Improvement in mental health following total hip arthroplasty: the role of pain and function

As part of a prospective THA cohort that began in 1996 at a large public hospital in Switzerland, data before and after surgery were systematically collected on all THAs performed at the institution [29]. For this current longitudinal study, all elective primary THAs (and no further contra-lateral hip arthroplasty during the follow-up year) operated at the Orthopedic Department between January 1 and December 31, 2010, and between January 1, 2012 and July 31, 2014 were eligible. Data from 2011 were not included because preoperative questionnaires had not been sent out routinely during that year. All eligible THA patients (n = 1045) received questionnaires, which were sent between 10 and 14 days prior to surgery. Of those, 848 questionnaires were returned (81.1%). One year after surgery follow-up questionnaires were sent to all eligible patients, to which 785 (75.1%) responded. Overall, 636 (60.9%) of the eligible patients with THA responded to both the preoperative and the 1-year postoperative questionnaire, with 610 people having data on mental health status and pain or function data and were included in this study.

At baseline and 1 year after surgery, patients completed patient-reported outcomes using questionnaires. Mental health status was assessed using the mental component score (MCS) of the Medical Outcomes Study Short Form-12 (SF-12) [17], which is a generic health-related quality of life measure. Pain and function were assessed using the reduced form of the Western Ontario McMaster Universities (WOMAC) [30], which is a disease-specific instrument for the assessment of osteoarthritis of the hip and knee. The WOMAC pain and function scales were normalized to a range of 0 (lowest possible score) to 100 (highest possible score), with an increasing score indicating better outcome.

The main outcomes of interest were the MCS at baseline, 1 year after THA, and 1-year change. The MCS ranges from 0 to 100, with higher scores indicating better outcomes. We calculated the change as the absolute difference in MCS scores between baseline and 1 year after THA. The mean ± SD population value in this geographic study area [31] was 46.3 ± 10.1.

The main predictors of interest included the WOMAC pain and function scores at baseline, 1 year after THA, and the 1-year change. We calculated differences in WOMAC pain and function scores as the absolute difference between baseline and 1 year post-op pain and function scores, respectively. We took into consideration participants’ age, sex, body mass index (BMI: < 25, 25–29.9, 30–34.9, 35.0+), education level (< 9, 9–12, > 12 years of education), insurance status (private or public), smoking status (ever or never), medical co-morbidities such as diabetes (yes or no), the American Society of Anesthesiologists (ASA) score (1 = normal healthy patient, 2 = patient with mild systemic disease, 3 = patient with severe systemic disease, or 4 = patient with severe systemic disease that is a constant threat to life) [32], medications used including antidepressants or opioids, Charnley disability grade (A = involving one hip, B = involving both hips, or C = multiple joints or other disabilities leading to difficulties in ambulation) [33], and reason for THA (primary vs. secondary OA, the latter including dysplasia, inflammatory arthritis, aseptic necrosis or post-traumatic origin).

Preoperatively, the questionnaire was sent out to all patients undergoing elective THA approximately 10 to 14 days prior to surgery. The follow-up questionnaire was sent out 1 year after surgery to all patients still alive. For patients who did not return their 1-year questionnaires, another follow-up questionnaire was sent about 3 months after the first mailing. Information on the baseline characteristics including age, sex, and insurance status was recorded at the time of admission. Reason for OA and Charnley disability grades were recorded on a pre-specified form by the operating surgeon. Medical co-morbidities, medication use at the time of admission, BMI, ASA score and smoking status were obtained from the anaesthesia report and discharge summary. Information on education level was obtained from the patient via the preoperative questionnaire.

Regarding the first aim (to examine the change in mental health from before to 1 year after surgery and to identify variables associated with improvement), we calculated means and standard deviations (SD) for baseline, 1 year after surgery, and 1-year change in MCS scores overall, and by subgroups defined by age, sex, BMI, education, insurance, smoking status, ASA scores, diabetes, antidepressant or opioid use, Charnley scores, and OA status. To show the magnitude of the overall 1-year difference, we estimated Cohen’s effect size where 0.2, 0.5, and 0.8 were considered respectively as small, medium, and large differences between baseline and 1 year after THA [34]. We also plotted the distributions of the baseline and 1 year after THA MCS scores, using kernel density plots.

Regarding the second aim (to examine the association between change in mental health and change in pain and function over time), we calculated means and standard deviation (SD) for baseline, 1-year, and 1-year change in WOMAC pain and function scores. We then examined 1-year change in MCS by quartiles of 1-year changes in pain and function. We also used 2 separate linear regression models to predict 1-year change in MCS as the main outcome of interest, one model with 1-year change in pain and the other with 1-year change in function as the main predictor of interest. We performed both unadjusted linear regression, and adjusted for education, age, BMI, sex, smoking status (ever vs. never), insurance status (private vs. public), diabetes, ASA score (1 vs. 2+), and Charnley score (C vs. A and B), OA status (primary vs. secondary). By visual inspection, the distributions of regression residuals of these models were reasonably bell-shaped. Finally, as education level was an important covariate and approximately one fourth of our participants were missing education data, we further performed sensitivity analyses using a simultaneous multiple imputation for the education level. In brief, we entered education, age, and BMI as continuous variables in addition to sex, insurance, tobacco, ASA, diabetes, hypertension, and Charnley score into the model with 1-year change in pain or function predicting 1-year change in MCS score to perform multiple imputation of missing data for multivariable adjusted linear regression. We used IBM Windows SPSS V.22 (IBM Corp., Armonk, NY, USA) for all statistical analyses.