Patient population
After receiving approval from the ethics review board (Ziekenhuis Oost Limburg Genk - CTU2020022) a retrospective cohort study on a series raTKA or mTKA cases performed by one of three high-volume (> 200 cases per year) arthroplasty surgeons was conducted. The raTKA cases included the first 30 consecutive patients for each operating surgeon (n = 90) that met all inclusion criteria and none of the exclusion criteria (below). The mTKA group consisted of a consecutive series of 30 cases for each surgeon that occurred concurrently with the raTKA cases (n = 90). All patients provided informed consent. Patients were allocated to either raTKA or mTKA based on availability of the robotic device on the date of their surgery. All cases were performed between December 2019 and September 2020. Patients were included in the study if they were between the ages of 18 and 80 years, had an indication for primary TKA due to osteoarthritis, and had surgery by one of the surgeon authors (LV, EN, JT). Patients with congenital deformity, underlying neurological dysfunction, severe deformity (> 15 degrees of preoperative varus/valgus alignment or a non-correctable deformity), a prior infection or osteotomy around the knee, prior unicompartmental procedure or osteotomy, or fracture as the primary indication were excluded (n = 18). No cases were excluded due to conversion to mTKA for mechanical failures. Ultimately, 180 patients (Fig. 1) were included, with 60 (30 raTKA and 30 mTKA) patients from each surgeon. To further evaluate the learning curve, the raTKA group was split into consecutive groups of 10.The STROBE guidelines were followed [28].
Surgical technique and recovery
Surgeons underwent cadaveric training on the robotic system (ROSA® Knee System, Zimmer Biomet, Montreal, Quebec, Canada) prior to performing the procedure in the operating theatre. The surgeons also provided support amongst themselves by assisting each other on the first two cases, before performing the next cases on their own. Two surgeons had prior experience with other robotic systems, but this was minimal < 20 cases for one using a robotic-arm assisted surgery for partial knee arthroplasty and < 10 cases for the other using a hand-held burring robotic device. None of the surgeons had prior experience with other computer assisted surgery technologies. The same surgical technique with the subvastus approach, extension gap first, was used in all cases. Additionally, all patients received a patellar resurfacing. Every case for all surgeons included the same operating staff with one assisting nurse, a resident, and the same industry representative for all surgeons. The same prosthesis design (Persona® Posterior Stabilized; Zimmer Biomet, Warsaw) was used in all cases.
All raTKA cases were performed image-free using intraoperative data to achieve the preferred component placement and size (Fig. 2). Distal femoral and proximal tibial bi-cortical registration pins were inserted, and fixed optical trackers mounted. Bone registration was performed using boney landmarks displayed on the computer screen to verify anatomy and establish bone geometry. Joint balancing captured femoral and tibial poses with corrective forces, assessed kinematics through the arc of motion, and enabled fine tuning of implant positioning based on laxity of the soft tissue envelope. A robotic arm was used to execute the perioperative plan by placing the cutting jig in the appropriate plane, removing the need for traditional intramedullary guides (Fig. 3). Tibial and femoral osteotomies in the coronal plane were performed to achieve the overall alignment as desired by the surgeon (Fig. 3). In the sagittal plane, 3°- 5° of femoral component flexion were used to optimize implant sizing whilst preventing notching. The tibial slope was initially set to 4.5 degrees for the initial cut and then adjusted as required based on intra- operative assessment of the flexion gap (Zimmer FuZion® Tensor, Zimmer Biomet, Warsaw, IN, USA) and range of motion (ROM). Optical motion capture technology was used to assess limb alignment (Fig. 4), ROM (Fig. 5), flexion and extension gaps (Fig. 5), and arc of motion with trial implants prior to definitive selection and cement implantation (Refobacin® R, Zimmer Biomet, Warsaw, IN, USA).
Conventional mTKA was performed using standard instrumentation and all were fully cemented (Refobacin R, Zimmer Biomet, Warsaw, IN, USA). Intramedullary referencing was used to perform tibial and femoral bone resections. Flexion and extension gaps were checked using the same tensor device as raTKA and appropriate soft tissue releases performed to ensure balanced gaps. No further intraoperative adjustments or tailoring of implant positioning were performed to account for individual patient anatomy.
Every patient followed the same postoperative protocol, bed rest the day of surgery. Bed rest at day one with twice a day continuous passive motion (CPM) for 45-60 min. On day two, mobilization with crutches or a walker was added. Ascending/descending stairs with aid was added on day three. If pain was under control, wound was dry, knee flexion was over 60° and patients were self-sustainable with crutches, they could leave the hospital at postoperative day three or four.
Data collection and analysis
As some studies have reported learning curves in raTKA for surgical time between 20 and 40 cases, we included the first 30 raTKA cases with their concurrent mTKA cases for each surgeon as a convenience sample [18, 26]. Due to the limited follow-up time (minimum three-month) no further adjustments were made to the sample size. Due to geographical or logistical reasons some patients lacked postoperative radiographs (Fig. 1). Additionally, two patients lacked full extension (n = 1 for raTKA and mTKA) at 6 weeks postoperatively and were unable to obtain the full-length hip-knee-ankle (HKA) radiographs.
The learning curve for raTKA was assessed according to the recommendations of Hopper et al. [7] where both measures of surgical performance and early patient outcomes were assessed. Chart reviews were performed to obtain operative times for the surgical performance and 90-day postoperative complications to assess early patient outcomes. Radiographs were reviewed to assess final limb alignment (HKA angle) and planned vs. validated measures in the raTKA cases were also reviewed.
Operative time was defined as time from initial surgical incision to final wound closure. In raTKA, surgical times for the following parts of the procedure were also recorded: set-up (surgical tray, robotic device, and instruments), registration (surgical approach, insertion, registration of pins and bone registration), joint balancing, bone preparation, implant trialing, cement implantation of final prosthesis, and overall operative time.
Accuracy of achieving the planned alignment of the leg was determined based on full-length HKA radiographs. They were evaluated according to methods of Cooke et al. [6]. The hip center was obtained using concentric Mose circles. The goal for mTKA was to achieve zero degrees for the HKA angle, resulting in a neutral mechanical alignment (MA). However, this goal was adjusted in some cases with more severe deformity resulting in cases that were under corrected (not planned nor achieved a neutral mechanical alignment). Radiographic outliers for mTKA were considered as patients whose mechanical alignment was > ±3° of planned neutral MA. In raTKA, outliers were defined as being > ±3°of the intraoperative planned HKA angle. This was due to the surgeon’s using an adjusted mechanical alignment in raTKA to minimize ligamentous releases.
Implant positions were assessed as described by Moon et al. [15]. The femoral coronal implant alignment was measured as the lateral angle subtended by the femoral mechanical axis and the line connecting the distal points of the medial and lateral condyles of the femoral component known as the distal lateral femoral angle (DLFA). The femoral sagittal implant alignment (femoral flexion angle) was calculated as the angle subtended between the perpendicular line running proximally from the distal femoral surface in contact with the femoral component and the femoral mechanical axis. The tibial coronal implant alignment was measured as the medial angle subtended by the tibial mechanical axis and the medial to lateral axis of the tibial implant known as the medial proximal tibial angle (MPTA). The tibial sagittal alignment (tibial slope) was calculated as the angle between the tibial mechanical axis and anterior to posterior axis of the tibial implant. Anteroposterior radiographs were used to measure the joint line height by calculating the perpendicular distance from a line extending through the distal points of the femoral condyles and a parallel line extending to the fibular head. True lateral knee radiographs were used to calculate the posterior tibial slope [2] and posterior condylar offset ratio (PCOR) [8].
The radiological measurements were reviewed by two independent observers and MA was assessed for interrater agreement. Each observer was blinded to the other’s measures. The average of the two was used as the final measure. When looking at interrater agreement for mechanical alignment, a Bland-Altman plot demonstrated good agreement around the zero bias line and within the 95% confidence interval.
The cumulative summation sequential analysis tool (cumsum) [5, 29] was used to assess the learning curves in raTKA for operative time. Standardized target values for the cumsum analyses were set using the (surgeon-specific) mean values for the overall operative time from the raTKA group. Cumsum values represent a running total of the differences between the value of each data point and the standardized target. To determine the point at which the learning phase was over, we used a piecewise linear regression model. This model describes the cumsum function as two lines and the point where the lines connect (‘change point’) is the transition between learning phase and post-learning phase (mastered).
Continuous variables are presented by means and standard deviations. The difference between the planned and validated angles are reported as absolute values. Counts and percentages are used for categorical data. The raTKA and mTKA groups were compared with respect to patient characteristics using the chi square test and Fisher’s exact test for categorical data; and by means of an independent t test or Mann–Whitney test for continuous variables.
The mTKA, learning raTKA, and mastered raTKA groups were compared for operative time by means of two-way ANOVA models. To correct for a possible surgeon effect the ANOVA models included, both a group (mTKA, learning raTKA, and master raTKA) effect and a surgeon effect. Least-square means and 95% confidence intervals obtained for this model are presented. Statistical significance was set at p < 0.05 for all statistical tests. Analysis was performed using SAS for Windows software version 9.4 by an independent statistician (LB).