Effects of elbow immobilization on upper extremity activity

Abstract

Background:

It has been dogma that handedness greatly impacts daily activities.

Interruptions in the ability to utilize the dominant arm due to neuromusculoskeletal injuries could negatively impact performance of activities of daily living. Daily activity can be measured using activity monitors. This study aimed to examine how arm dominance impacts function by immobilizing the arms of healthy individuals.

Methods:

Ten individuals wore four upper extremity activity monitors for three days—one day without immobilization, one day with their dominant arm immobilized, and one day with their non-dominant arm immobilized. Triaxial acceleration data was used to calculate average daily activity and an asymmetry index. Between-condition differences were examined.

Findings:

During dominant arm immobilization, the dominant forearm had significantly less average daily activity when compared to the no immobilization day (p = .0001) and the non-dominant immobilized day (p < .0001). A similar trend was observed at the non-dominant forearm when the non-dominant arm was immobilized. Immobilization of an arm increased asymmetry index and reliance on the non-immobilized arm. Significant differences in asymmetry index were not observed between the two casted conditions.

Interpretation:

When an upper extremity was casted, immobilized forearm and upper arm average daily activity was comparable. Dominance did not have an effect on asymmetry index. Immobilization affected asymmetry index compared to no immobilization. This study has demonstrated that regardless of arm immobilized, there will be a reliance on the contralateral limb about twice as much as the immobilized limb. This could prove problematic if the contralateral limb is restricted or injured, limiting independence.

1. Introduction

It has been dogma that upper extremity dominance greatly impacts daily functional activities. Previous studies have demonstrated dominant and non-dominant arms are each optimized for performance of unique functions (Bagesteiro and Sainburg, 2002; Przybyla et al., 2012; Sainburg, 2005). The dominant arm is optimized for control of limb trajectory, while the non-dominant limb is optimized for limb positioning. The coordination of these complementary systems allows for optimizing neural resources to seamlessly execute activities of daily living, especially bimanual tasks (Przybyla et al., 2012). Interruptions in the ability to utilize the dominant arm due to neuromusculoskeletal injuries or conditions could negatively impact performance of activities of daily living. Many neuromusculoskeletal injuries and conditions affect multiple joints and motions, making it difficult to identify any impact arm dominance has on function in these cases.

Daily functional activity can be measured and recorded using activity monitors. This application of accelerometry utilizes triaxial accelerometers to record motions of a segment in the free-living environment. These methods have been used for both upper (Bailey and Lang, 2013; Hurd et al., 2013; Uswatte et al., 2005) and lower (Fortune et al., 2014; Fortune et al., 2017; Mathie et al., 2003) extremity applications to provide an indication of functional usage. Various quantities can be extracted and calculated from the raw acceleration data recorded by activity monitors. Average daily activity has been indicative of overall function (Uswatte et al., 2005; Uswatte et al., 2006). The use of an asymmetry index has been used to gauge impact of treatment in shoulder arthroscopy (Hurd et al., 2013), stroke (Lang et al., 2007; Noorkõiv et al., 2014), and brachial plexus injury (Webber et al., 2019). This measure allows for reliance on the dominant side to be quantified. Previous studies using this real world data have reported that healthy individuals have symmetrical usage of their upper extremities throughout the day, regardless of dominance (Hurd et al., 2013; Lang et al., 2007; Rand and Eng, 2010).

To explore the effect of upper extremity dominance on daily functional activity in an injured population, this study immobilized the upper extremities of a group of healthy individuals to simulate restricted elbow and wrist function that could be cause by a neuromusculoskeletal injury or condition. It was hypothesized that upper extremity dominance affects average daily activity. The null hypothesis was that no differences in asymmetry index would exist between immobilized and non-immobilized conditions.

2. Methods

2.1. Subjects

Ten subjects (9 male; 1 female) with no prior neuromusculoskeletal conditions of their back, neck, shoulders, or arms participated in this Institutional Review Board (IRB) approved study after providing informed consent. Subjects (mean ± standard deviation age: 28.2 ± 5.3 years) were recruited by word of mouth to reflect the clinical population studied in a previous investigation into upper extremity activity (Webber et al., 2019). They were remunerated for their time upon completion of the study.

2.2. Data acquisition

Subjects wore four tri-axial activity monitors (wGT3X-BT, ActiGraph, Pensacola, FL, USA) on bilateral forearms and upper arms for the three day study. Forearm monitors were positioned like a wristwatch on the subjects, while upper arm monitors were positioned to lie on the mid-belly of the triceps brachii on the lateral aspect of each arm. A long arm cast to prohibit elbow, wrist, and forearm motion was applied on the two immobilization study days (elbow at 90° of flexion, forearm in neutral rotation, wrist at 30° of extension, and digits and thumb free). The subject’s dominant arm was immobilized on one study day. A second study day, the subject’s non-dominant arm was immobilized. During a third study day, the subject did not receive any arm immobilization. The order of these study days was randomized to minimize any learning effects. Each subject traveled to an outpatient orthopedic cast clinic for cast application and removal on the two study days requiring immobilization.

For immobilization days, activity monitors were worn after cast application for a minimum of ten hours. Subjects wore activity monitors during waking hours when there was no immobilization. Subjects went about their daily tasks as normally as they could throughout the three day study. Tri-axial acceleration data was collected at 50 Hz and stored onboard the monitors.

2.3. Data processing

Data were downloaded (ActiLife, ActiGraph, Pensacola, FL, USA) from the monitors. This raw, tri-axial acceleration data was processed with a custom MATLAB code to determine daily activity and asymmetry indices. The data were filtered using a 4th order Butterworth filter (ω_c = 0.1 Hz). Utilizing established methods (Hurd et al., 2013), data from each sensor were divided into minute epochs and activity (m/s²) was calculated (Eq. 1) for each arm segment (forearm and upper arm). This value was averaged across each day for use when calculating asymmetry indices (A_x) for each segment (Eq. 2). Average activity for each segment each day was calculated, along with forearm and upper arm A_x and wear time.

Equation 1 Activity per 60 s epoch (n: number of data points within each epoch; a_x, a_y, a_z: tri-axial acceleration components).

$$\mathit{Activity}\left(m/s^2\right)={\sum }_{i=1}^n\sqrt{a_{x_i}^2+a_{y_i}^2+a_{z_i}^2}$$

Equation 2 Asymmetry Index (A_x) (U: activity of the uninvolved limb; I: activity of involved limb). (Kaufman et al., 1996).

$$\mathit{If}U\ge I,\mathit{then}{A}_x=\frac{U}{I}-1$$

$$\mathit{If}U<I,\mathit{then}{A}_x=1-\frac{I}{U}$$

2.4. Statistical analysis

A Shapiro-Wilk W Test was conducted in JMP (SAS Institute, Cary, NC, USA) to determine normality of average activity for each segment, and A_x. Depending on the normality of each data set, immobilization conditions were compared using either an analysis of variance (normally distributed data) or a Kruskal-Wallis test (non-normally distributed data). For normally distributed data, a Tukey-Kramer post hoc analysis was performed. Post hoc analysis using Dunn’s Method for Joint Ranking between all pairs was conducted for data that was not normally distributed. This method included a Bonferroni correction for multiple comparisons. For all analyses, p ≤ .05 indicated the presence of statistically significant differences.

3. Results

3.1. Wear time

Wear time averaged 10 h for each condition (Table 1). Since the wear time was constrained by study design for the days when subjects were immobilized, between condition differences were not explored.

Table 1.

Wear times by immobilization type for each condition, mean and standard deviation (SD) reported along with median and interquartile range (IQR).

	Wear time (Minutes)
	Mean	SD	Median	IQR
No immobilization	600	30	645	70
Dominant	610	10	610	30
Non-dominant	660	60	610	10

3.2. Average daily activity

Average daily activity (Fig. 1) was normally distributed for all segments (dominant forearm, p = .269; non-dominant forearm, p = .649; dominant upper arm, p = .526; and non-dominant upper arm, p = .155). When the dominant arm was immobilized, the dominant forearm had significantly less average daily activity (234 m/s²) when compared to the no immobilization day (472 m/s²; p = .0001) and the non-dominant immobilized day (540 m/s²; p < .0001). A similar trend was observed at the non-dominant forearm when the non-dominant arm was immobilized. Significant differences in average daily activity of the non-dominant forearm (254 m/s²) were observed when compared with the no immobilization day (466 m/s²; p = .001) and the dominant arm immobilized day (479 m/s²; p = .001).

Fig. 1.

Average daily activity (m/s²) presented as median, first, and third quartiles for a) dominant forearm monitor, b) non-dominant forearm monitor, c) dominant upper arm monitor, and d) non-dominant upper arm monitor. Immobilization of the dominant elbow and wrist decreased dominant forearm activity, while non-dominant forearm activity was slightly elevated compared to the no immobilization case, albeit not significantly different. A similar trend was observed for the upper arms. When the non-dominant arm was immobilized, similar trends for forearm and upper arm activity existed.

The magnitude of upper arm average daily activity was approximately 2/3 than that of forearm activity overall (across conditions and both arms, forearm average 400 m/s² and upper arm average 270 m/s²). When the dominant arm was immobilized, subjects utilized their dominant upper arm (222 m/s²) significantly less than when their non-dominant arm was immobilized (319 m/s²; p = .038).

Mean forearm and upper arm average daily activity was the same when the monitors were on the immobilized arm. This trend was evident for dominant arm monitors when the dominant arm was in a cast (forearm 234 m/s²; upper arm 222 m/s²; p = .762), as well as non-dominant arm monitors when the non-dominant arm was immobilized (forearm 254 m/s²; upper arm 244 m/s²; p = .786).

3.3. Asymmetry index

Forearm A_x data (Fig. 2) were normally distributed (p = .133), thus were analyzed using an analysis of variance, followed by Tukey-Kramer post hoc analysis. Upper arm A_x data were not normally distributed (p = .043), prompting use of the nonparametric Kruskal-Wallis test and post hoc analysis using Dunn’s Method for Joint Ranking between all pairs. Overall, A_x was greater when a subject’s arm was immobilized. Significant differences in A_x were not observed between the two casted conditions (Table 2).

Fig. 2.

Average asymmetry index by subject (light grey) overlaid with condition median, first, and third quartiles (black). Immobilization of an upper extremity resulted in increased reliance on the non-immobilized arm, increasing A_x. Regardless of arm immobilized, A_x was affected similarly.

Table 2.

p-values for between condition differences in asymmetry index. Significant differences (bolded) existed between the no-immobilization condition and both immobilized conditions.

	Forearm	Upper arm
No-immobilization/non-dominant	< 0.0001	0.001
No-immobilization/dominant	< 0.0001	0.004
Non-dominant/ dominant	0.988	1.000

4. Discussion

During this study, upper extremity activity of healthy subjects was monitored in three different conditions—no immobilization, dominant arm immobilized, and non-dominant arm immobilized. The immobilized conditions simulated restricted elbow and wrist motion that could be caused by any neuromusculoskeletal injury. Casted healthy subjects were used to make the present study more broadly applicable, eliminating potential effects that concomitant injuries in a clinical population could have on daily activity. This application of accelerometry demonstrates a straightforward use of raw tri-axial acceleration data derived quantities, such as average daily activity and A_x. Subsequent analysis of the data could provide insight into specific activities conducted throughout the day, similar to the use of accelerometry of the lower extremities (Fortune et al., 2014).

Wear time, average daily activity for each segment, forearm A_x, and upper arm A_x were calculated for each condition. Significant differences existed between no immobilization and immobilized conditions for average daily activity, forearm A_x, and upper arm A_x. Study design constrained the time when subjects were immobilized, so wear time between conditions was not compared.

4.1. Average daily activity

Immobilization of an upper extremity with a long arm cast resulted in subjects relying on the contralateral arm, with increased average daily activity for free segments. Forearm activity is a result of both motions originating at the shoulder and at the elbow. As such, average daily activity for the upper arm was generally less than that observed at the forearm. When a cast was applied, forearm and upper arm average daily activity were essentially the same since the monitors were on the same segment with motion only originating at the shoulder. The linear accelerations used to calculate average daily activity were the same because of the small translational motion generated at the shoulder moving the arm as one rigid body. Comparing daily activity using the A_x can mitigate this effect, resulting in a description of solely forearm or upper arm usage. When compared with published healthy control (non-immobilized) values for average daily activity (forearm 1025 m/s²; upper arm 607 m/s²), the values in this study (forearm 490 m/s²; upper arm 294 m/s²) were approximately half (Hurd et al., 2013). The ratio between upper arm and forearm average daily activity remained the same between the two studies. Differences in the magnitude of the average daily activity could be due to differing subject groups, with occupation or activities of daily living differing significantly between the two studies. To compare these two groups further would require additional subject demographics not readily available.

4.2. Asymmetry index

Assessing asymmetry of upper extremity usage can provide an insight into patient function in their free living environment. Without the constraints of the lab, individuals are free to go about their daily activities as they so choose. This provides a quantitative description of function to supplement patient reported outcome measures in a clinical setting.

Immobilization with a long arm cast increased the asymmetry of both forearm and upper arm usage, regardless of the arm casted. Subjects relied on their non-casted forearm to complete activities throughout their day twice as much as their casted forearm. Upper arm asymmetry due to immobilization was also present. It is apparent that a dysfunctional elbow substantially impacts symmetrical arm usage. The functional shoulders of subjects allowed for some usage of the immobilized limb, resulting in decreased A_x compared to values reported previously for a clinical population with brachial plexus injuries (Webber et al., 2019). When subjects were not casted, their A_x was almost 0, indicating almost perfectly symmetrical usage, as published previously (Hurd et al., 2013).

4.3. Clinical implications

Hand dominance is not as impactful a factor in function as measured by A_x. Furthermore, this study has shown that dominance can be compensated for in healthy individuals who are immobilized. Anecdotally, this has been observed in clinical scenarios in individuals with upper extremity amputation and brachial plexus injuries. This study validates the quick adaptation. Disruption of arm dominance can have substantial impact on clinical outcomes. This study has demonstrated that regardless of arm immobilized, there will be a reliance on the contralateral limb about twice as much as the immobilized limb. Clinicians should take note of this, especially when treating patients whose contralateral limb is restricted or injured, as this could substantially limit the patient’s ability to function independently.

4.4. Study limitations

It is important to interpret the results presented in this study keeping in mind its limitations. Subject compliance with wearing the activity monitors in the proper orientations could skew some of the data. Additionally, casting of a healthy individual does not completely represent a specific upper extremity condition with limited motion. Patients typically present with various other limitations due to their injury. This would further decrease their average daily activity compared to the subjects in this study, as well as increasing A_x.

Nevertheless, the information presented in this study can aid in contextualizing accelerometry studies in clinical populations, like those in subjects with shoulder arthroplasty (Hurd et al., 2013), stroke (Lang et al., 2007; Noorkõiv et al., 2014), and BPI (Webber et al., 2019). Immobilization of the elbow in an individual with an otherwise healthy shoulder and hand forces the individual to rely more heavily on their unimpaired arm, regardless of handedness. They are, however, able to appropriately position their upper extremity with motion originating at the shoulder to utilize their hand function.

Moreover, this study presents quantitative evidence indicating that average daily forearm activity is a result of motion originating at both the shoulder and the elbow. Because of the relatively small translations associated with upper extremity movement, there is not an impact on the linear accelerations experienced by a segment, regardless of where the activity monitor is placed.

5. Conclusion

Activity of the upper extremity during periods of restricted motion was collected and analyzed. There was no effect of dominance on A_x; however, there was an effect of immobilization on A_x when compared to a no immobilization condition, as expected. Regardless of the arm immobilized, an individual will rely on the contralateral side about twice as much as the immobilized arm. This must be taken into consideration clinically, as independent function after trauma will be dependent upon the ability to use this contralateral side. The information presented in this study allows for a comparison point for future studies utilizing accelerometry, especially those investigations into populations with neuromusculoskeletal injuries or conditions.