A convenience sample composed of men and women with anterior shoulder instability (Instability Group, n=20), healthy swimmers (Athletes Group, n=19), and healthy sedentary subjects (Sedentary Group, n=21) were invited to participate on this cross-sectional study. The groups were homogenous regarding age, height, weight, body mass index. Anthropometric data and peak of strength are presented in Table 1. This study was conducted at the Universidade Federal de São Carlos (UFSCar), São Carlos, SP, Brazil according to the Helsinki Declaration and was approved by a Local Ethics Committee in Human Research (Process number: 186.596). All participants provided a signed consent form prior to data collection.

Individuals with instability were included if they had a diagnosis of traumatic anterior shoulder instability given by a specialized-board Orthopedic physician and had not received physical therapy treatment within the 6 months prior to this study. Swimmers were included if they had no history of upper limb injuries, had been training at a competitive level for at least a year, at a frequency of three to five times a week, with a training volume greater than twenty kilometers a week.7,17 The inclusion criteria for sedentary participants were no involvement in physical activities related to the upper limbs and no history of upper limb injuries. Exclusion criteria for all subjects were atraumatic instability, inferior or posterior traumatic instability, generalized laxity, a history of shoulder surgery, or previous use of corticosteroid injections.

Seventy-seven individuals were recruited in this study and seventeen were enrolled after been screened for inclusion. Ten individuals were excluded for presenting another type of glenohumeral instability, four subjects presented positive shoulder provocative tests, two had received physical therapy treatment or a shoulder injection within the previous 6 months, and one had undergone shoulder surgery.

Each subject performed all the tests on the same day and were assessed by the same examiner (i.e. a physical therapist with experience in these assessments). Protractor muscle maximal isometric strength was measured using an HHD (Lafayette Instrument, Lafayette, IN) and an isokinetic dynamometer (Biodex System 3, Computer Sports Medicine Inc., Stoughton, MA). While the dominant side was assessed in healthy subjects, the affected side was assessed in subjects with shoulder instability. Before the strength assessment, participants performed a standardized warm-up with 10 repetitions on the following movements: arm circumduction, and shoulder flexion, extension, abduction, and adduction. Three submaximal familiarization trials were performed with the isokinetic dynamometer and HHD prior to maximal tests in each condition assessed.18

The protraction strength assessments were performed in two different positions: seated19–21 and lying supine,16,22,23 in a random order. In the seated position, the participant was instructed to sit on a stool, 60cm high with no back support, keeping their feet on the floor and knees at 90° of flexion. The back and head were supported on a wall and the contralateral arm was positioned on the leg (Fig. 1A). In the lying supine position, the participants were instructed to lie on a mattress with the knees at 90° of flexion (Fig. 1B). In both positions, the assessed shoulder was positioned at 90° of flexion, full elbow extension, and neutral forearm.24

Figure 1.

Protraction isometric strength test using HHD in the seated position (A) and in the lying supine position (B) and using the isokinetic dynamometer.

(0.1MB).

During the assessments, the examiner provided manual resistance by holding the HHD between their trunk and the palmar face of the participant's hand. The test position was set as the midpoint between maximal scapular protraction and retraction, which was determined by the examiner. First, the examiner moved the arm of the participant through the maximal range of motion of protraction and retraction and then oriented the subject to maintain the arm approximately in the middle point of the range. Participants were asked to provide maximal strength like punching against the dynamometer after the evaluator's verbal command.

For the tests using the isokinetic dynamometer, participants were seated with the trunk and waist fixed by non-elastic belts. The assessment was performed with the shoulder at 90° of flexion, full elbow extension, and neutral forearm position, using the closed kinetic chain device of the isokinetic dynamometer (Fig. 1C). The dynamometer axis was aligned to the estimated glenohumeral joint center and fixed at zero degrees. Initially, maximal protraction and retraction ranges of motion were measured in this position, using the Biodex system. The mean position between maximum protraction and retraction was set as the strength test position. For maximal test performance, subjects were oriented to hold the handgrip of the device and push like punching with maximal strength.

For each test position, subjects performed three maximal isometric contractions of 5s each, with a 30-s interval between them. All participants rested for a period of 30-min between each position of HHD and between the instruments. The three groups were evaluated in three different days. The peak force of each trial was recorded in Newton (N).

Sample size calculation was performed a posteriori using the correlation test to determine the statistical power for the number of individuals in each group. In all comparisons between the HHD and isokinetic, effect sizes greater than 0.7 were verified, resulting in a statistical power of at least 80%. All anthropometric data, isometric peak force recorded with the HHD and with the isokinetic dynamometer were found to present normal distribution in the Shapiro–Wilk test for all three groups. Participants age, body mass, BMI and height were compared between groups using one-way ANOVAs. The mean of the three repetitions of HHD and isokinetic dynamometer was registered and normalized by the participant's body mass.

The correlations between HHD and isokinetic dynamometer strength measures were calculated using Pearson's correlation estimates. The correlations between HHD and isokinetic strength measures were calculated using Pearson's correlation estimates. Results were interpreted as follows: values less than 0.3 meaning weak correlation; 0.3–0.7, moderate correlation; and more than, 0.7 strong correlation.25

In addition, linear regression coefficient (r2) was calculated to determine if a variation in the results of the tested instrument (HHD) could be explained by the variation in the gold standard (isokinetic dynamometer).

Bland–Altman plots were generated for each of the two HHD positions contrasted with the isokinetic dynamometer in order to assess agreement between both instruments. Limits of agreement were calculated as well as the linear regression coefficient (r2) and their respective bias in relation to the maximum agreement. The analysis was conducted using – SPSS version 21 and Microsoft Excel and the significance level was set at 5% for all the tests.

A moderate correlation was observed between HHD measurements in the seated (r=0.59) and lying supine (r=0.54) positions compared to the isokinetic dynamometer for all individuals. Moreover, a strong correlation was observed between the HHD and isokinetic dynamometer in the seated position in the Sedentary Group (r=0.79), Instability Group (r=0.80), and Athletes Group (r=0.76). All these correlations were significant (p<0.001) (Table 2).

For all individuals, the limits of agreement (LoA) between the HHD and isokinetic dynamometer in the seated position had a range of 281.51N to −194.89N with a bias of 43.31N (95% CI: −0.18N to 86.60N, p>0.05) (Fig. 2A). The LoA between the HHD and isokinetic dynamometer in the lying supine position was between 303.96N and −196.01N with a statistically significant bias of 53.98N (95% CI: 8.34N to 99.62N, p<0.05) (Fig. 2B). Although almost all individuals are in the 95% limits of agreement range of the Bland–Altman plots, the CIs are considered very wide, indicating that the measures tended to disagree. Furthermore, the mean difference between HHD and isokinetic dynamometer were biased positively and statistically significant just in the lying position of the HHD, i.e., measurements were overestimated in HHD evaluation.

Figure 2.

(A) Bland–Altman plot with the agreement between the isokinetic dynamometry versus HHD assessment in the seated position for all individuals. (B) Bland–Altman plot with the agreement between the isokinetic dynamometry versus HHD assessment lying supine position for all individuals.

(0.31MB).

After the verification of lower agreement between the HHD and isokinetic dynamometer in the lying supine position in all groups together and considering that the sitting position is closest to the position adopted in the isokinetic assessment, we opted to include the results of Bland–Altman plots for each group only in the seated position of the HHD.

The LoA between the HHD in the seated position and isokinetic dynamometer for the Instability Group had a range of 249.25N to −86.22N with a bias of 81.51N (95% CI: 50.88N to 112.13N, p<0.05) (Fig. 3A). The LoA for the Athletes Group in the same position was between 155.88N and −264.43N with a bias of −54.27N (95% CI: −92.63N to −15.90N, p<0.05) (Fig. 3B). Finally, the LoA for the Sedentary Group was between 318.10N and −127.70N with a bias of 95.20N (95% CI: 54.51N to 135.89N, p<0.05) (Fig. 3C).

Figure 3.

(A) Bland–Altman plot of the Instability Group measurements with the comparison between the isokinetic dynamometry versus Seated HHD. (B) Bland–Altman plot of the Athletes Group measurements with the comparison between the isokinetic dynamometry versus Seated HHD. (C) Bland–Altman plot of Sedentary Group measurements with comparison between the isokinetic dynamometry versus Seated HHD.

(0.32MB).

Almost all individuals are in the 95% limits of agreement range of the Bland–Altman plots of each study group, but the CIs are also considered very wide, indicating that the measures tended to disagree. The mean difference between HHD and isokinetic strength were biased positively (p<0.05) for the Instability and Sedentary groups, indicating that HHD measurements were higher compared to isokinetic dynamometer measurements. In the Athletes Group, the mean difference between HHD and isokinetic strength were negatively biased (p<0.05), indicating that HHD measurements were lower compared to isokinetic dynamometer measurements.