Participants
Participants for the THA group and the control group were recruited through advertisements and articles in local newspapers. The THA group comprised patients who had undergone unilateral THA four to five years ago. Further inclusion criteria were an age between 50 and 70 years as well as being physically active at least two times a week. Healthy, age-matched and physically active participants were included in the control group. Exclusion criteria were neurological or cardiovascular diseases and acute injuries of the musculoskeletal system. Healthy controls were also excluded if they had diagnosed osteoarthritis in lower extremity joints. All participants gave written consent to participate in this study after being informed about the procedure, its purpose and possible risks linked to the participation. The study was approved by the local ethics committee of the Otto-von-Guericke-University Magdeburg presided by Dr. med. Norbert Beck and carried out in line with the Declaration of Helsinki (no. of vote: 155/18 on December 3, 2018). It was retrospectively registered in the German Registry of Clinical Trials under the ID: DRKS00016945.
Measurement protocol
For this cross-sectional study, data collection was carried out between January 2019 and June 2019. The participants were asked to attend one testing, in which all measurements were conducted. The measurements consisted of examinations of isometric strength of the hip muscles, hip ROM, balance and gait. In the THA group and the control group, data were collected on both legs. Regarding the THA group, the legs were differentiated in the operated and the non-operated side.
Maximum isometric hip strength analysis and active hip ROM analysis
The examinations of the isometric strength of the hip muscles and of the active hip ROM were performed in a self-developed diagnostic machine (Fig. 1). The pelvis support helps the patients to maintain a fixed position and to avoid compensational movement during the measurements. The diagnostic machine contains a 270° rotatable baseplate, which enables the strength measurement of patients in different directions while the patients can just remain in their position.
The reliability of the isometric hip strength measurement as well as of the hip ROM measurement of the diagnostic machine had been investigated before the examinations. 24 healthy individuals participated in the reliability studies (Study 1 (hip ROM): n=10, 28.4±5.7 years; Study 2 (hip strength): n=14, 21.3±2.1 years). Reliability of the hip strength and ROM measurement were examined in a test-retest design with seven days between measurements. To examine the test-retest reliability, intra-class correlation coefficients (ICCs) were calculated based on a single-rating, absolute agreement and a two-way mixed effects model [20]. The ICCs showed values ranging between 0.85 and 0.95 for the isometric hip strength measurement (hip flexion, extension, abduction, and adduction) and values ranging from 0.65 to 0.84 for the hip ROM measurement (hip flexion, extension and abduction). According to Koo and Li, these ICCs indicate good to excellent reliability [20] with one exception for the measurement of hip flexion (ICC=0.65), which is interpreted as moderate reliability. The results suggest that the diagnostic machine provides an environment to reliably quantify maximum isometric hip strength and active hip ROM.
For the examination of the maximum isometric strength of the hip muscles, subjects were instructed to stand in an upright position in the diagnostic machine. The pelvis support helped the participants to remain in this position. A neoprene brace was placed distally at the thigh as an attachment possibility for the hauling rope. An integrated force transducer (Hottinger Baldwin Messtechnik GmbH, Darmstadt, Germany) measured the isometric strength in the respective pulling directions hip flexion, extension, abduction and adduction in the neutral hip position (Fig. 3).
For each motion direction, one pretest and two main tests were performed. Subjects were instructed to build up strength and contract maximally without an abrupt push. A resting period of one minute between each trial was maintained. Force data from the strength analysis were acquired at 1000 Hz and filtered in Matlab with a 4th order Butterworth low-pass filter (5 Hz). Out of the two main trials, the trial with the highest torque was normalized to the body mass of the participants and used for further analyses. The distance between the greater trochanter and the point of applied force (the middle of the neoprene brace) served as the lever arm (Fig. 3).
For the examination of the hip ROM, subjects were also standing in the diagnostic machine fixated right above the pelvis in order to avoid compensational movements with the upper body but still providing free movement of the hip joint. Active ROM of the hip was measured in flexion, extension and abduction in a standing position. Adduction was excluded due to potential risk of luxation of the prosthesis. The angles of the three movement directions were quantified with an acceleration sensor (PLUX-Wireless Biosignals S.A, Lisbon, Portugal) placed distally on the lateral side of the thigh. After initializing the sensor in the neutral zero position, participants were instructed to slowly perform three maximal hip flexion movements followed by three maximal extension and abduction movements (Fig. 2). Particular attention was paid to the participants to not modify their upper body position and to cleanly execute the motion (hip abduction, flexion, extension) in the respective motion axis. Data from the motion analysis were acquired at 1000 Hz and filtered in Matlab with a 4th order Butterworth low-pass filter (5 Hz). Out of the three trials, the maximum hip angles in flexion, extension and adduction on each side were extracted for further analyses.
Balance assessment
Static balance was assessed in the bipedal and single-leg stance using a force plate (PLUX-Wireless Biosignals S.A, Lisbon, Portugal). For the bipedal stance, subjects were asked to take off shoes and stand with both legs, hip width apart, on the force plate with the arms hanging down at the sides. Two trials with a duration of thirty seconds were recorded. For the single-leg stance, the participants were instructed to position one leg in the center of the force plate, slightly lifting off the other foot and fixating the wall in front of them. Before collecting data, the participants were asked to practice this posture. Two trials on each leg were captured with a duration of ten seconds. The acquisition time in the single-leg stance was limited to ten seconds as most subjects were not able to hold the position for much longer. Balance data were sampled at 250 Hz and further processed using Matlab (Version 2018b, The Math-Works Inc., Natick, MA). The dataset was filtered applying a 4th order Butterworth low-pass filter with a 10 Hz cut-off frequency. The total length of the center of pressure (COP) during bipedal and single-leg stance was computed as well as the standard deviations (SD) of the COPx and COPy for mediolateral (ML) and anteroposterior (AP) directions [21]. The best trials of each leg were chosen for further analyses.
Gait analysis
The gait analysis was performed with InvestiGAIT, an inertial sensor-based system consisting of four Shimmer3 sensors (Shimmer, Realtime Technologies Ltd, Dublin, Ireland) and an in-house Matlab program for recording and analyzing gait data. Two of the inertial sensors were laterally placed above each ankle. In order to quantify the movement of the hip and the upper body, the third and the fourth Shimmer sensors were centered at the height of the posterior superior iliac spine and at the thoracic vertebra II [22]. The subjects were asked to walk a predefined distance (12.5 m) marked by two pylons at their self-selected, comfortable walking speed. For each participant, twelve gait sequences were recorded. Outcome spatiotemporal gait parameters involved step length, stance and swing duration as well as one-leg-stance as a percentage of the gait cycle. These parameters can also be used for inter-limb examinations as they are calculated for both legs (affected/non-affected) and therefore provide information about gait symmetry or asymmetry [23, 24].
The gait parameters of the InvestiGAIT system are calculated based on the identification of gait events including initial contact (IC), midswing point and terminal contact (TC). These events are detected as local minima (IC, TC) or local maxima (midswing) in the signals of the z-axis of the ankle gyroscopes, which describe the angular velocities of the shanks in the sagittal plane. More detailed information about the detection of gait events and calculation of the gait parameters of the InvestiGAIT system are provided in Orlowski and Loose [25] and Orlowski et al. [26]. The InvestiGAIT system has been confirmed to be a valid and reliable system to investigate human gait in a clinical setting [22, 26].
Statistical Analyses
All analyses were performed using SPSS 25 (SPSS Inc., Chicago, IL) with a significance level set to p <0.05. The data were tested for normal distribution applying the Shapiro-Wilk test. To investigate potential differences between the operated and non-operated side of the THA group, a paired t-test was applied for each parameter. In case of violation of normal distribution, the nonparametric Wilcoxon test was used. For group comparisons, the demographic variables of the THA group and the control group were verified for significant differences applying the unpaired t-test, Mann-Whitney U and Chi-squared test. As an age difference- although not significant- was observed, analyses of covariance (ANCOVA) were applied to assess group differences in each parameter, using age as a covariate. The data of the operated side of the THA group were compared to the averaged data of the right and left leg of the control group. For intrasubjective comparisons, effect sizes were calculated using Cohen’s dz for within-subjects designs [27]. In case of non-normally distributed data, the effect sizes were determined with the correlation coefficient r. For group comparisons, effect sizes were calculated applying Cohen’s ds for in between-subjects designs [27]. Values for d=0.2 were interpreted as a small, d=0.5 as a medium and d=0.8 as a large effect. Effect sizes for r were interpreted as small (r=0.1), medium (r=0.3) and large (r=0.5) [28].