2.1 Population selection and characteristic
Patients for this study were identified retrospectively from a compiled arthroplasty cohort held at our clinical centre who underwent unilateral total knee arthroplasty(TKA) for knee osteoarthritis(OA) from Jan. 2017-Dec. 2017. To exclude the bias caused by different surgeons’ technical factors, patients’ procedures were performed by the same attending group. Inclusion criteria were as follows:1)diagnosed as osteoarthritis of the knee;2)primary TKA was scheduled without any surgical operations before;3)the availability of peri-operative X-rays of a standard anterior and lateral position of the knees and full-length radiographs of the lower limbs as well. Exclusion criteria were as follows:1)inflammatory joint diseases, septic arthritis or tuberculous arthritis; 2)osteonecrosis, fractures or bone tumor of the target knee(s) which requires TKA;3)unrelated death or unwillingness to answer questions during follow-up.
2.2 Date collection
Patients completed a questionnaire without any implication from the medical unit at baseline and 3-year follow-up section. The following preoperative information were collected:age, side, sex, disease duration, body mass index(BMI), education level and radiographic findings including the minimal joint space width(pre-mJSW), preoperative flexion contracture of knee(FCK) and varus/valgus angle of knee(VAK). Postoperative information were collected as follows: Satisfaction rate(percentage), complications(deep vein thrombosis, pulmonary embolism, incision infection or disunion, or any readmission related to the procedure) . The following parameters were collected both preoperatively and postoperatively: mental health, HSS score, 11-point NRS (rest and walking) and VAK in X-rays.
The patients’ education level were recorded according to the following pattern: “well-educated” (those with the degree of university or higher); ”basically-educated”(those with the degree of middle or high school); “poorly-educated”(those with the degree of primary school or no experience of education).
The evaluation of patients’ mental health was assessed at baseline and 3-years after operation. Two conditions would be regarded as ”Depression state” in our study: 1)patients diagnosed with MDD(major depressive disorder)at baseline; 2)during the follow-up section, we applied The Hamilton Depression Rating Scale (HAM-D, 24-item) for the patients without the history of depression, and patients with total score greater than 8 were defined as “depression state”[12].
Digital photographs were taken of the target knee in the standard anterior and lateral position. And lower-limb full-length radiographs were applied for all patients pre- and postoperatively. All preoperative minimal joint space width(pre-mJSW) and radiological alignment readings were completed by three independent experienced observers who were blinded to patient demographics and outcomes. The radiological parameters were all measured by the PACS system (See Supplementary Material).
Postoperative function and pain improvements were measured by Hospital Special Surgery(HSS)score and numerical analog score(11-point NRS)in rest and walking. In addition, we absorbed the changes in the alignments after TKA as another aspect of objective parameters. The follow-up questionnaires must be answered by the patients themselves.
2.3 Grade of satisfaction
Satisfaction was measured by asking the patients to rate their feedbacks towards the operation and postoperative rehabilitation after 3 years. The responses would be recorded in numerical form initially, but converted to 4-point scale eventually.
Patients’ satisfaction towards the overall procedures and rehabilitation was considered as the primary parameter for subjective outcomes. In our study, we divided the satisfaction rate into four grades:”very satisfied(91%-100%)”, ”satisfied(81%-90%)”, “acceptable(61%-80%)”, “disappointed(≤60%)”.
2.4 Surgery procedures and implant materials
All TKA procedures were performed in the same operation room(OR 501) by the same attending group. In order to control the co-variates in surgery, posterior-stabilized(PS) prosthesis was implanted in all patients. All the implants are made of Co-Cr-Mo alloy produced by the same corporation.
2.5 Defining the MCID and MIC
We applied the anchor-based approach to establish the MCID in our study[13]. Anchor-based methods determine the MCID by associating the changes in the numerical scale of independent assessment of improvement with patient-reported subjective outcome(s)[14]. At the end of our follow-up questionnaire, one question would be asked:”Given the overall rehabilitation three years after TKA, how much will you rate your satisfaction if 100% means perfect or very satisfied”? The MCID was calculated as the difference in the mean change between patients with “satisfied” group compared with those with “acceptable” group according to the anchor question.
In addition, we define the MIC(minimum important change) as the change in the score relative to the baseline for patients who reported "the minimal improvement" ("satisfied" group in this study)according to the anchor question. Receiver operating characteristic (ROC) curve was used to identify threshold which is equivalent to the point achieving the maximal sensitivity and specificity in predicting the minimal improvement.
2.6 Data analysis
Descriptive statistics were performed on all study data. Besides the 4-point-scale satisfaction, patients were also divided into two groups based on whether they were satisfied (“very satisfied and “satisfied” group) or dissatisfied (“acceptable” and ”disappointed” group). Unpaired Student’s t test and non-parametric Wilcoxon rank-sum(Mann–Whitney) test were used for continuous variables, and the Chi-square test for categorical variables. Paired Student’s t test was applied for preoperative and postoperative assessments. One-way analysis of variance (ANOVA) with correction for multiple testing (Bonferroni)was used to compare means between groups. Variables were entered into a multiple ordinal logistic regression predicting the independent associations of them with patient's satisfaction. Variables tested included: age, sex, BMI, HSS score change, NRS-walking change, pre-mJSW and depression state.
Spearman’s rank correlation and Kendall’s rank correlation were used to identify the relevance between continuous and/or categorical variables. Simple linear regression analysis was used to identify the MCID, using the slope of the line for the change according to different level of satisfaction in the HSS and NRS score. The receiver operating characteristics (ROC) curve was used to define the MIC(threshold) that best discriminated (maximum sum of specifcity and sensitivity of the model) between “satisfied” and “acceptable”group.
Statistical significance was set at p<0.05 and all tests were 2-tailed. All statistical analyses and illustrations were done using Stata (version 15.1) software (StataCorp) and GraphPad Prism(version 8.0).