The study protocol was approved by the Ethics Committee of Kitasato University Hospital (permit number: B18–088), and was performed in accordance with the Declaration of Helsinki. Informed consent was obtained from all individual participants included in the study.
This was a retrospective study. Primary and unilateral THA were performed in 336 patients at our hospital between November 2015 and March 2018. After excluding 11 patients who were unable to walk before surgery and 8 patients with perioperative complications including postoperative cerebral infarction (n = 1), load limitation due to bone vulnerability (n = 2), fractures during surgery (n = 2), and those with deep venous thrombosis that needed treatment early after surgery (n = 3), 317 patients were included in this study.
Patient characteristics, including age, gender, BMI, comorbidities, preoperative functional assessment, length of hospital stay, and discharge, were collected from electronic medical records. In addition, time taken to achieve walking over a distance of 50 m with or without a walking aid and the proportion of patients achieving unassisted walking on postoperative day 5 were recorded as outcome measures. Functional recovery was evaluated daily and defined as the ability to walk over a distance of 50 m without human assistance, regardless of the use of a walking aid.
Preoperative functional assessment
Hip pain, knee extensor and hip abductor muscle strength, hip flexion ROM, and gait speed were assessed 1–3 days before operation by one of seven physical therapists who had received training for more than one month. Hip pain on the affected side during walking was assessed using a visual analog scale. Muscle strength was determined by measuring maximum voluntary isometric knee extensor and hip abductor strength on the affected side using a hand-held dynamometer with a restraining belt  (μTas; ANIMA, Tokyo, Japan). Measurements were obtained twice, with the highest values expressed as relative to body weight (%BW) used in the analysis. Hip flexion on the affected side was used to evaluate ROM. Usual and maximal gait speeds were measured by timing the patients walking at their usual or maximal pace with any necessary assistive devices over the middle 10 m of a 16-m walkway. A digital stopwatch was used to time subjects as they walked over a 10-m distance. Subjects were provided with 3 m to accelerate and decelerate before and after the test distance. For the usual speed walking trials, they were instructed to walk at their normal comfortable speed. For the maximal speed walking trials, they were asked to walk as fast as they could safely without running. First, usual gait speed was measured, and then, maximal speed was measured. Between each measurement, 30-s rest was taken.
A rehabilitation program is shown in Appendix. A standard rehabilitation program comprising weight bearing, as tolerated with a walking aid, was started on the day after surgery, and patients were allowed to eliminate walking aids whenever comfortable. Physical therapy was performed once a day on weekdays, and it included progressively improving walking ability, other functional activities, and walking stairs according to the needs and progress of an individual patient. Patients participated in a progressive program involving range of motion exercises, strengthening exercises, and functional training. Patients were allowed to use analgesics for pain as needed. Patients were discharged from the hospital on consultation with their caregivers, surgeons, nurses, and physical therapists, according to the following criteria: ability to walk independently with a walking aid, and if necessary, climb stairs, remain in a stable medical condition, and exhibit adequate wound healing. Discharge disposition in this study was defined as a rehabilitation facility or home, and the patients were transferred to a rehabilitation facility if they could not achieve the criteria at 1 week after surgery.
Cox regression analysis constructing two predictive models was used to determine the prognostic capabilities of gait speed for predicting time to functional recovery: Model 1, age + sex; Model 2, Model 1 + hip and knee muscle strength + ROM.
We constructed receiver operating characteristic (ROC) curves to compare the predictive ability of usual and maximal gait speed for functional recovery on postoperative day 5. The areas under the curves (AUCs) of ROC curves were compared according to the method of DeLong et al.  We also constructed ROC curves for functional recovery at 5 days after the operation using two models: gait speed only and clinical model including age, sex, hip and knee muscle strength, and ROM.
The Kaplan–Meier method was used to evaluate functional recovery and comparisons were performed by the log-rank test. Patients were divided into 3 groups based on tertile of gait speed. Subgroup analyses of gait speed were performed to examine by means of Cox regression analyses with adjustment for age and sex as potential confounders, including age stratified at < 65 years, 65–74 years, and ≥ 75 years, and sex. Hazard ratios (HRs) are reported with corresponding 95% confidence intervals (CIs).
Data are presented as the means ± standard deviation or as percentages. Statistical analyses were performed using SPSS version 22.0 (IBM Corporation, Armonk, NY), STATA version 13.0 (StataCorp, College Station, TX), and R version 3.1.2 (The R Foundation for Statistical Computing, Vienna, Austria). In all analyses, P < 0.05 was taken to indicate statistical significance.