50C 25.72%, distance walked/hour 63.13%) [16]. SV95C was stable after one year in controls but significantly and constantly decreases in subjects with DMD. In DMD patients, there is a significant correlation between SV95C and all timed tests [17]. We did not find such correlations in controls, which is probably related to the fact that weakness constitutes a common limiting factor on 6MWD and SV95C in patients with DMD, strengthening the correlation in this group.

Gathering normative data is crucial in the validation process of an outcome measure, as it allows identification of potentially confounding variables such as age, height, or gender. For a specific outcome, the normative data can either show a strong ceiling effect, as is the case for most of normative scales such as CHOP Intend, NSAA, and Motor Function Measure [31‐34], or increases with age, as is the case for strength [21, 35] and the 6MWD [20].

The limitations of this study reside in the small number of controls per subgroup (age and gender), principally related to the limited availability of recording devices. Given the number of subjects, we did not study the seasonal influence on the parameters. It is likely that a seasonal analysis would need to take weather conditions into account, as they are more likely to influence walking behaviour than does the time of year, and would require continuous recording over one year, which could be burdensome in controls. Another limitation is that the BMI evolution of participants tends toward a more gradual increase than those of age-matched DMD, mainly after 8 years of age [36]. However, this is not a major issue since there is no correlation between BMI and SV95C in both the adult and children subgroups. During the study, participants wore devices on both wrist and ankle. The analysis of upper limb movement in control patients is complex, since there is an interference with movement produced during the ambulation. The results will therefore be presented elsewhere, with comparison to the upper limb measures in ambulant and non-ambulant DMD patients.

This study provides evidence that normative data regarding stride length and velocity in a home-based environment with wearable sensors is feasible. These normative data will allow us to express a patient’s data as a percentage of predicted value for age or height and to identify those variables that have important confounding factors. There is still a large amount of work to be done to enable analysis of ambulation and upper limb function in various diseases using medical device for real-life mobility measurement. Future research should focus on exploring new fields of neurological disease and the creation of algorithms to study upper limb function, falls, or ataxia.