Association of the 36-Item Short Form Health Survey Physical Component Summary Score With Patient Satisfaction and Improvement 2 Years After Total Knee Arthroplasty

IMPORTANCE Increases in total knee arthroplasty (TKA) utilization rates suggest that its indications have been expanded to include patients with less severe symptoms. A recent study challenged the cost-effectiveness of TKA in this group of patients. OBJECTIVE To determine the association of the 36-Item Short Form Health Survey physical component summary score (SF-36 PCS) with patient satisfaction 2 years after TKA. DESIGN, SETTING, AND PARTICIPANTS This cohort study reviewed registry data from 2 years of follow-up of patients who underwent unilateral TKA from January 1, 2010, to December 31, 2014, at a single-center tertiary institution in Singapore. Data were acquired on April 27, 2017, and analyzed from August 15, 2017, to December 22, 2017. MAIN OUTCOMES AND MEASURES Patient satisfaction and SF-36 PCS. Preoperative disability and postoperativefunctionasmeasuredbytheSF-36PCSwerecorrelatedwithΔ(2-yearendpointscore minus baseline score) and patient satisfaction, scored on a 6-point Likert scale, with lower scores indicating greater satisfaction. from August 15 to December 22, 2017. Patient demographics, side of surgery, and individual body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) were established from records retrieved. Race/ethnicity was assessed because it has been shown to have an effect on both preoperative scores and functional outcomes after primary TKA even after adjusting for confounders. We classified race/ethnicity according to self-reports by the patients. 5-8 Preoperative scoring was determined at the preoperative evaluation to provide an accurate benchmark for comparison. We excluded patients with secondary arthritis from posttraumatic, inflammatory, or infective causes and those with postoperative prosthetic joint infections or fractures. Patients with these conditions are known to have a different prognosis from those with primary osteoarthritis. Different considerations (presence of long-term systemic medications) and surgical techniques are also needed to account for anatomical and physiological differences. 9,10 This study was conducted with approval from the Singhealth Centralized Institutional Review Board, which waived informed consent because the study was a retrospective review of a large sample with no effect on or interaction with patient treatment. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. The SF-36 for preoperatively and at the 2-year follow-up. The 2-year mark was chosen as the end point it approximates the time that early potential plateaus with concurrent reduction in the likelihood of the occurrence of an musculoskeletal event that could affect SF-36 scores at a longer follow-up. The SF-36 a generic 36-item, patient-reported survey of health used to determine care outcomes in adult patients. 11 The survey contains 8 domains: vitality; physical functioning; bodily pain; general health


Introduction
Osteoarthritis is reported to be the 11th greatest contributor to disability globally and the 38th greatest in disability-adjusted life-years. This demand on health services is set to increase with the aging and increasing obesity of the world's population. 1 For patients with refractory knee osteoarthritis, total knee arthroplasty (TKA) remains an effective treatment with predictable outcomes compared with other treatment modalities. 2 Losina et al 3 reported that the increase in TKA utilization rates in the United States was not fully explained by population growth and obesity and suggested that the rapid increase among younger patients may be a result of expanding indications for this procedure.
In a recent publication, Ferket et al 4 challenged the cost-effectiveness of TKA in this group of younger patients with less severe symptoms. By simulating a model, they reported that providing total knee replacement to patients with 12-Item Short Form Health Survey (SF-12) physical component summary (PCS) scores less than 35 was the optimal scenario, given a cost-effectiveness threshold of $200 000 per quality-adjusted life-years (QALYs) with a cost savings of $6974 and a minimal loss of 0.008 QALYs compared with the current practice. Restricting TKA to more severely affected patients (ie, those with lower preoperative scores) enabled the procedure to be more costeffective than its current use.
Cost-effectiveness based solely on improvement in outcome scores may not adequately take into account the expectations and needs of the individual patient. Our study aimed to evaluate the association of the 36-Item Short Form Health Survey (SF-36) PCS with patient satisfaction after TKA by examining the association of preoperative disability and postoperative function as measured by the SF-36 PCS with its Δ (2-year end point score minus baseline score) and patient satisfaction scores.

Methods
Prospectively collected institutional registry data were reviewed for all patients who underwent primary unilateral TKA for Kellgren-Lawrence grade 3 to 4 knee degenerate osteoarthritis at a single tertiary institution, Singapore General Hospital, from January 1, 2010, to December 31, 2014. On April 27, 2017, data from hospital electronic medical records were retrieved. Data analysis was conducted from August 15 to December 22, 2017. Patient demographics, side of surgery, and individual body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) were established from records retrieved. Race/ethnicity was assessed because it has been shown to have an effect on both preoperative scores and functional outcomes after primary TKA even after adjusting for confounders. We classified race/ethnicity according to self-reports by the patients. [5][6][7][8] Preoperative scoring was determined at the preoperative evaluation to provide an accurate benchmark for comparison. We excluded patients with secondary arthritis from posttraumatic, inflammatory, or infective causes and those with postoperative prosthetic joint infections or fractures. Patients with these conditions are known to have a different prognosis from those with primary osteoarthritis. Different considerations (presence of long-term systemic medications) and surgical techniques are also needed to account for anatomical and physiological differences. 9,10 This study was conducted with approval from the Singhealth Centralized Institutional Review Board, which waived informed consent because the study was a retrospective review of a large sample with no effect on or interaction with patient treatment. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
The SF-36 scores for each patient were assessed preoperatively and at the 2-year follow-up. The 2-year mark was chosen as the end point because it approximates the time that early rehabilitation potential plateaus with concurrent reduction in the likelihood of the occurrence of an unrelated musculoskeletal event that could affect SF-36 scores at a longer follow-up. The SF-36 is a generic 36-item, patient-reported survey of health commonly used to determine care outcomes in adult patients. 11 The survey contains 8 domains: vitality; physical functioning; bodily pain; general health perceptions; physical functioning; emotional functioning; social functioning; and mental health.

JAMA Network Open | Orthopedics
Possible scores range from 0 to 100, with higher scores representing better health status. The SF-36 PCS was compiled from individual scores of the 8 domains using a formula validated for the study population. The SF-36 PCS was chosen as the primary variable because it has the advantage of lower inherent variability than the individual domain scores and allows elimination of both floor and ceiling effects. 12 The difference in the SF-36 PCS (ΔPCS) after TKA was calculated by comparing SF-36 PCS score at 2 years after surgery with its preoperative value. Patients with a ΔPCS meeting or exceeding the minimal clinically important difference (MCID) of 10 were considered to have a clinically significant improvement after surgery. 13 This cutoff was used to dichotomize patients into 2 groups: improvement and no apparent improvement. Based on the dichotomized MCID outcome as a response, a preoperative PCS cutoff was determined using receiver operating characteristic analysis as a diagnostic tool for predicting improvement or no improvement after surgery. Subgroup analysis was performed between patients with scores below and above the preoperative PCS cutoff to determine differences in patient characteristics and outcomes.
Patient satisfaction scored at the 2-year follow-up was recorded by an interviewer masked to previous scores. Satisfaction was scored on a 6-level Likert scale (1, excellent; 2, very good; 3, good; 4, fair; 5, poor; and 6, terrible). Adapted from question 53 of the North American Spine Society Questionnaire, the satisfaction question was specifically worded as, "How would you rate the overall results of your treatment for leg pain?" To reduce interviewer bias, this question was asked exactly as phrased in a neutral tone without further substantiation. A score of 4 or less was considered to be indicative of patient satisfaction.

Statistical Analysis
Two-year patient satisfaction least-squares means were compared among preoperative SF-36 PCS groups using analysis of covariance with adjustment for age, BMI, sex, race/ethnicity, and side of surgery. A significant omnibus F test was followed by post hoc pairwise comparisons. We calculated 95% CIs for the mean differences. Univariate logistic regression was used to assess the association between preoperative SF-36 PCS score and dichotomized MCID. Receiver operating curve analysis was performed to identify a statistically optimal diagnostic cutoff based on the Youden J statistic. For this analysis, logistic regression was used. Spearman correlation was calculated for SF-36 vs ΔSF-36.
Statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc). The level of statistical significance was taken to be 5% (2-sided test).

Results
During the study period from January  The boxplot of patient satisfaction against postoperative SF-36 PCS revealed a decreasing mean postoperative SF-36 PCS with poorer patient satisfaction (Figure). There were considerable variance and outliers within each satisfaction group, with significant numbers of patients with either low satisfaction despite high scores or high satisfaction despite low scores.

Discussion
Total knee arthroplasty is typically indicated as a salvage option for end-stage arthrosis with reasonably good and predictable outcomes. Beyond an aging Baby Boomer generation and associated conditions, such as obesity, in the Western world, technological advances contributed to the ease and availability of TKA in most developed countries. This access led to TKA being performed on patients with less severe symptoms, ultimately increasing the volume of TKA performed. 1,3 The inevitable increase in health care costs has led to a focus on value-based reimbursement as opposed to volume-based reimbursement. With that, measures of cost-effectiveness and patientreported outcomes have become important considerations in funding health care procedures.
However, health care costs differ among systems and countries. In a system with lower-cost access to health care, the cutoff for cost-effectiveness should correspondingly be higher. Patient expectations are also known to differ among cultural and racial/ethnic groups. 14-17 Thus, a cutoff determined by a certain patient demographic characteristic might not be translatable owing to cost and population differences. Furthermore, owing to variations in costs, disability below the cutoff as defined based on cost-effectiveness alone does not necessarily imply that no medical or psychological benefits can be derived from the procedure at the individual patient level. In our study, preoperative scores do not correlate well with patient satisfaction after surgery. The MCID is a more clinically oriented concept that attempts to go beyond statistically significant changes at the group level that may not be significant at the individual level to find the smallest difference between the questionnaire scores that the patient perceives to be beneficial. 13 the extent of improvement that they have experienced. 23,24 Higher absolute postoperative scores and meeting patient expectations are more consistent determinants of patient satisfaction. [25][26][27] Likewise, we found that postoperative SF-36 PCS scores better correlated with patient satisfaction.
Knee-specific scores, such as WOMAC, Oxford, and Knee Society Scores, may also be helpful in quantifying functional ability. However, these scores will not replace a methodical history taking, comprehensive physical examination, and thorough discussion with the patient with careful understanding of his or her expectations and goals with surgery. In determining the costeffectiveness of TKA in addition to knee function, the general health and emotional state of the patient should be considered holistically to estimate his or her final functional status after surgery. 25 Future studies looking into this interplay of factors may allow us to identify novel factors associated with high absolute postoperative scores as a surrogate for patient satisfaction.

Limitations
Our study is limited in the following areas. First, to quantify satisfaction, our center used a 6-level Likert scale, with lower scores indicating better outcomes, adapted from the North American Spine Society questionnaire. A simple question may lack the sensitivity in detecting subtle differences accurately compared with a comprehensive questionnaire. Patients also have different opinions on what constitutes excellent and very good responses. We were also unable to delineate the specific reasons behind patient satisfaction or dissatisfaction. However, comprehensive questionnaires tend to be vulnerable to bias. Robertsson et al 28 reported the usable return rate of a more comprehensive questionnaire to be 18% to 45% lower than that of a simple satisfaction questionnaire. Patients not responding to the comprehensive questionnaires were more often unsatisfied with their operated knee than patients responding.