Subjects
Subjects were female patients with hip OA who had undergone primary unilateral THA at two hospitals (Ebina General Hospital and Zama General Hospital) between December 2015 and September 2017, and who agreed with the purpose of the study. Exclusion criteria were (1) a history of lower limb surgery, e.g., valgus osteotomy and rotational acetabular osteotomy or spine surgery; (2) previously diagnosed painful orthopedic disease other than hip joint disease; (3) postoperative complications such as fracture, dislocation, infection, or nerve paralysis; (4) previously diagnosed mental disease or neuromuscular disease; and (5) MJS ≤0.5 mm as measured on hip X-ray images 11. To minimize the effects of sex differences in lifestyle behaviors, which may affect daily step count and MVPA 27, only female patients were included in this study, as roughly 90% of patients in the two hospitals are female. Four highly experienced surgeons performed all THAs without a navigation system through a minimally invasive anterolateral approach independent of the implant type, with patients placed in the half lateral position (i.e., pelvis tilted 60 degrees relative to the floor).
This study complied with the ethical standards of the Declaration of Helsinki (1964) and its subsequent amendments, and was approved by the Ethics Review Committee of our institution. All patients provided written informed consent to participate in this study.
Procedures
This prospective cohort study was conducted over a period of 18 months from six to 24 months after THA. The primary outcome was change in MJS (ΔMJS: MJS at 24 months - MJS at six months). The baseline was set as six months after THA, as this corresponded to the time at which about 90% of healthy subjects showed improvement in the number of daily steps in a previous study 4. Baseline measurements included background factors, radiographic assessment, pain assessment using a visual analog scale (VAS), and functional measures obtained before and six months after THA. PA (daily step count and MVPA) was also measured six months after THA. ΔMJS, the primary outcome, was calculated at 24 months, i.e., the end of the observation period.
Postoperative rehabilitation was performed according to the clinical pathway of the two hospitals. On the day after surgery, all patients were allowed to bear full weight and underwent inpatient rehabilitation including gait exercises, passive range of motion (ROM) exercises, and muscle strengthening exercises. The length of hospital stay according to the clinical pathway was three weeks. In addition to inpatient rehabilitation, most patients underwent outpatient rehabilitation roughly once a week after discharge for five months.
Measurements
Background information including age, height, body weight, and body mass index was collected from medical records. Functional measures including the Japanese Orthopaedic Association hip score (JOA score), hip flexion ROM, hip abductor muscle strength, and 10-m gait speed were obtained by two physical therapists who had thoroughly practiced measurement procedures. Hip abductor muscle strength was measured using a hand-held dynamometer (μTas F-1; Anima Corp., Tokyo, Japan) by referring to the previously proposed method 28. Patients were positioned on a platform in a supine position at a 0° angle, with a sensor pad attached to the distal lateral side of the thigh. After practice, muscle strength during 5 seconds of isometric contraction was measured twice, and the higher value was used for analysis. The length of the lever arm was measured from the greater trochanter to the center of the sensor. Muscle strength was calculated as torque per body weight (Nm/kg). For 10-m gait speed, patients were instructed to walk as fast as possible, and the higher value of two measurements was used for analysis.
For PA, a digital pedometer with 3-axis acceleration sensors (TH-400; YAMASA, Tokyo, Japan) was used to measure daily step count (steps/day) and MVPA (min/day). The validity and reliability of another device by the same manufacturer (EX-510; YAMAX, Tokyo, Japan), which uses the same algorithm, has been demonstrated previously 29. MVPA was measured at a cadence of ≥120 (steps/minute). There is a strong correlation between walking cadence and exercise intensity 30 and a cadence of ≥120 (steps/min) corresponds to moderate or higher activity intensity (≥3-4 METs) 31. Patients were instructed to wear the device on their body for 24 hours a day except when bathing or sleeping (sleep time was set at 8 hours), for one to two months after THA. Non-wearing time (in hours), if any, was recorded in patient reports. If non-wearing time was ≥4 hours, the day was treated as a missing value 32. The average value, excluding missing values, for two weeks from five to six months after THA was calculated.
Radiographic assessment
An anteroposterior pelvic radiograph in the supine position was obtained by skilled radiology technicians. A single well-experienced examiner measured MJS, center edge (CE) angle, sharp angle, acetabular head index (AHI), leg length discrepancy (LLD), and KL grade using software (SYNAPSE Enterprise-PACS ver. 2.4.1, Fujifilm, Japan). We referred to the measuring method proposed previously in two studies 11, 24. The minimal detectable change of ΔMJS using this method is 0.5 mm 11, 24. MJS was measured at three locations: (1) at the lateral margin of the subchondral sclerotic line, (2) at the apical transection of the weight-bearing surface by a vertical line through the centre of the femoral head, and (3) at the medial margin of the weight-bearing surface bordering on the fovea, or as a fourth measurement if the minimum MJS was found outside the three standard locations of measurement. MJS was selected as the smallest of these three measurements.
Statistical analysis
All values are presented as mean (standard deviation) or median (1st quartile to 3rd quartile). ΔMJS was calculated as the primary outcome of this study; an increase in MJS (i.e., positive ΔMJS) indicates that OA has progressed, whereas a decrease in MJS (i.e., negative ΔMJS) indicates that OA has improved. In order to clarify the presence or absence of progression of OA over the course of 1.5 years, the number of patients with ΔMJS ≥0.5 mm and the proportion of these patients among all patients were calculated. In addition, annual ΔMJS (ΔMJS / 1.5) (mm / yr) was calculated to allow for comparisons with ΔMJS results from previous studies. Patients were divided into two groups according to KL grade: 0 (KL 0 group) or ≥1 (KL ≥1 group). The t-test or Mann-Whitney U test was used to compare background factors, radiographic assessment, pain assessment, functional measures, and PA between the two groups. When there was a significant difference in any background factor, a linear model was used to adjust for that factor. The generalized linear model (GLM) was used to clarify the influence of PA on ΔMJS. GLM is an extension of the linear model, and the probability distribution of any exponential family can be selected for the error structure. In this study, we selected normal and gamma distributions. The link function was selected for the inverse function {f(x) = 1/(a+x), a = fixed number}. First, we created a model in which the objective variable was ΔMJS, with PA (either daily step count or MVPA) as the only explanatory variable (Step 1: univariate analysis). Second, the forced entry method was used to create a model in which adjustment factors (age, body weight, and MJS at six months), KL grade, and PA (either daily step count or MVPA) were used as explanatory variables (Step 2: multivariate analysis). A dummy variable was assigned to KL grade. Third, interaction terms between PA (either daily step count or MVPA) and KL grade were included in explanatory variables along with adjustment factors (Step 3: multivariate analysis with interaction terms). If the interaction term was significant, and residual deviations as calculated by the χ2 test in Steps 2 and 3 significantly decreased, a simple slope analysis using KL grade (0 or ≥1) was performed. Finally, using the forced entry method, a model with a significant interaction term, PA (both daily step count and MVPA), KL grade, and adjustment factors was created (Step 4). The model with the lowest Akaike’s Information Criterion (AIC), a selection criterion, was used as the final prediction model. In addition, residual analysis was performed using quantile regression. The sample size was calculated using the calculation method for the linear model (α = 0.05, detection power = 0.95, effect size = 0.3, 7 input explanatory factors), yielding a required sample size of 81. Since radiographic assessment at 24 months after THA was performed as part of periodic health examinations, we estimated that there were few dropouts (about 10%). Therefore, we selected 90 patients for baseline measurements at six months after THA. R version 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria) was used for all statistical analyses.