Description of population
This retrospective observational study was approved by the Institutional Review Board of Yokohama City University (B190700017), and informed consent was obtained from all patients. Inclusion criteria for the current study included patients whose knees were treated with consecutive, simultaneous bilateral TKA between January 2014 and December 2018 in a single hospital. In total, 212 patients met the inclusion criteria. We excluded 18 patients with rheumatoid arthritis, seven patients who could not participate in the rehabilitation program for reasons, such as postoperative thrombus embolism, and three patients with missing preoperative physical data. Finally, a total of 191 patients were enrolled in the study (Fig. 1). All patients were diagnosed with osteoarthritis.
Surgical interventions and postoperative treatments
Bilateral TKAs were performed sequentially by one team comprising five experienced orthopedic surgeons. General anesthesia with peripheral nerve block was used for all patients, except for those who did not agree with or could not undergo peripheral nerve block due to complications. A lateral curved longitudinal skin incision and midvastus approach was employed. Intra- and extramedullary techniques were used for placement of femoral and tibial cutting guides, respectively. A modified gap-balancing technique was employed. A tourniquet was inflated before incision and deflated after cement hardening. The Scorpio NRG® posterior-stabilized prostheses (Stryker Orthopedics, Mahwah, NJ, USA) were used in all cases.
All patients, except those with anemia (hemoglobin [Hb] < 11.0 g/dL), underwent two 400 mL autologous blood collections prior to surgery. Postoperative autologous blood transfusion of 800 mL was performed within 3 days after surgery, independent of Hb levels. In patients with Hb level < 11 g/dL, we could not collect autologous blood samples due to our hospital’s autologous blood guidelines, and allogenic blood transfusion was performed in patients with Hb level < 7.5 g/dL or those who were expected to progress to anemia because of postoperative bleeding.
Postoperative pain management comprised non-steroidal anti-inflammatory drugs for internal use with an additional acetaminophen infusion when required. Rehabilitation was started on the day after surgery and continued until discharge. Full weight-bearing was allowed from the day after the surgery. Patients underwent a stepwise program to achieve the ability to use a wheelchair, followed by walking using a walker and T-cane, depending on compliance. In our facility, the rehabilitation goal was set within 3 weeks post-surgery following the clinical pathway. Generally, patients could not be discharged on time if stable independent walking with a T-cane and stepping up and down the stairs were not achieved within 3 weeks postoperatively. Patients were informed that a hospital transfer would be required if rehabilitation programs needed more than 3 weeks. Patients were discharged to their homes when they were able to walk using a T-cane and step up and down the stairs independently.
Data collection
Patient data, including age, sex, body mass index (BMI), living alone, Knee Society function score (KSS) [8], American Society of Anesthesiologists (ASA) class, preoperative Hb levels (g/dL), serum albumin (Alb) level (g/dL, measured 8 weeks prior to surgery before autologous blood collection), mean range of motion (average of the two sides), and mean Kellgren–Lawrence classification (K–L grade, average of the two sides), were collected from the hospital’s computer databases [9]. All parameters were obtained from the medical records at the time of preoperative outpatient examination and through patient interviews. The rehabilitation program utilization was calculated as the duration from the date of operation to the date of discharge from the hospital to the patient’s home. Patients who were transferred to other hospitals due to delayed rehabilitation were analyzed separately as transferred cases. Patients who could return to their residence within 3 weeks after surgery were assigned to the compliance group (n = 132), whereas those who could not and transferred patients were assigned to the delayed group (n = 59).
Statistical analysis
Values are expressed as mean ± standard deviation (SD). Compliance and delayed groups were defined as dependent variables. Age, sex, BMI, living alone, KSS, the ASA class, Hb levels, Alb levels, mean range of motion, and the K–L grade were defined as predictor variables. Normally distributed data were analyzed using a non-paired t-test, whereas non-normally distributed data were analyzed using the Mann–Whitney U test; the chi-square test was used for between-group comparisons of the preoperative parameters. Multivariate logistic regression analysis using a forward-backward stepwise selection method was performed to determine factors influencing successful compliance or prolonged rehabilitation programs. Standardized regression coefficients (β) and associated P-values were determined to assess statistical significance. P-value < 0.05 was considered statistically significant. According to the methods of Sullivan et al. [10] and Oba et al. [11], a prediction scoring formula was created using the multivariate logistic regression model results. Steyerberg et al. reported that stepwise selection with low alpha (P < 0.05) leads to a relatively poor model performance in cases with small samples [12]. Therefore, we created a prediction formula using factors resulting from the regression analysis (P < 0.20). Similarly, Zemek et al. selected variables using factors with P < 0.20, which was considered to have some correlation with the results [13].
Scores were calculated by adding the results of multiplications between each independent factor and its coefficient β. To simplify the formula, up to three significant digits of the coefficient β were used. Ten points were added to the formula to make the score positive. C-statistics and likelihood ratio were used to assess the predictive ability of the model. Patients were stratified according to their score, and the observed probability of being in the compliance or delayed group was assessed. All statistical analyses were performed using the JMP 12.2 software (SAS Institute Inc., Cary, NC, USA).