Predictors of falls in patients during the first year after total hip arthroplasty: A prospective cohort study

Abstract Background and Purpose Since falls after total hip arthroplasty (THA) cause severe complications such as dislocation and fractures around the femoral stem, it is important to investigate what factors predict of falls. Thus, investigating predictors of falls in patients waiting for THA would be valuable as it lead to more strategic interventions to prevent these problems. The purpose of this study was to evaluate the predictors of falls in patients during the first year after THA. Methods This is a prospective cohort study. A total of 157 patients who underwent THA for unilateral hip osteoarthritis were analyzed. The incidence of falls during the first year after THA was monitored, and patients were classified into a “faller” and “non‐faller” group. The following factors were compared between the two groups: demographic data (age, sex, body mass index, leg length discrepancy, length of hospital stay, and history of falling), preoperative hip abductor muscle strength, functional performance (single leg stance and maximum walking speed), pain during walking, and physical activity. Results On multivariate logistic regression analysis, preoperative hip abductor muscle strength on the affected side and a history of falling were predictors of falls during the first year after THA. On subsequent receiver operating characteristic curve analysis, preoperative hip abductor muscle strength on the affected side was retained as a significant predictor, with a cut‐off strength of 0.46 Nm/kg differentiating the faller and non‐faller groups with a specificity of 73.6%, specificity of 50.0%, and area under the curve of 70.2%. Conclusions Finding from the present study suggested that clinicians should focus on low preoperative hip abductor muscle strength on the affected side and a history of falling to prevent falls during the first year after THA.


| Study design
This is a prospective cohort study. We followed the strengthening the reporting of observational studies in epidemiology statement for reporting a cohort study.

| Patients
Patients were recruited at our institute, between May 1 and (c) complications during or after surgery; and (d) presence of a severe leg length discrepancy (>3 cm) before surgery.
All patients were treated with a primary THA using an anterolateral approach under general anesthesia combined with epidural anesthesia. A standardized multimodal protocol was used for postoperative analgesia. Intravenous acetaminophen (1000 mg) was first administered at the time of wound closure and then every 6 hours during the first 24 hours after surgery. Oral acetaminophen (1000 mg, three times a day) was then started. Celecoxib (400 mg) was given during the morning on post-operative day 1; then, 200 mg was administered twice a day.
Postoperative rehabilitation was performed using a clinical path and booklet. Full weight-bearing was allowed from the day of the surgery, with use of a walker initially, transitioning to the use of a cane on postoperative day 1 or 2. All patients were able to walk independently with a cane, and were discharged home within 5 days of the surgery.
After discharge, they underwent regular follow-up examination at 2, 6 months, and 1 year after surgery. At the time of the examinations, they were performed exercise instruction by a physical therapist for about 20 to 40 minutes. These exercises were performed open kinetic chain exercises (hip extension, external rotations, and abduction) aimed at improving the range of hip motion, increasing around these hip muscle strength. At the end of each intervention, the physical therapists instructed them to continue appropriate exercise at home. A specific rehabilitation program to prevent falls was not provided.

| Ethical considerations
This study was approved by the institutional review board of the authors' affiliated institutions (TGE 00996-115). We obtained verbal and written informed consent from all participants before enrollment.
This study was conducted in accordance with the Helsinki Declaration.

| Outcome measures
Demographic data of patients (age, sex, body mass index [BMI], leg length discrepancy, length of hospital stay, and history of falling) were recorded using a preliminary form. Three physiotherapists were separately responsible for each patient. Hip abductor muscle strength, functional performance (one leg stance time and maximum walking speed), hip pain during walking, and physical activity were evaluated at 1 month before surgery. Additionally, we assessed the incidence, circumstances, injury duration, and frequency of falls, prospectively, during the first year after surgery.

| Fall assessment
All patients completed a questionnaire recording the incidence of falls.
Falls were defined as "a person falling onto the same level or a lower level on their own, without external force from other person, loss of consciousness, paralysis from a sudden stroke, or an epileptic seizure." 17 Additionally, the circumstances of the falls was also recorded, including the location (indoors, outdoors, or on stairs), time of day (morning, daytime, or nighttime), cause (tripping, slipping, or loss of balance), injury sustained (none, wound or bruise, or fracture), postoperative duration (less than 6 months, or ≧6 months), and frequency of falls (1 time or ≧2 times) during the first year after surgery. 10

| Hip abductor muscle strength
Hip abductor muscle strength was evaluated using a hand-held dynamometer (MicroFET2, Hoggan Health Industries, Salt Lake City), with the patient in a supine position. The dynamometer was placed lateral to the fibula (2.5 cm proximal to the malleolus). Three trials of maximum effort were performed, and the highest value was used for the analysis. The torque-to-body weight ratio of abductor muscle strength (Nm/kg) was calculated from the body weight and spina-malleolar distance. 18 Good interrater and test-retest reliability of handheld dynamometer measurements have been verified in previous studies in healthy adults 19 and patients after THA. 20

| Hip pain during walking
Hip pain during walking was evaluated using a 100-mm visual analog scale score. 24

| Physical activity
Patients were asked the mean number of days and hours of physical activity they performed in 1 week, using the International Physical Activity Questionnaire. 25 The intensity of physical activity was classified as follows: high-intensity exercise, 8 metabolic equivalents (Mets); moderate intensity, 4 Mets; and walking, quantified as 3.3 Mets. The activity intensity was converted to a respiratory quotient, and the number of calories consumed in activity per week was calculated from the respiratory quotient and body weight. In addition, based on the report by Brach et al, 26 patients were classified into a high or low physical activity group (high, burning ≥1000 kcal, or low, burning <1000 kcal per week).

| Sample size calculation
Based on a priori power analysis, the minimal sample size for the multivariate logistic regression analysis to examine significant factors (α = .05, power = .95, effect size = 0.3, potential predictor variables = 3) was calculated, assuming a 25% to 30% fall rate after THA. Hence, a sample of approximately 150 participants was required. To account for potential drop out (10%-15%) over the 12-month follow-up period, a sample of 170 participants was needed.

| Statistical analyses
The incidence rate of falling and fall-related injuries was calculated.
Initially, the following variables assessed before surgery were analyzed and compared between fallers and non-fallers using Student's t tests or chi-squared (χ 2 ) tests, as appropriate. Then, we performed a multivariate logistic regression analysis to assess the preoperative factors predicting falls during the first year after THA. For significant preoperative factors identified, a receiver operating characteristic (ROC) curve was constructed to determine their accuracy in distinguishing between fallers and non-fallers, where accuracy was evaluated by the area under the curve, and the cut-off value determined by the highest sum of sensitivity and specificity was used.
All statistical analyses were performed using SPSS Version 24 (SPSS, IBM, Inc, Chicago), and a P value < .05 was considered significant.

| RESULTS
Patient selection and the flow of the study are shown in Figure 1.
After excluding four patients who did not return for regular follow-up examination, 157 patients (15 men and 142 women, mean age 63.9 ± 9.9 years) were included in analyses (follow-up rate, 97.5%). None of these patients developed a postoperative infection or dislocation, and none required revision arthroplasty.
Comparison of demographic characteristics, hip abductor muscle strength, functional outcomes, and physical activity among fallers and non-fallers are reported in Tables 1 and 2. The incidence of at least one fall occurred in 32 patients (20.4%). There was no significant difference in age, sex, BMI, leg length discrepancy, or length of hospital stay, between the two groups. Compared to non-fallers, however, fallers had lower hip abductor muscle strength on the affected side (0.45 ± 0.17 vs 0.60 ± 0.22 Nm/kg, respectively; P < .001) and on the unaffected side (0.64 ± 0.22 vs 0.77 ± 0.26 Nm/kg, respectively; P = .008), and a more prevalent prior history of falling, with 37.5% of patients in the fallers group reporting a prior fall, compared to 17.6% in the non-fallers group (P = .017).
The results of the multivariate logistic regression analysis are summarized in Table 3. Preoperative hip abductor muscle strength on the affected side and a prior history of falling were independent predictors of a fall during the first year after THA. Figure 2 shows the ROC curves constructed to determine the optimal cut-off value for preoperative hip abductor muscle strength on the affected side that F I G U R E 1 Flowchart of patients throughout the study This underlines the potential for a preoperative history of falling to induce a debilitating downward spiral, marked by loss of confidence, prolonged functional recovery, and an increased risk of falling after THA. It is important to further note that a preoperative history of falling is not only a risk factor for falling per se after THA, but also a factor of functional recovery after THA.
Finding from the present study suggested that surgeons and physical therapists should focus on low preoperative hip abductor