A pilot study using actigraphy to examine activity performance and its relationship with activity capacity in adults with facioscapulohumeral muscular dystrophy

Abstract

Background

Weakness associated with facioscapulohumeral muscular dystrophy (FSHD) impacts daily activities. Impact is often measured using standardized clinical assessments, documenting activity capacity, and only captures a snapshot of function. Wearable sensors may enable assessments of real-world activity performance.

Objective

To examine activity performance (actigraphy) and its relationship with capacity (clinical measures of strength and function) in adults with FSHD.

Methods

Remote assessments were piloted in a subgroup from Motor Outcomes to Validate Evaluations in FSHD. Participants wore waist-worn activity monitors for 7 days. Activity metrics included moderate-to-vigorous intensity physical activity (MVPA), time in activity levels, and step counts. Descriptive statistics summarized activity, Mann Whitney U tests compared groups, and correlation analyses examined relationships between activity performance and capacity.

Results

Thirteen subjects wore the device for a median of 7 days. Most time was sedentary, with minimal vigorous activity. Participants spent a median of 23.1 min daily (IQR: 23.8) and 161.7 min weekly (IQR: 157.9) in MVPA. Median daily step count was 4245 (IQR: 2892), with a median maximum step count of 20 (IQR: 7) within a 10-second epoch. Correlations were found between total MVPA and 10mw/r (ρ=−0.720, p = 0.006) and TUG-comfortable (ρ=−0.720, p = 0.006), and between maximum step count and several functional measures.

Conclusion

This pilot study provides insights into activity performance and its relationship with capacity in a small cohort of adults with FSHD. Total MVPA and maximum step count appear most informative for evaluating activity performance; larger studies are needed to confirm findings and assess psychometric properties of these metrics.

Introduction

Facioscapulohumeral muscular dystrophy (FSHD) is the third most common form of muscular dystrophy, with a prevalence of approximately 1 in 15,000 to 1 in 20,000.^1,2 FSHD is characterized by progressive, asymmetrical muscle weakness and atrophy. Muscles of the face, shoulder girdle, and upper arm are often initially affected.^3,4 Other commonly involved areas include muscles of the legs, pelvic girdle, and core. Weakness in the legs often leads to reported difficulties with mobility, walking, and participation in daily activities. ²

Currently, our understanding of the impact of FSHD on daily activities is primarily gathered through in-person clinic visits. During which clinical outcome assessments (COAs) ⁵ are used to assess activity capacity – what an individual is capable of doing in a structured environment at a given time.^6,7 These assessments typically rely on standardized tests and only provide a snapshot of function. In contrast, activity performance refers to what an individual actually does in their real-world environment, offering a more accurate reflection of their lived experience. ⁶ Measuring activity performance in the real-world using wearable sensors may provide additional information on the impact of FSHD.

Advances in technology have led to the development of wearable sensors, which use tri-axial accelerometers to produce quantitative metrics of activity. Their small size and long battery life allow for continuous data collection in real-world settings over multiple days. These sensors generate metrics, such as time spent in moderate-to-vigorous intensity physical activity (MVPA) and step counts. These metrics may provide a more comprehensive view of an individual's activity performance. For instance, wearable sensor data can be used to assess whether individuals meet physical activity guidelines. These guidelines recommend at least 150 min of moderate-to-vigorous intensity aerobic activity per week for adults, including those with chronic disease or disabilities. ⁸ Additionally, step count data for special populations, including those with neuromuscular diseases, ranges from 1200 to 8800 steps per day, with a midpoint of 5000 steps per day. ⁹

Evidence in other populations, such as amyotrophic lateral sclerosis, multiple sclerosis, Duchenne muscular dystrophy, and pulmonary arterial hypertension, has shown that actigraphy-derived metrics are valid,^10–13 distinguish between individuals with disease and healthy controls,^14–16 and are sensitive to disease progression and treatment effects.^10,15,17 Furthermore, a study in children with congenital muscular dystrophies demonstrated the feasibility of using accelerometry to assess real-world physical activity. ¹⁸ A recent review of Duchenne muscular dystrophy found that all included studies reported meaningful relationships between clinical outcomes and wearable sensor-derived metrics. ¹⁹ In FSHD, actigraphy has been used as a secondary outcome measure in interventional studies,^20–22 where post-treatment improvements in mean registered physical activity were observed in both aerobic exercise and cognitive-behavioral groups. ²⁰ Furthermore, a recent study illustrated the acceptability of smartphones and wearables to patients with FSHD. ²³ These previous findings suggest that data derived from wearable sensors may provide useful information on the impact of FSHD. However, there is limited knowledge of day-to-day physical activity, representing activity performance, in individuals with FSHD. Additionally, there is little known about how physical activity metrics relate to measures of strength and function in FSHD. Therefore, the purpose of this pilot study was to better understand activity performance in individuals with FSHD using actigraphy and to explore relationships between performance measures and activity capacity, as measured by COAs.

Materials and methods

This remote assessment pilot study was conducted at the University of Kansas Medical Center (KUMC) and University of Rochester Medical Center (URMC). It included a sub-group of participants enrolled in the Motor Outcomes to Validate Evaluations in FSHD (MOVE: NCT04635891) study. Participants were recruited from clinic visits and enrolled sequentially. Institutional Review Board approval was obtained, and all participants provided informed consent for this sub-study. MOVE is a multi-site study designed to accelerate therapeutic development and improve care by establishing a standardized, clinic-based protocol. The remote assessments aimed to capture functional tasks typically performed during routine in-person visits, including reaching and lifting objects, midarm and hand tasks, and gait and function measurements. Actigraphy was used to monitor physical activity in the participants’ natural environment, serving as a measure of activity performance, which is the primary focus of this paper.

Participants

Participants included in the remote assessment pilot study were adults (age 18–75 years) with genetically confirmed FSHD (type 1 or 2) or were clinically diagnosed with FSHD with characteristic findings on exam and an affected first degree relative; enrolled in MOVE; with internet access for remote pilot assessment. Participants were excluded if they were unwilling or unable to provide informed consent, or had any other medical condition which, in the opinion of the investigator, would interfere with study participation. Participation was voluntary, and participants were allowed to withdraw from the study at any time.

MOVE outcome measures

In-person visits were conducted at KUMC and URMC. Measures included disease severity classification using the FSHD clinical score, ²⁴ manual muscle testing (MMT), and functional assessments. MMT consisted of 16 bilateral muscle groups: shoulder abductors and external rotators, elbow flexors/extensors, wrist flexors/extensors, common finger extensors, thumb flexors, hip flexors/extensors, hip abductors/adductors, knee flexors/extensors, ankle dorsiflexors/plantar flexors plus neck flexors/extensors. Muscles were graded using the modified 10-point Medical Research Council (MRC) scale. ²⁵ Average MMT scores were calculated using all muscle groups, and average lower extremity (LE) scores included the hip, knee, and ankle muscles. Functional assessments consisted of mobility items from the FSHD-COM. ⁵ These included: timed supine to sit, timed-up and-go (TUG); performed twice: (1) TUG comfortable – participants self-selected, everyday walking pace and (2) TUG maximum – participants self-selected, fast walking pace, 10-meter walk/run test (10mw/r), timed ascend and descend of 4 stairs, and timed sit to stand.

ActiGraph

The wGT3X-BT is ActiGraph's original wearable device for real-world monitoring of physical activity. ²⁶ The device has a 25-day battery life, with a sampling rate of 30 Hz. ²⁶ Raw data from accelerometers can be processed using algorithms that produce activity counts. Algorithm-based cut points are used to categorize data into intensity buckets, which classifies an individual's energy expenditure. ²⁶ Cut points were defined as sedentary: 0–99 counts/minute (cpm), light: 100–1951 cpm, lifestyle: 760–5724 cpm (manually added), moderate: 1952–5724 cpm, vigorous: 5725–9489 cpm, and very vigorous: 9499-∞ cpm.^26,27 These cut points were developed to correspond to different intensities of activity using commonly employed Metabolic Equivalent of Task (MET) categories. ²⁷ MET range categories were use to define light as less than 3 METs, moderate as 3–5.99 METs, vigorous as 6–8.99 METs, and very vigorous as greater than 8.99 METs. ²⁷ Examples of activities that fall into these categories include light household chores or walking at a slow pace, brisk walking or walking upstairs, cycling or lap swimming, and running or weight training, respectively. ²⁸ The lifestyle cut point was created to overlap with the light and moderate cut points, ²⁶ and encompasses both light and moderate level activities.

Participants received a toolkit via mail containing equipment to conduct the remote assessments including an ActiGraph device after completion of their in-person MOVE visit. At-home outcome measures were conducted, after which participants were provided with instructions on ActiGraph device placement and wear time requirements. Specifically, subjects were instructed to wear the device over their non-dominant hip continuously for 7-days. A 7-day monitoring period was selected because of its consistent use in physical activity studies. ²⁹ These monitoring periods are acceptable by participants, provide an adequately large number of days to reach an inter-class correlation of 80% or more in most populations, and allows for sampling of behavior during week and weekend days. ²⁹

Day 1 was defined as the first day the participant began wearing the device, and Day 7 was defined as the final complete day of device use. Upon return of the device, raw data were downloaded at 10- and 60-second epochs. 10-second epochs were used to analyze activity, as they appear to be more well-suited to capture short bursts of activity.^30,31 The 60-second epochs were used to measure time spent sleeping. The raw .agd files were uploaded into ActiLife for analysis; proprietary algorithms were used to process data into activity counts and intensities.^27,32,33

Derived metrics included total MVPA (minutes spent between 1952–9489 cpm), step count (sum of steps), maximum step count (greatest number of steps in a 10-second epoch). Additionally, time spent in various activity levels (sedentary, light, lifestyle, moderate, vigorous, and very vigorous) as well as total sleep time (total number of minutes scored as asleep) were derived. Data were available by hour, day, and as a summary of total wear time.

Statistical analysis

Wear time requirements were set at 600 min (10 h) per day, as previously reported. ³⁴ Day 1 was excluded for all participants since it only captured a portion of the day's activity; initialized after a mid-day virtual visit. After removing Day 1 and any other days with less than 600 min of wear time, participants’ data were included if they met the wear time criteria for at least 3 days. A 3-day minimum was selected as it has been shown to reliably predict physical activity in adults. ³⁵ Average values of wear time per day and across the total wear period were then calculated, with sleep time removed from sedentary time.

Data were analyzed using SPSS version 29.0. Continuous variables are reported as medians with interquartile ranges (IQR). Mann-Whitney U tests were used to compare characteristics between different groups using defined thresholds of activity (150 min of MVPA per week) and step counts (5000 average steps per day), and the median maximum step count of the sample. Spearman's correlation coefficient analyses were used to assess the relationship between actigraphy-derived metrics and clinical measures of strength and function. The significance level for all analyses was set at p < 0.05, two-tailed.

Results

Demographics

Twenty-three subjects were enrolled in the remote assessment pilot sub-study. Of these, three withdrew, and two were non-ambulatory, making them not appropriate for the ActiGraph component. However, those two were able to complete other at-home outcome measures. This left a total of eighteen participants, thirteen of whom (7 males, 53.8%) met the wear time requirements for inclusion in the analysis. We have partial insights into why five participants did not meet the wear requirements. Two participants reported not wearing the device, while three simply had insufficient wear time. Although no post-study interviews were conducted to gather further details, one of the five participants mentioned during the physical exam that they found the device uncomfortable. This information was provided without any further details recorded.

Among the included participants, the median age was 65 years (IQR: 27.5), and the median FSHD Clinical score, measured on a 10-point scale, was 5 (IQR: 5.5). Median clinical measures of strength and function, representing activity capacity, are detailed in Table 1. The median time between in-person clinical visits and the start of ActiGraph data collection was 17 days (IQR: 13).

Table 1.

Demographics.

Characteristics	Median (Interquartile Range) (n = 13; *n = 10; **n = 11)
Average Strength	4.46 (0.49)
Average Lower Extremity Strength	4.39 (0.69)
FSHD-Composite Outcome Measure Score	*15.50 (13.75)
Timed Up and Go-Comfortable (seconds)	8.95 (2.74)
Timed Up and Go-Maximum (seconds)	**6.89 (2.92)
10-meter Walk/Run Test (seconds)	6.01 (4.33)
Ascend Stairs (seconds)	**2.75 (1.26)
Descend Stairs (seconds)	**2.39 (0.74)
Timed Sit to Stand (seconds)	1.37 (0.76)

Activity levels and step counts

Participants wore the ActiGraph for a median of 7 days (range: 5–7 days). The most time was spent in sedentary activity, followed by light, lifestyle, moderate, vigorous, and very vigorous activity levels. Details of the time spent in various activity levels as well as total sleep duration, are presented in Table 2.

Table 2.

Time spent in various activity levels.

Activity Levels (minutes)	Median (Interquartile Range) (n = 13)
Sedentary	654.3 (591.0)
Light	164.5 (82.7)
Lifestyle	71.4 (21.4)
Moderate	22.9 (23.6)
Vigorous	0.2 (0.3)
Very Vigorous	0.0 (0.0)
Moderate-to-Vigorous per day	23.1 (23.8)
Moderate-to-Vigorous per week	161.7 (157.9)
Total Sleep Time	444.0 (128.1)

Participants spent a median of 23.1 min per day (IQR: 23.8) and 161.7 min per week (IQR: 157.9) in MVPA. In total, 7 participants (53.8%) met or exceeded the recommended 150 min of MVPA per week. Among these participants, times on the TUG-comfortable (U = 5.0, z = −2.286, p = 0.022), 10mw/r (U = 4.0, z = −2.429, p = 0.014), and descend 4 stairs (U = 3.0, z = −2.191, p = 0.030) were statistically significantly faster compared to those who did not meet 150 min of MVPA per week.

The median daily step counts were 4245 steps (IQR: 2892), with a median maximum step count of 20 steps (IQR: 7) within a 10-second epoch. A total of 5 participants (38.5%) averaged 5000 steps per day or more. However, clinical measures of strength and function did not significantly differ between those who averaged 5000 steps per day or more and those who did not. Using the median maximum step count as a threshold, 7 participants (53.8%) achieved at least 20 steps within a 10-second epoch. These participants demonstrated significantly lower scores on the abbreviated FSHD-Composite Outcome Measure (U = 1.5, z = −2.305, p = 0.016), and significantly faster times on the timed supine to sit (U = 4.0, z = −2.429, p = 0.014) and TUG-maximum (U = 3.0, z = −2.191, p = 0.030) compared to those who did not meet the 20-step threshold.

Relationships between activity performance and activity capacity

Moderate to strong correlations were found between time spent in various activity levels with measures of strength and function. Only time spent in very vigorous activity level showed a significant correlation with strength (ρ=0.629, p = 0.021, 95% CI = 0.101, 0.880). Time spent in lifestyle correlated with the most functional measures, including TUG-comfortable (ρ=−0.604, p = 0.029, 95% CI = -0.871, −0.062), and TUG-maximum (ρ=−0.709, p = 0.015, 95% CI = -0.921, −0.170). The strongest correlations were observed between total MVPA and both the 10mw/r (ρ=−0.720, p = 0.006, 95% CI = -0.913, −0.263) and TUG-comfortable (ρ=−0.720, p = 0.006, 95% CI = -0.913, −0.263), which were numerically identical for moderate activity level.

Maximum step count also showed moderate to strong correlations with various measures of function. Maximum step count was associated with several functional tasks, including TUG-comfortable (ρ=−0.610, p = 0.027, 95% CI = -0.873, −0.071), TUG-maximum (ρ=−0.636, p = 0.035, 95% CI = -0.899, −0.038), 10mw/r (ρ=−0.555, p = 0.049, 95% CI = -0.852, −0.013), and timed sit to stand (ρ=−0.666, p = 0.013, 95% CI = -0.894, −0.164).

Discussion

In this pilot study, we used wearable sensors to examine activity performance in a cohort of adults with FSHD. Most participants spent a large portion of their time in sedentary activities, with minimal to no time in vigorous activity, highlighting the potential impact of FSHD on daily activity. Previous research indicates that individuals with FSHD display reduced cadence and gait speed compared to healthy controls, ³⁶ which may align with our findings of limited time spent in vigorous activity in this small cohort. Median daily step counts were also comparable to those reported in a study of 30 individuals with slowly progressive neuromuscular diseases. ³⁷ Additionally, over half of the participants achieved at least 150 min of MVPA per week and recorded maximum step counts of at least 20 steps within 10-second epochs. These thresholds were able to distinguish between participants, with significant differences observed in activity capacity measures, highlighting their potential utility, as previously described.^20–22 Moreover, the maximum step count within 10-second epochs may be less influenced by behavior and external factors and more closely reflect a person's ability. ¹⁷

We found moderate to strong correlations between actigraphy-derived metrics and clinical measures of strength and function. Specifically, step count, maximum step count, total MVPA, and time spent in lifestyle, moderate, and very vigorous activity levels were associated with various functional metrics that assess activity capacity. Notably, maximum step count and MVPA showed strong correlations with TUG-comfortable and timed sit-to-stand, both of which involve transitioning from sitting to standing. This suggests that difficulty with this task may negatively impact overall activity performance. In addition, maximum step count correlated with the greatest number of functional metrics, indicating that it may provide a more comprehensive view of activity performance compared to other actigraphy-derived metrics. The preliminary results suggest that MVPA and maximum step count may serve as useful measures of physical activity in real-world settings.

Limitations and mitigation strategies for future studies

This study has some limitations. First, adherence was limited, as five out of eighteen participants did not meet the minimum wear-time criteria and were excluded from the analysis. While the study did not examine reasons for non-adherence, it may reflect inherent challenges related to remote data collection. Future studies may benefit from implementing additional supports, such as reminders or daily logs, ¹⁸ to improve compliance. Optimizing adherence is critical for ensuring high-quality data in wearable sensor studies. Second, the small sample size is another limitation, increasing the potential influence of outliers on the results. For example, one participant demonstrated markedly high activity levels, which may have driven some of the observed correlations. To reduce the impact of such outliers, non-parametric analyses were employed. Third, activity monitoring may have been influenced by behavioral changes in response to the awareness of being observed (the Hawthorne effect ³⁸ ), which has been noted in other studies.^10,17 This may have led to elevated activity levels during the monitoring period. Future studies may consider longer monitoring durations to mitigate this effect. Fourth, the use of proprietary algorithms, developed in healthy populations, may also limit the accuracy of the findings. These algorithms may not fully account for the altered gait parameters ³⁶ and reduced activity tolerance ⁴ seen in individuals with FSHD. Future research may pursue the development of disease-specific algorithms and cut points. Lastly, given the exploratory nature of this pilot study, the large number of correlation tests conducted may have increased the risk of experiment-wise error.

Future research directions

There is a growing shift towards remote monitoring in clinical research. Our preliminary findings suggest that actigraphy-derived metrics may offer an alternative for capturing physical activity data. However, larger studies are needed to confirm these findings and to establish the reliability, validity, and responsiveness of these wearable sensor-derived metrics. Remote monitoring offers several advantages, including reduced participant burden by minimizing travel and eliminating the need to perform at a single time point. Various future research directions could advance this shift. For example, a larger, longitudinal sub-study stemming from the MOVE study could expand the sample and allow for further examinations of construct validity and responsiveness. Additionally, wearable sensors could be incorporated into clinical trials to monitor physical activity levels, offering a more comprehensive view of activity patterns in this population. This may be particularly valuable to sponsors when participants are instructed to maintain consistent activity levels, and, depending on their responsiveness, may offer the ability to detect changes in real-world physical activity. Finally, integrating these devices into physical therapy practice may enhance intervention development by enabling real-world physical activity monitoring. Examining correlations between wearable sensor-derived metrics and quality of life could further support the design of targeted interventions to potentially improve the overall management of individuals with FSHD.

Conclusion

This pilot study examines activity performance and explores the relationship between activity performance and capacity in a small cohort of adults with FSHD. Our findings suggest that wearable sensors may play a valuable role in assessing physical activity in this population. At the same time, capacity-based COAs during in-person visits remain essential. Used together, these complementary approaches offer a more complete picture of function by capturing both peak ability in a controlled setting and activity performance in a person's natural environment. However, in looking ahead, with the growing interest in digital health outcomes, actigraphy-derived metrics may present an opportunity to enhance the evaluation and management of individuals with FSHD.