Journal of Surgery

Introduction

Orthopaedic surgeons perform physically demanding and repetitive tasks during surgery. The occupational hazards [1-4], risk of burnout [5-7], ergonomic factors related to risk of injury [8-11] and fatigue [12] associated with orthopaedic surgery have been well documented in the literature, and work-related musculoskeletal disorders and injuries have been shown to be a leading cause of injury and absenteeism [13]. A recent study also showed that one hundred percent of surveyed orthopaedic surgeons who had been in practice between 21 and 30 years reported some form of musculoskeletal overuse disorder. Shoulder overuse disorders (rotator cuff disease, biceps tendonitis, and other tendinopathies) were the most common reported injuries for adult reconstructive surgeons [4]. Research has also been conducted to explore the relation of case-order in total joint replacement and patient complications [14]. In primary Total Hip Arthroplasty (THA), a mallet is traditionally used for multiple steps in the procedure: femoral broaching, stem insertion, cup impaction, and impaction of head onto stem. It is estimated that a surgeon may swing a mallet as many as 300 times during each case [2], and 12 ± 9 times for each femoral broach [1].A surgical automated impactor was developed to automate impaction steps in THA procedures.

A potential benefit of this technology is the ability to reduce surgeon muscle fatigue, and over time, reduce the risk of surgeon injury for orthopaedic surgeons. The impactor’s effect on the surgeon’s muscle fatigue has been investigated in simulated broaching [15] duringTotal Hip Arthroplasty(THA) with the use of Surface Electromyography (sEMG). Results from this study showed that a group of surgeons performing a set of impactions in a lab-controlled setup had significantly lower levels of muscle fatigue compared to manual impactions. However, this has not yet been evaluated in real-life THA procedures.Muscle fatigue is a result of an intensive activity maintained over time and at a high effort. However, effort is user dependent [16]. Factors contributing to fatigue include posture, strategy, force, stature, training level, type of contraction (e.g eccentric, concentric) and is a result of the relationship between power and duration [17].

In addition, motivation, physical fitness, nutritional status, and the types of motor units (i.e., fibres) recruited based on the intensity and duration of activity have an impact on the user’s performance. An increase in muscle activation is the result of a progressive increase in muscle fiber recruitment, with attempt to complete a given task, and muscle fiber fatigue results in an inability to produce the same force level [18]. For this reason, a change in muscle activation at surgical steps of interest, can be used to characterize the effect on the user of the specific workflow. On the contrary, a decrease, or no change in muscle activation can suggest that muscles can maintain the required force. For these reasons, when muscle fatigue is investigated in a real-world environment, all these factors should be considered, and analysis should be performed focusing on identified workflow steps and user-specific analysis.Therefore, the aim of this study was to quantify changes in muscle activation during manual and automated THA workflows to characterize surgeon specific performance, by means of EMG. Our hypothesis was that 1) increased muscle activation would be seen in manual broaching compared to automated broaching across the surgeons being studied; and 2) different surgeons would have different patterns of muscle activation depending on their broaching position or style.

Methods

Level of Evidence: II

As part of a prospective, Randomized, Controlled Trial (RCT), four male orthopaedic surgeons performed 77 direct anterior approach THA cases randomized into two groups:39 automated impaction cases (KINCISE^TM; DePuy Synthes, Warsaw, IN, USA) in the study group, and 38 manual impaction cases in the mallet control group. Cases were performed at different locations over a period of 2 years (2019-2021).The trial is registered on ClinicalTrials.gov (NCT04191733). Patients and surgeons provided signed consent to participate and Institutional Review Board (IRB) approval was in place for the duration of the study. For one patient only, osteoporotic bone was reported (DORR Class C), whereas the remainder of cases reported bone quality rated between Good and Normal (DORR Class A and B). Two cases reported calcar fracture. To record data on muscle performance, sEMG signals were collected using a compression shirt (Athos, Mad Apparel, Inc.) containing built-in surface electrodes situated over various muscle groups, including bilateral Biceps Brachii (RBB, LBB),Anterior Deltoid (RAD, LAD), Latissimus Dorsi (RLD, LLD), Triceps Brachii (RTB, LTB) and Pectoralis (RP, LP) muscles. During each case, surgeons wore the Athos shirt underneath the scrubs to allow the electrodes to be in direct contact with the muscle compartments identified (Figure1).

Figure 1: Athos T-shirt with surface EMG sensors embedded

Surface EMG data was recorded uninterruptedly for the entirety of each case. The time corresponding to the start and end of broaching and start and end of the procedure were collected and saved separately.Data provided from Athos was already pre-conditioned and in the form of sEMG Root Mean Square (RMS) envelopes. For this study, each envelope was then segmented by extracting burst of activity corresponding to four time points: 1) start of the procedure, 2) start of broaching, 3) end of broaching and 4) end of the procedure. The average amplitude was then quantified for each time point, to investigate the trend over time and changes of muscle activation in the automated and manual groups during THA. Data analysis was performed with MATLAB (Mathworks, Inc).EMG average lower than noise level (~30µV) were discarded from statistical analysis. Together with quantitative data, qualitative data, in the form of a questionnaire, and pictures of surgical ergonomic form were gathered from the group of surgeons. The questions focused on personal strategy in performing manual and automated THA, such as the posture used during both workflows, whether there were sensations of fatigue, and the percentage of manual and automated cases. The answers of the questionnaire are reported in the results section.Statistical analysis was performed to investigate the difference in muscle activation features (number of muscles recruited and amplitude of active muscles) as a way of describing muscle fatigue and the strategies adopted by a group of orthopaedic surgeons. This was performed with an ANOVA General Linear Model (Minitab 18, Statistical Software (2010). State College, PA: Minitab, Inc.), where the response was set as RMS values and fixed factors set as modalities (automated, manual), muscles and surgeons. The confidence level was α<0.05. To discern surgeons’ strategy and surgical ergonomic effect on performance, results are presented for each surgeon to identify the number of muscles with highest activation for each workflow at each stage.

Patient and Surgeon Demographics

Osteoporotic bone was reported for one patient (DORR Class C; automated), whereas the remainder of cases reported bone quality rated either Good or Normal. DORR class was A in 32 patients, B in 43, and C in 2. One case reported calcar fracture in the manual group. Patient demographic data were not different between groups (Table 1). (Surgeon 1) reported previous history of musculoskeletal injury at the time of the clinical trial (shoulder scope decompression for subacromial impingement).

Table 1: Patient characteristics Results

The answers to the questionnaire completed by surgeons are summarized in (Table 2). Surgeons recruited muscle compartments at different intensities. (Figure 2) shows the difference in muscle activation between automated and manual cases. Statistical significance was observed for Surgeon 2 (p=0.02). Both (Surgeon 3 and 4) showed higher activation for manual cases at different degrees but without statistical significance.

Table 2: Surgeon Details

Figure 2: Range of muscle activation for each surgeon for manual cases and automated cases (statistically significant difference noted for Surgeon 2 (p<0.001)).

In terms of differences over time, (Figure3) shows that for Surgeon 1, 3 and 4 statistical significance is not observed over time, however differences in activation were observed. Surgeon 3 shows a decrease in muscle activation for automated cases during broaching (time point 2 and 3) and an increase in muscle activation during manual broaching. For manual cases, similar muscle activation was observed for Surgeon 1 (p=0.463), 3 (p=0.110) and 4 (p>0.05), however Surgeon 2 showed the highest muscle activation and it presented at the end of the procedure (p<0.001).

Figure 3: Range of muscle activation for each surgeon for manual and automated cases for each time point (statistically significant higer muscle activation at the end of the surgery for Surgeon 2 (p<0.001)).

As shown in (Figure 4), no difference was observed for Surgeon 1 between manual and automated cases (p=0.365). However, LAD, LBB, RBB, RTB, and LTB, showed the highest average muscle activation (p<0.001).

Figure 4: Surgeon 1 summary results (*denotes significant difference).

Figure 5: Surgeon 2 summary results (*denotes significant difference). (Figure 5) shows (Surgeon 2) presenting significantly higher muscle activation for manual cases (p<0.0001).Muscles with highest muscles activation for manual cases (p=0.008) were LAD, LP, RBB, RP and RTB. Surgeon 3 (Figure 6) showed statistically significant higher muscle activation was observed for manual cases (p=0.023) for LAD, LBB, RBB, RAD and LLD. LTB showed the highest activation for both groups, with no statistical difference (p>0.05) between manual and automated cases. Overall, (Surgeon 3) showed lower levels of muscles activation, resulting in 4 out of 10 muscle compartments discarded from analysis (EMG amplitude lower than noise level). For automated cases, LAD, LLD and RBB muscle activation was higher than the noise level for one case only, as shown in (Figure 6) by the crossed circle.As shown in (Figure 7), muscle activation data for (Surgeon 4) showed similar range values for muscle activation with no overall statistical significance. For manual cases, LBB showed the highest muscle activation, with no statistical difference (p=0.140). Highest muscle activation was observed for automated cases, specifically for LAD, LP, LTB, RAD, RBB and RP, with statistical difference only for RBB (p<0.01), with average muscle activation ranging between 62.3 uV and 131.2 uV.

Figure 6: Surgeon 3 summary results (*denotes significant difference).

Figure 7: Surgeon 4 summary results (*denotes significant difference).