Abstract
Background
The purpose of this study was to use at-home, portable, continuous monitoring technologies to record arm motion and activity preoperatively and postoperatively after reverse total shoulder arthroplasty (RTSA).
Methods
Thirty-three patients indicated for RTSA were monitored preoperatively and 3 and 12 months postoperatively. Inertial measurement units were placed on the sternum and upper arm of the operative limb, recording humeral motion relative to the torso for the duration of a waking day. Elevation events per hour (EE/h) > 90°, time spent at >90°, and activity intensity were calculated and compared between time points. Patient-reported outcome measures were also collected at all time points.
Results
At 3 (P = .040) and 12 (P = .010) months after RTSA, patients demonstrated a significantly greater number of EE/h > 90° compared with preoperatively. There were no significant differences (P ≥ .242) in the amount of time spent at different elevation angles at any time point or in arm activity intensity. Overall, 95% of the day was spent at elevation angles < 60°, and 90% of the day was spent in a low- or moderate-intensity state. Pearson correlations demonstrated relationships between forward elevation and the number of EE/h (r = 0.395, P = .001) and the number of EE/h > 90° (r = 0.493, P < .001).
Conclusion
After RTSA, patients significantly increase the frequency of arm elevation to higher angles. However, we found no differences in the amount of time spent at different elevation angles. Overall, after RTSA, >95% of the day was spent at elevation angles < 60° and <1% of the day was spent at >90° of elevation.
Keywords: Reverse shoulder arthroplasty, inertial measurement unit, wearable sensors, motion tracking, upper extremity, clinical outcomes, shoulder arthritis, cuff tear arthropathy
Reverse total shoulder arthroplasty (RTSA) is a promising surgical solution for a growing number of pathologies and patient populations, effectively restoring shoulder function and reducing pain.19 , 30 , 32 Postoperative active range of motion is frequently used by clinicians in evaluating the success of the procedure, although it may not describe the functional range of motion used by patients during activities of daily living. To this effect, the past decade has seen a number of research studies adopting portable at-home technologies for activity monitoring. By eliminating the controlled clinical environment, patient activity patterns may be more accurately represented.8 , 20 Inertial measurement units (IMUs), historically used in aircraft navigation applications, have been proposed as a method for accurate, continuous human motion tracking. Composed of triaxial accelerometers, gyroscopes, and magnetometers, IMUs can determine the orientation of a rigid body to within a few degrees.4 , 11 , 21 Unlike optical or electromagnetic tracking, IMUs are cost-effective, portable, noninvasive, and commercially available; do not require an unobstructed line of sight; and are comparably accurate.2 , 27
IMUs have been used extensively to evaluate the lower limb, with a growing interest in their use to evaluate the upper limb.24 , 28 , 34 A handful of studies have applied IMUs to the shoulders of both healthy controls and patient populations, investigating the feasibility of IMU application to the upper limb.5, 6, 7 , 10 , 12 , 14 , 17 , 18 , 29 , 35 Unlike anatomic total shoulder arthroplasty, for which osteoarthritis is a primary indication, the primary indication for RTSA is cuff tear arthropathy, symptomatically presenting as the inability to actively elevate the arm, in addition to pain.15 , 26 For this reason, restoring active elevation to allow patients to independently complete activities of daily living is a primary goal of RTSA.9 , 25 To our knowledge, no study has prospectively examined continuous at-home arm motion in an RTSA population preoperatively until the healed phase postoperatively. The purpose of this study, therefore, was to assess operative arm motion and activity preoperatively, during the early healing phase 3 months postoperatively, and at the healed phase 1 year postoperatively in an RTSA patient population. We also sought to determine any relationships between arm activity metrics and clinical outcomes.
Materials and methods
Patient selection and instrumentation
This was a prospective case series investigating shoulder elevation prior to and after RTSA as a secondary outcome measure to a randomized trial investigating glenosphere migration between structural bone graft and porous metal augmentation lateralization techniques. A priori power analysis for the randomized trial defined a sample size of 40 participants. Of these, 33 participants (19 men and 14 women; mean age, 71.2 ± 8.2 years) consented to shoulder activity monitoring. The procedures were not consecutive, as primary recruitment was for the randomized trial, and were performed from June 2017 through June 2019. The inclusion criteria were any patient who was scheduled for a primary RTSA and had the capacity to provide written, informed consent. The exclusion criteria were as follows: patients unwilling to wear the motion-tracking garment; patients requiring revision arthroplasty; patients with insufficient bone stock for either a standard-length cemented or press-fit humeral stem, avascular necrosis, proximal humeral fracture, or cognitive or neurological decline; women who were pregnant or planning to become pregnant, and patients unable to read and/or write English.
All surgical procedures were performed by a single, fellowship-trained orthopedic surgeon (G.S.A.), and all patients received the Aequalis Ascend Flex reverse shoulder system (Wright Medical-Tornier Group, Memphis, TN, USA) with a lateralized glenosphere. Postoperatively, the shoulder was immobilized in a sling for 6 weeks. During this time, passive range-of-motion exercises were performed. Following sling removal, active range-of-motion exercises were initiated. Rehabilitation exercises for strength began 3 months postoperatively, concluding at the patient's discretion.
Approximately 1 month prior to surgery, as well as at the 3-month and 1-year clinical follow-up visits, patients were instrumented with 2 IMUs (35 × 60 × 15 mm) to measure shoulder elevation throughout the day. One IMU was positioned at the sternum to provide a reference for the torso's orientation in space; the other IMU, along the lateral shaft of the humerus of the operative shoulder, facing outward. The IMUs were fixed in a snug pocket sewn into a tight-fitting compression shirt (Nike, Beaverton, OR, USA) to inhibit IMU motion relative to the skin while also providing a comfortable solution for long-term wear by the patient. Following instrumentation, patients were encouraged to perform their regular activities of daily living, with the exception of activities that would submerge the IMUs in water (eg, bathing or swimming).
Data acquisition and analysis
The triaxial IMUs (3-Space Data Logger; Yost Labs, Portsmouth, OH, USA) used in this study have a manufacturer-specified orientation accuracy of ±1° for dynamic conditions along all orientations, with a range of 360° about all axes, and have previously been used in a study investigating shoulder motion.18 Real-time Kalman filtering was used to determine orientation. An onboard micro secure digital card enabled real-time data logging at a frequency of 10 Hz. An external power supply was connected to each IMU to increase data-logging capacity for the duration of a day of wear. This external power supply was approximately 40 × 15 × 80 mm and fitted into a separate snug pocket sewn into the compression shirt so as not to inhibit shoulder range of motion. Prior to patient use, IMUs were charged and time synchronized via wired connection to a computer. After the patient donned the compression shirt, a research associate ensured the IMUs were properly positioned. The patient was asked to assume a normal standing position with the arm of interest as close to his or her side as possible. The research associate then began simultaneous data logging from both the torso and humeral reference IMUs. At the end of the day, the patient was asked to remove the instrumented shirt and return it to the research institute via prepaid expedited parcel. The patient was also asked to complete a daily activity log, record when the garment was removed, and include this in the envelope.
On receipt, data from each IMU were uploaded to a personal computer for processing. Output logs from each IMU consisted of time-stamped quaternions: 4-element vectors comprising orientation information in 3 dimensions. Quaternion representations are more robust than traditionally used Euler angles as quaternions do not suffer from singularities such as gimbal lock when extracting anatomic joint rotations (in the case of 2 rotation axes aligning in a parallel configuration, creating an indeterminate system) and are more memory efficient than logging raw rotation matrices.2 In MATLAB (The MathWorks, Natick, MA, USA), quaternions from both the torso and humeral IMUs with corresponding time stamps were matched and a rotational difference between sensor pairs was computed with respect to the torso. With the knowledge that the first quaternions were recorded when the patient was in a standing position with the arm at 0° of abduction and neutral alignment, subsequent arm positions can be determined by taking the relative difference in the orientation of the humeral IMU to that of the torso IMU at matched times. Given a quaternion and rotation sequence, anatomic joint angles can then be determined and the angle of interest isolated—in this case, the angle of shoulder elevation.
Following processing, metrics isolated from the data included the number of elevation events per hour; the proportion of time spent within different elevation ranges (0°-20°, 20°-40°, 40°-60°, 60°-80°, 80°-100°, or >100°); the proportion of elevation events that occurred within these elevation ranges; and the intensity of arm activity, rated as low, moderate, or high intensity (defined later). Elevation events were measured as discrete peaks in the time series with a minimum peak height of 20° and minimum peak width of 1 second. Thus, subsequent peaks must be separated by a trough of ≥20° to be considered separate events. A peak height of 20° and peak width of 1 second were chosen to highlight arm movements that are deliberate—arm elevation changes of <20° and <1 second are likely normal variations in a static condition that encompass any orientation error from the IMUs and any change in orientation of the IMUs within the compression shirt due to slight body movements.23 To evaluate the intensity of arm activity, the elevation time series underwent discretization into epochs of 60 seconds.13 For each epoch, arm activity was classified as low if ≤3 elevation events occurred; moderate, 4-9 elevation events; and high, ≥10 elevation events. The number of epochs with low-, moderate-, or high-intensity classifications was normalized to the total number of epochs, defining the proportion of the day spent at each activity level.
Clinical outcomes
Preoperatively and 3 months and 1 year postoperatively, patients were asked to complete a series of validated patient-reported outcome questionnaires. These included a pain rating from 0 to 10 and the Subjective Shoulder Value; American Shoulder and Elbow Surgeons score; Simple Shoulder Test score; Disabilities of the Arm, Shoulder and Hand score; and Constant-Murley score. Active forward elevation, lateral abduction, and adducted external rotation were measured using a 30-cm manual goniometer. Active internal rotation was recorded as the highest point along the spine attainable with the thumb pointing upward.
Statistical analysis
Mixed-effects analysis was applied to each clinical outcome and data metric to evaluate any differences between time points, accounting for missing values. Mixed-effects analysis in this instance is analogous to a 1-way repeated-measures analysis of variance with the Tukey post hoc test for multiple comparisons; however, it is robust to missing values by applying a maximum likelihood model. The maximum likelihood model represents the combination of model parameter values with the highest probability of drawing the provided data sample, assuming that values are missing for random reasons.22 The effect of arm dominance was also investigated. The Pearson correlation coefficient was assessed to determine any relationships between clinical outcomes and IMU metrics at each time point. Statistical significance was set at P < .05, and all analysis was completed using Prism software (version 8; GraphPad Software, La Jolla, CA, USA).
Results
The mean age at the time of surgery was 71 ± 8 years, and 58% of patients were men. In 70% of patients, the arm undergoing RTSA was the dominant arm. Full patient demographic characteristics are reported in Table I .
Table I.
Patient demographic characteristics
| Characteristic | Data |
|---|---|
| Age, mean ± SD, yr | 71 ± 8 |
| Sex: M/F | 19/14 |
| Indication | |
| CTA | 14 |
| OA | 11 |
| MRCT | 4 |
| OA and RCT | 3 |
| RA | 1 |
| Operative arm: dominant/nondominant | 23/10 |
| BMI, mean ± SD | 31 ± 6 |
SD, standard deviation; M, male; F, female; CTA, cuff tear arthropathy; OA, osteoarthritis; MRCT, massive rotator cuff tear; RCT, rotator cuff tear; RA, rheumatoid arthritis.
IMU data
Preoperatively, 29 patients donned the instrumented garment, with a mean time of wear of 7.1 hours (range, 4.2-13.2 hours; >6 hours in 72%). Three months postoperatively, 27 patients donned the garment for an average of 6.6 hours (range, 1.8-11.2 hours; >6 hours in 74%), and one year postoperatively, 19 patients wore the garment for an average of 5.9 hours (range, 3.7-9.8 hours; >6 hours in 79%). Preoperatively, instrumentation was not performed in 4 patients because their visit to the clinic coincided with further same-day preoperative assessment for which application of the IMUs would hinder appropriate examination. Three months postoperatively, instrumentation was not performed in 6 patients because of unavailability of the study IMU devices, which were all currently deployed owing to a sudden surge in recruitment. One year postoperatively, instrumentation was not performed in 3 patients as a result of contralateral arthroplasty, preventing donning of the over-the-head garment; 2 patients refused further follow-up; and 9 patients were not assessed because of the transition from in-person to remote follow-up resulting from the COVID-19 (coronavirus disease 2019) pandemic.
Three months (P = .040) and one year (P = .010) postoperatively, patients demonstrated a significantly greater number of elevation events per hour > 90°, approximately doubling the preoperative instance. No other significant differences were observed between time points (Table II ). Patients spent the greatest proportion of the day performing low-intensity activity and the least amount of time performing high-intensity activity, regardless of whether they were in the preoperative, early healing, or healed phase of RTSA. One year postoperatively, <1% of the day, on average, was spent at >90° of elevation.
Table II.
Inertial measurement unit data metrics
| Preoperative | Postoperative |
P value | ||
|---|---|---|---|---|
| 3 mo | 1 yr | |||
| No. of elevation events/h | 125 ± 53 | 127 ± 59 | 150 ± 72 | .105 |
| No. of elevation events/h > 90° | 5 ± 4 | 10 ± 8 | 11 ± 9 | .006∗ |
| % of time spent at >90° of elevation | 0.73 ± 1.24 | 1.25 ± 1.67 | 0.78 ± 1.01 | .321 |
| % of day spent performing low-intensity activity | 62 ± 15 | 64 ± 15 | 61 ± 17 | .451 |
| % of day spent performing moderate-intensity activity | 30 ± 9 | 27 ± 11 | 29 ± 10 | .592 |
| % of day spent performing high-intensity activity | 8 ± 7 | 8 ± 9 | 10 ± 11 | .360 |
Data are presented as mean ± standard deviation.
Statistically significant (P < .05).
Similarly, no significant differences were observed in the proportion of the day spent within ranges of elevation angles at any time point, with the greatest amount of time spent between 0° and 20° of elevation (Table III ) and monotonically less time spent at higher elevation angles (Fig. 1 ). Considering the peak elevation angle at which elevation events occurred, although not significant for angles between 0° and 80°, there was a trend toward fewer elevation events at lower angles, as well as a greater number of elevation events at higher angles, between the preoperative and postoperative conditions (Table IV , Fig. 2 ). This was most notable for elevation events > 100°, showing a significantly greater proportion of such events 3 months postoperatively compared with preoperatively (P = .020). Although a significant difference in the proportion of elevation events was observed between 80° and 100° of arm elevation, the results were underpowered to detect any significant difference within the multiple comparisons between time points (all P ≥ .075).
Table III.
Proportion of day spent at varying arm elevation angles
| Elevation angle, ° | Preoperative, % | Postoperative, % |
P value | |
|---|---|---|---|---|
| 3 mo | 1 yr | |||
| 0-20 | 52 ± 14 | 55 ± 16 | 50 ± 15 | .496 |
| 20-40 | 31 ± 12 | 29 ± 11 | 33 ± 7 | .334 |
| 40-60 | 12 ± 8 | 10 ± 7 | 13 ± 10 | .515 |
| 60-80 | 3 ± 4 | 4 ± 4 | 3 ± 3 | .890 |
| 80-100 | 1 ± 1 | 2 ± 3 | 1 ± 1 | .282 |
| >100 | 0.5 ± 1 | 1 ± 1 | 0.5 ± 1 | .242 |
Data are presented as mean ± standard deviation. Percentages may not add up to 100% because of rounding.
Figure 1.
Binning of elevation angles throughout day preoperatively (Preop, solid blue), 3 months postoperatively (striped yellow), and 1 year postoperatively (checkered red). No significant differences (all P ≥ .242) were observed within any bin between time points.
Table IV.
Proportion of elevation events at varying arm elevation angles.
| Elevation angle, ° | Preoperative, % | Postoperative, % |
P value | |
|---|---|---|---|---|
| 3 mo | 1 yr | |||
| 20-40 | 49 ± 11 | 46 ± 15 | 43 ± 16 | .218 |
| 40-60 | 32 ± 7 | 29 ± 6 | 31 ± 5 | .192 |
| 60-80 | 12 ± 4 | 14 ± 7 | 15 ± 8 | .206 |
| 80-100 | 4 ± 2 | 6 ± 5 | 6 ± 4 | .041∗ |
| >100 | 3 ± 2 | 5 ± 4 | 4 ± 3 | .030∗ |
Data are presented as mean ± standard deviation. Percentages may not add up to 100% because of rounding.
Statistically significant (P < .05).
Figure 2.
Binning of elevation events preoperatively (Preop, solid blue), 3 months postoperatively (striped yellow), and 1 year postoperatively (checkered red). ∗Three months postoperatively, a significantly greater proportion of elevation events occurred at >100° of elevation compared with preoperatively (P = .020).
When the results were separated into cohorts based on hand dominance and operative side, the dominant cohort demonstrated significantly more elevation events per hour > 90° and a greater proportion of elevation events > 100° at 3 months (mean difference, 5 [P = .043] and 2.6% [P = .034], respectively) and 1 year (mean difference, 5 [P = .031] and 1.6% [P = .045], respectively) compared with preoperatively (Table V ). There were no significant differences between time points for any IMU metric within the nondominant cohort (all P ≥ .200) (Table VI ).
Table V.
Dominant-cohort metrics
| Preoperative | Postoperative |
P value | ||
|---|---|---|---|---|
| 3 mo | 1 yr | |||
| IMU data metric | ||||
| No. of elevation events/h | 128 ± 52 | 137 ± 58 | 152 ± 77 | .197 |
| No. of elevation events/h > 90° | 6 ± 4 | 11 ± 8 | 11 ± 9 | .012∗ |
| % of time spent at >90° of elevation | 0.63 ± 0.94 | 1.29 ± 1.67 | 0.81 ± 1.12 | .255 |
| % of day spent performing low-intensity activity | 61 ± 15 | 62 ± 15 | 62 ± 17 | .677 |
| % of day spent performing moderate-intensity activity | 31 ± 9 | 30 ± 11 | 28 ± 10 | .652 |
| % of day spent performing high-intensity activity | 9 ± 7 | 9 ± 9 | 10 ± 12 | .452 |
| Proportion of day at varying arm elevation angles, % | ||||
| 0°-20° | 54 ± 13 | 53 ± 16 | 49 ± 14 | .576 |
| 20°-40° | 29 ± 11 | 28 ± 9 | 33 ± 7 | .157 |
| 40°-60° | 12 ± 8 | 12 ± 7 | 13 ± 10 | .968 |
| 60°-80° | 4 ± 4 | 4 ± 4 | 3 ± 3 | .666 |
| 80°-100° | 1 ± 1 | 2 ± 3 | 1 ± 1 | .199 |
| >100° | 0.5 ± 0.5 | 1 ± 1 | 0.5 ± 1 | .177 |
| Proportion of elevation events at varying arm elevation angles, % | ||||
| 20°-40° | 48 ± 12 | 43 ± 14 | 42 ± 15 | .297 |
| 40°-60° | 32 ± 7 | 30 ± 7 | 32 ± 4 | .315 |
| 60°-80° | 13 ± 5 | 15 ± 7 | 15 ± 8 | .400 |
| 80°-100° | 4 ± 2 | 7 ± 5 | 6 ± 4 | .067 |
| >100° | 3 ± 2 | 5 ± 3 | 4 ± 2 | .017∗ |
IMU, inertial measurement unit.
Data are presented as mean ± standard deviation.
Statistically significant (P < .05).
Table VI.
Nondominant-cohort metrics
| Preoperative | Postoperative |
P value | ||
|---|---|---|---|---|
| 3 mo | 1 yr | |||
| IMU data metric | ||||
| No. of elevation events/h | 117 ± 61 | 103 ± 58 | 143 ± 57 | .251 |
| No. of elevation events/h > 90° | 5 ± 4 | 7 ± 6 | 9 ± 11 | .407 |
| % of time spent at >90° of elevation | 1.34 ± 1.89 | 0.75 ± 1.05 | 0.51 ± 0.55 | .656 |
| % of day spent performing low-intensity activity | 68 ± 14 | 72 ± 13 | 56 ± 14 | .224 |
| % of day spent performing moderate-intensity activity | 26 ± 9 | 22 ± 8 | 33 ± 8 | .224 |
| % of day spent performing high-intensity activity | 6 ± 7 | 6 ± 7 | 10 ± 8 | .300 |
| Proportion of day at varying arm elevation angles, % | ||||
| 0°-20° | 46 ± 15 | 59 ± 17 | 50 ± 20 | .341 |
| 20°-40° | 39 ± 14 | 32 ± 15 | 31 ± 7 | .441 |
| 40°-60° | 11 ± 8 | 6 ± 4 | 14 ± 14 | .244 |
| 60°-80° | 1 ± 1 | 1 ± 1 | 3 ± 4 | .887 |
| 80°-100° | 1 ± 1 | 1 ± 1 | 1 ± 1 | .908 |
| >100° | 0.5 ± 1 | 1 ± 1.5 | 0.5 ± 0.5 | .760 |
| Proportion of elevation events at varying arm elevation angles, % | ||||
| 20°-40° | 53 ± 9 | 52 ± 14 | 47 ± 20 | .679 |
| 40°-60° | 31 ± 6 | 28 ± 4 | 30 ± 8 | .251 |
| 60°-80° | 10 ± 4 | 11 ± 6 | 15 ± 9 | .200 |
| 80°-100° | 3 ± 1 | 4 ± 3 | 5 ± 4 | .426 |
| >100° | 3 ± 2 | 5 ± 6 | 3 ± 4 | .414 |
IMU, inertial measurement unit.
Data are presented as mean ± standard deviation.
Clinical outcomes
Patients showed significant improvement in all clinical outcome measures preoperatively through 3 months postoperatively, between 3 months and 1 year postoperatively, and preoperatively to 1 year postoperatively, with the exception of external rotation at all time points and internal rotation preoperatively to 3 months postoperatively (P > .999) (Table VII ). The gain in function and reduction in pain between time points were also greater than the minimal clinically important difference for each outcome for which the minimal clinically important difference following RTSA has previously been reported (forward elevation, 12°; lateral abduction, 7°; pain rating, 1.6; American Shoulder and Elbow Surgeons score, 13.6; Simple Shoulder Test score, 1.5; and Constant-Murley score, 5.7), with the exception of the pain rating between 3 months and 1 year.31
Table VII.
Patient-reported outcome measures
| Preoperative | Postoperative |
P value | ||
|---|---|---|---|---|
| 3 mo | 1 yr | |||
| Forward elevation, ° | 73 ± 31 | 98 ± 29 | 123 ± 18 | <.001∗ |
| Lateral abduction, ° | 65 ± 27 | 90 ± 24 | 107 ± 24 | <.001∗ |
| External rotation, ° | 28 ± 22 | 29 ± 16 | 38 ± 17 | .059 |
| Internal rotation (1-6)† | 3 ± 2 | 3 ± 2 | 4 ± 1 | .001∗ |
| Pain score (0-10) | 7.1 ± 2.3 | 2.6 ± 2.1 | 1.2 ± 1.5 | <.001∗ |
| SSV (0-100) | 33 ± 22 | 70 ± 20 | 85 ± 15 | <.001∗ |
| ASES score (0-100) | 34 ± 16 | 65 ± 17 | 83 ± 13 | <.001∗ |
| SST score (0-12) | 2.7 ± 2.1 | 5.2 ± 3.1 | 8.2 ± 2.7 | <.001∗ |
| DASH score (0-100) | 54 ± 16 | 41 ± 20 | 18 ± 17 | <.001∗ |
| Constant-Murley score (0-100) | 28 ± 13 | 51 ± 16 | 66 ± 71 | <.001∗ |
SSV, Subjective Shoulder Value; ASES, American Shoulder and Elbow Surgeons; SST, Simple Shoulder Test; DASH, Disabilities of the Arm, Shoulder and Hand.
Data are presented as mean ± standard deviation.
Statistically significant (P < .05).
Based on landmarks from Constant-Murley shoulder score: 1, lateral thigh; 2, buttock; 3, lumbosacral junction; 4, waist; 5, T12; and 6, T7 or interscapular.
On correlation of clinical outcomes to metrics derived from the IMUs at all time points, maximal forward elevation (r = 0.493, P < .001) (Fig. 3, a ) and lateral abduction (r = 0.394, P < .001) (Fig. 3, b) were significantly moderately correlated with the number of elevation events per hour > 90°, as was pain (r = –0.347, P = .002) (Fig. 3, c). Maximal forward elevation (Fig. 3, d) was moderately associated with the number of elevation events per hour (r = 0.395, P = .001) and the proportion of the day spent in low (r = –0.332, P = .004) and high (r = 0.347, P = .002) activity states (Fig. 3, e). The Subjective Shoulder Value (Fig. 3, f) was weakly though significantly correlated with the number of elevation events per hour > 90° (r = 0.290, P = .032) but no other IMU metrics.
Figure 3.
Scatter plots illustrating the relationships between forward elevation and number of elevation events/ per hour > 90° (a), lateral abduction and number of elevation events per /hour > 90° (b), pain and number of elevation events per /hour > 90° (c), forward elevation and number of elevation events per /hour (d), forward elevation and activity state (e), and Subjective Shoulder Value and number of elevation events per /hour > 90° (f).
Discussion
The purpose of this study was to investigate the continuous shoulder motion during normal activity prior to, in the early healing phase of, and in the late healing phase of RTSA using IMUs. Characterizing such shoulder motion can provide insight into the effect of shoulder joint replacement on daily living outside the clinical environment. There is a trend toward a greater number of elevation events per hour as the shoulder heals, with significantly more events at higher elevation angles. Extrapolating the daily value of 150 elevation events per hour, assuming a 16-hour waking day, the reverse shoulder undergoes approximately 876,000 cycles per year. Two previously published studies have extrapolated the estimated annual duty cycle of the shoulder using IMUs, ranging from approximately 65,000 to 4.5 million cycles per year.17 , 18 Although the presented annual count is within the range of these estimates, variations in the definition of “motion cycle” ought to be standardized for future further comparison between studies. These values, in turn, can be used in modeling or in vitro kinematics and wear testing to predict implant longevity.
It is interesting to note that there were no significant differences in shoulder activity intensity after RTSA as compared with preoperatively. One would theorize that after undergoing surgery designed to alleviate pain and improve range of motion, patients would increase their intensity of activity. This finding of no difference may be a reflection of the relatively elderly and obese patient population with limited functional demands, as well as the large standard deviations observed within the group. This may also explain why no differences were observed with respect to the proportion of the day spent at different elevation angles—both preoperatively and 1 year postoperatively, 95% of the day was spent at elevation angles < 60°. A previous study assessing shoulder biomechanics in the healthy elderly population found similar results, with 97% of the day spent at elevation angles < 90°, suggesting that even if the elderly population has the capacity to perform activities with a large range of motion, these patients seldom do so.4 It is also noteworthy that although no differences were observed with time, significantly more elevation events occurred at ranges > 100° at 3 months postoperatively compared with preoperatively, though not 1 year postoperatively compared with preoperatively. Increases at 3 months are likely a result of patients intentionally completing rehabilitation exercises in an effort to actively improve strength and range of motion. Not surprisingly, the Pearson correlations demonstrated that as active range of motion increased, so too did the number of elevation events per hour, the number of events > 90°, and activity intensity. The number of elevation events per hour > 90° was also significantly correlated to the Subjective Shoulder Value, and for this reason, rehabilitation practices following RTSA should continue to focus on improving active range of motion, in particular forward elevation. Overall, it appears that the increased elevation events to obtain motion at 3 months are transient and that patients settle back into their original preoperative movement event frequency.
Assessment of traditional patient-reported outcome measures (PROMs) demonstrated that RTSA is an effective technique for reducing pain and restoring active range of motion and shoulder function as early as 3 months postoperatively, with improved benefit through 1 year. Our outcomes at 1 year are consistent with those reported at a mean 2.8-year follow-up in a cohort with lateralized glenospheres.1 It is well known that although PROMs can improve the understanding of how a disease or treatment impacts patients' daily lives, patients' emotion, personal bias, missing data, and comorbidities may influence their response.16 Specifically, the PROMs used in this study were not designed intentionally for elderly patients, and therefore, responses garnering higher scores (eg, the ability to perform a specific activity) may have been selected postoperatively despite patients not actually performing the activity in their normal daily routine. This may have contributed to the subjectively higher PROM scores without an observed change in arm activity.
When considering the effect of arm dominance on IMU metrics, we found that the dominant cohort showed gains at 3 months and 1 year postoperatively in terms of the number of elevation events per hour > 90° and the proportion of elevation events > 100° whereas the nondominant cohort presented no observable differences in any metric. A handful of studies have used IMUs to compare the effects of dominance on functional metrics in both healthy and patient populations. In a study evaluating healthy subjects, Nam et al23 showed that the working volume of the dominant hand for activities of daily living was significantly greater than that of the nondominant hand. Coley et al6 reported no significant difference in the frequency and duration of arm positions during daily activities in healthy participants, although there was a trend toward a higher frequency with the dominant extremity. Hurd et al12 applied a triaxial accelerometer to the upper arms of both a control group and patient group scheduled to undergo shoulder arthroplasty, measuring a difference in activity between dominant and nondominant arms (control), as well as involved and uninvolved arms (patient). They reported that the dominant and uninvolved upper arms showed significantly greater activity. Our results reflect the findings of these studies, suggesting that limb dominance does influence postoperative function.
To our knowledge, this is one of the first prospective studies investigating continuous arm motion in an RTSA patient population, acquiring both clinical outcomes and objective metrics from IMUs. The primary limitations of this study are the patient dropout rate at 1 year—only 14 patients were assessed at all time points—and the short duration of monitoring at each time point. The garment was donned on the same day as clinical follow-up, as many patients resided beyond the study center and did not want to make multiple trips into the city. Consequently, a full day's activities may not be fully represented by the logged data and may also include a period of driving home. To account for missing values at each time point, the mixed-effects model of analysis was used, as is frequently done in the case of analysis of unbalanced repeated measures,33 although it should be noted that the heterogeneity in wear time at different time points may have affected the results. In particular, this may be responsible for the increase in the number of elevation events > 90° both 3 months and 1 year postoperatively without an observed increase in time spent at >90°. Another explanation for this apparent incongruence is that patients may use their arm more efficiently as a result of both increased range of motion and reduction in pain. Postoperatively, it may take less time to complete overhead tasks than preoperatively, and therefore, more events may take place in the same amount of time.
To ensure there was sufficient space on the onboard secure digital card to record all data, quaternions were logged at a frequency of 10 Hz. Although a higher logging rate would provide a smoother time series, as the study cohort comprised elderly patients, it was proposed that activities would likely not be of very high intensity. This assumption was reflected in the results, showing that approximately 90% of the day was spent in low or moderate activity phases, with <10 elevation events per minute.
A source of error in the data may have arisen from the use of walking aids, artificially increasing the duration spent at higher elevation angles, such as in the case of a walker. Another source of error is motion artifact of the garment while it is worn. Although a tight-fitting shirt was chosen for each patient, it is possible for the position of the IMUs relative to the humerus and torso to shift slightly during the day. The calibration procedure also may have led to an underestimation of elevation angles. When the closest position of the arm to the body in adduction is taken as 0° of elevation, patients with more adipose tissue and associated adduction deficit would record fewer data points at higher elevation angles than patients without this adduction deficit.
Another limitation of this study is that the IMUs were not specifically validated against optical tracking techniques. A previous study by El-Gohary and McNames10 compared the use of IMUs with an unscented Kalman filter sensor fusion orientation estimation algorithm vs. 3-dimensional optical tracking for upper-extremity motion with human subjects. They demonstrated a root-mean-square angle error < 8° and average correlation coefficient (r) ≥ 0.95 between the 2 tracking techniques, showing good reliability and validity for the use of IMUs in upper-extremity motion tracking. This error is greater than the manufacturer-specified accuracy of ±1° and likely more representative of organic human movement. Future studies ought to validate the use of their sensor techniques prior to clinical application.
Moreover, future studies should consider standardizing data acquisition and analysis methods to facilitate comparison between studies and identify metrics of clinical importance. They could also investigate the effect of sex, different planes of motion, and the angle at which intentional arm movements occur using the results from healthy age-matched controls to provide targets for rehabilitation practices.
Overall, IMUs provide an objective measure of patient arm use and activity. With an increasing number of studies reporting objective measurements following joint replacement, there is the potential to define their typical postoperative performance. This may help patients establish a realistic set of expectations following the procedure, which in turn may decrease the risk of patient dissatisfaction.3 The application of IMUs can be extended to the assessment of different shoulder pathologies and treatments, complementing the clinician's assessment and addressing the limitations of PROMs. Taken together, objective and subjective measurements, in addition to the degree to which patient expectations have been met, can help define a successful procedure.
Conclusion
After reverse shoulder arthroplasty, patients significantly increase the frequency of arm elevation events to higher elevation angles. However, we found no differences in the amount of time spent at different elevation angles when comparing preoperative with postoperative data. Overall, after RTSA, >95% of the day was spent at elevation angles < 60° and <1% of the day was spent at >90° of elevation.
Acknowledgments
The authors thank Daniel Langohr, PhD, for his insight into the use of inertial measurement units for the shoulder.
Disclaimer
This study was funded by the Arthritis Society Young Investigator Operating Grant. No outside source of funds was involved in data collection, data analysis, or the preparation or editing of the manuscript.
George S. Athwal has received research support funding, intellectual property royalties, and stock or stock options from Wright Medical Group. No company had any input into the study design, protocol, testing, data analysis, or manuscript preparation. All the other authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.
Footnotes
This study was approved by the Western University Health Sciences Research Ethics Board (project ID 105908).
References
- 1.Athwal G.S., MacDermid J.C., Reddy K.M., Marsh J.P., Faber K.J., Drosdowech D. Does bony increased-offset reverse shoulder arthroplasty decrease scapular notching? J Shoulder Elbow Surg. 2015;24:468–473. doi: 10.1016/j.jse.2014.08.015. [DOI] [PubMed] [Google Scholar]
- 2.Bloomfield R.A., Fennema M.C., McIsaac K.A., Teeter M.G. Proposal and validation of a knee measurement system for patients with osteoarthritis. IEEE Trans Biomed Eng. 2019;66:319–326. doi: 10.1109/TBME.2018.2837620. [DOI] [PubMed] [Google Scholar]
- 3.Bourne R.B., Chesworth B.M., Davis A.M., Mahomed N.N., Charron K.D.J. Patient satisfaction after total knee arthroplasty: who is satisfied and who is not? Clin Orthop Relat Res. 2010;468:57–63. doi: 10.1007/s11999-009-1119-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Chapman R.M., Torchia M.T., Bell J.E., Van Citters D.W. Assessing shoulder biomechanics of healthy elderly individuals during activities of daily living using inertial measurement units: high maximum elevation is achievable but rarely used. J Biomech Eng. 2019;141:1–7. doi: 10.1115/1.4042433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Chapman R.M., Torchia M.T., Bell J.E., Van Citters D.W. Continuously monitoring shoulder motion after total shoulder arthroplasty: maximum elevation and time spent above 90° of elevation are critical metrics to monitor. J Shoulder Elbow Surg. 2019;28:1505–1514. doi: 10.1016/j.jse.2019.01.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Coley B., Jolles B.M., Farron A., Aminian K. Arm position during daily activity. Gait Posture. 2008;28:581–587. doi: 10.1016/j.gaitpost.2008.04.014. [DOI] [PubMed] [Google Scholar]
- 7.Cutti A.G., Giovanardi A., Rocchi L., Davalli A., Sacchetti R. Ambulatory measurement of shoulder and elbow kinematics through inertial and magnetic sensors. Med Biol Eng Comput. 2008;46:169–178. doi: 10.1007/s11517-007-0296-5. [DOI] [PubMed] [Google Scholar]
- 8.Dobkin B.H., Martinez C. Wearable sensors to monitor, enable feedback, and measure outcomes of activity and practice. Curr Neurol Neurosci Rep. 2018;18:1–8. doi: 10.1007/s11910-018-0896-5. [DOI] [PubMed] [Google Scholar]
- 9.Doğan M., Koçak M., Onursal Kılınç Ö., Ayvat F., Sütçü G., Ayvat E. Functional range of motion in the upper extremity and trunk joints: nine functional everyday tasks with inertial sensors. Gait Posture. 2019;70:141–147. doi: 10.1016/j.gaitpost.2019.02.024. [DOI] [PubMed] [Google Scholar]
- 10.El-Gohary M., McNames J. Shoulder and elbow joint angle tracking with inertial sensors. IEEE Trans Biomed Eng. 2012;59:2635–2641. doi: 10.1109/TBME.2012.2208750. [DOI] [PubMed] [Google Scholar]
- 11.Fennema M.C., Bloomfield R.A., Lanting B.A., Birmingham T.B., Teeter M.G. Repeatability of measuring knee flexion angles with wearable inertial sensors. Knee. 2019;26:97–105. doi: 10.1016/j.knee.2018.11.002. [DOI] [PubMed] [Google Scholar]
- 12.Hurd W.J., Morrow M.M., Kaufman K.R. Tri-axial accelerometer analysis techniques for evaluating functional use of the extremities. J Electromyogr Kinesiol. 2013;23:924–929. doi: 10.1016/j.jelekin.2013.03.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hurd W.J., Morrow M.M., Miller E.J., Adams R.A., Sperling J.W., Kaufman K.R. Novel approaches to objectively assess shoulder function. J Shoulder Elbow Surg. 2014;23:e251–e255. doi: 10.1016/j.jse.2014.02.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Jung Y., Kang D., Kim J. Upper body motion tracking with inertial sensors. 2010 IEEE Int Conf Robot Biomimetics. 2010:1746–1751. doi: 10.1109/ROBIO.2010.5723595. [DOI] [Google Scholar]
- 15.Kiet T.K., Feeley B.T., Naimark M., Gajiu T., Hall S.L., Chung T.T. Outcomes after shoulder replacement: comparison between reverse and anatomic total shoulder arthroplasty. J Shoulder Elbow Surg. 2015;24:179–185. doi: 10.1016/j.jse.2014.06.039. [DOI] [PubMed] [Google Scholar]
- 16.Kingsley C., Patel S. Patient-reported outcome measures and patient-reported experience measures. BJA Educ. 2017;17:137–144. doi: 10.1093/bjaed/mkw060. [DOI] [Google Scholar]
- 17.Kirking B., El-Gohary M., Kwon Y. The feasibility of shoulder motion tracking during activities of daily living using inertial measurement units. Gait Posture. 2016;49:47–53. doi: 10.1016/j.gaitpost.2016.06.008. [DOI] [PubMed] [Google Scholar]
- 18.Langohr G.D.G., Haverstock J.P., Johnson J.A., Athwal G.S. Comparing daily shoulder motion and frequency after anatomic and reverse shoulder arthroplasty. J Shoulder Elbow Surg. 2018;27:325–332. doi: 10.1016/j.jse.2017.09.023. [DOI] [PubMed] [Google Scholar]
- 19.Law T.Y., Rosas S., George F., Kurowicki J., Formaini N., Levy J. Short-term projected use of reverse total shoulder arthroplasty in proximal humerus fracture cases recorded in Humana's National private-payer database. Am J Orthop (Belle Mead NJ) 2017;46:E28–E31. [PubMed] [Google Scholar]
- 20.Liao Y., Thompson C., Peterson S., Mandrola J., Beg M.S. The future of wearable technologies and remote monitoring in health care. Am Soc Clin Oncol Educ B. 2019;39:115–121. doi: 10.1200/edbk_238919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lin J.F.S., Kulić D. Human pose recovery using wireless inertial measurement units. Physiol Meas. 2012;33:2099–2115. doi: 10.1088/0967-3334/33/12/2099. [DOI] [PubMed] [Google Scholar]
- 22.Myung I.J. Tutorial on maximum likelihood estimation. J Math Psychol. 2003;47:90–100. doi: 10.1016/S0022-2496(02)00028-7. [DOI] [Google Scholar]
- 23.Nam H.S., Lee W.H., Seo H.G., Kim Y.J., Bang M.S., Kim S. Inertial measurement unit based upper extremity motion characterization for action research arm test and activities of daily living. Sensors (Switzerland) 2019;19:1–10. doi: 10.3390/s19081782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.O'Reilly M., Caulfield B., Ward T., Johnston W., Doherty C. Wearable inertial sensor systems for lower limb exercise detection and evaluation: a systematic review. Sport Med. 2018;48:1221–1246. doi: 10.1007/s40279-018-0878-4. [DOI] [PubMed] [Google Scholar]
- 25.Oosterwijk A.M., Nieuwenhuis M.K., van der Schans C.P., Mouton L.J. Shoulder and elbow range of motion for the performance of activities of daily living: a systematic review. Physiother Theory Pract. 2018;34:505–528. doi: 10.1080/09593985.2017.1422206. [DOI] [PubMed] [Google Scholar]
- 26.Pandya J., Johnson T., Low A.K. Shoulder replacement for osteoarthritis: a review of surgical management. Maturitas. 2018;108:71–76. doi: 10.1016/j.maturitas.2017.11.013. [DOI] [PubMed] [Google Scholar]
- 27.Porciuncula F., Roto A.V., Kumar D., Davis I., Roy S., Walsh C.J. Wearable movement sensors for rehabilitation: a focused review of technological and clinical advances. PM R. 2018;10:S220–S232. doi: 10.1016/j.pmrj.2018.06.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Robert-Lachaine X., Mecheri H., Muller A., Larue C., Plamondon A. Validation of a low-cost inertial motion capture system for whole-body motion analysis. J Biomech. 2019;99:109520. doi: 10.1016/j.jbiomech.2019.109520. [DOI] [PubMed] [Google Scholar]
- 29.Roldán-Jiménez C., Martin-Martin J., Bennett P., Mcphail S., Vargas A.C., Kerr G. Assessment of abduction movement in older people with painful shoulder: an analysis based on inertial sensors. BMC Musculoskelet Disord. 2019;20 doi: 10.1101/398123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rugg C.M., Coughlan M.J., Lansdown D.A. Reverse total shoulder arthroplasty: biomechanics and indications. Curr Rev Musculoskelet Med. 2019;12:542–553. doi: 10.1007/s12178-019-09586-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Simovitch R., Flurin P.H., Wright T., Zuckerman J.D., Roche C.P. Quantifying success after total shoulder arthroplasty: the minimal clinically important difference. J Shoulder Elbow Surg. 2018;27:298–305. doi: 10.1016/j.jse.2017.09.013. [DOI] [PubMed] [Google Scholar]
- 32.Thon S.G., Seidl A.J., Bravman J.T., McCarty E.C., Savoie F.H., Frank R.M. Advances and update on reverse total shoulder arthroplasty. Curr Rev Musculoskelet Med. 2020;13:11–19. doi: 10.1007/s12178-019-09582-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Vonesh E.F., Carter R.L. Mixed-effects nonlinear regression for unbalanced repeated measures. Biometrics. 1992;48:1. [PubMed] [Google Scholar]
- 34.Walmsley C.P., Williams S.A., Grisbrook T., Elliott C., Imms C., Campbell A. Measurement of upper limb range of motion using wearable sensors: a systematic review. Sport Med Open. 2018;4 doi: 10.1186/s40798-018-0167-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Zhou H., Hu H., Harris N.D., Hammerton J. Applications of wearable inertial sensors in estimation of upper limb movements. Biomed Signal Process Control. 2006;1:22–32. doi: 10.1016/j.bspc.2006.03.001. [DOI] [Google Scholar]