This study shows that planning of TKA using PSI by different surgeons results in an excellent agreement for implant sizes between surgeons as well as in repeated planning by the same surgeon. Next to implant size, intra- and interobserver reliability demonstrated good to excellent agreement (ICC > 0.75) for 7 out of 12 remaining settings and 6 out of 12 parameters, respectively. Hence, it may be stated that PSI is a reliable method for planning of a TKA.
Previous studies have shown that PSI planning accurately predicts the implant size used intraoperatively [5, 8, 12]. The current study shows that planning of the implant size, within and between orthopaedic surgeons, is reliable. The maximum size difference was 1 implant size for the femur, and 2 implant sizes for the tibia, compared to the other plannings of the same patient.
Changes to the default plan can result in different implant sizes. Overall, more changes were made to the tibia than the femur. This may explain the greater difference in default and approved implant size for the femur and tibia: 1 size versus 2 sizes, respectively. TKA surgery can be planned more effectively by understanding the size change frequency of implants, in combination with intraoperative concordance to the preoperative plan. Consequently, when implant sizes can be accurately predicted, and planning implant sizes itself is reliable, the operating team will be able to minimize intraoperative implant size errors in advance. This may lead to improved operating room efficiency due to a decreased number of required operative trays (eq. reliable patient-specific trays), less inventory planning, and possibly lowering hospital costs per TKA procedure. This may be of interest for future research.
Previous studies have emphasized that changes in the initial technician’s plan were necessary to get an accurate preoperative planning of the implant sizes. Intraoperative alterations in implant size were significantly lower for the plans approved by the surgeon compared to the default plans provided by the technician [4, 10, 12]. Based on this previous literature, the expertise of the surgeon is thus essential for evaluating and approving the default planning provided by the manufacturer.
Intra- and inter-observer reliability ICC were 1.00 for femoral rotation from the epicondylar axis, posterior tibial slope, and tibial rotation because none of the surgeons modified these parameters. Due to general consensus on these parameters, less variation will occur, with a higher agreement as result. When there is less consensus on a certain parameter, more changes are made which will result in a lower agreement. Surgeon 3 made the fewest alterations to the proposed plannings, resulting in an excellent agreement (ICC > 0.90) for intra-observer reliability for all settings. Thus, high ICC can be caused by good agreement between adjusted plannings, or due to no alterations made to the proposed plan. Additionally, the adjustment of one parameter can derive alterations of other parameters. For example, an increase of resection might result in the need for a smaller implant size and adjustments in placement of the newly chosen implant size. Awareness of this effect is essential when interpreting the results of this article. This ‘snowball effect’, as well as less consensus on certain parameters with more changes to the default planning and therefore more differences within and between surgeons, may explain why parameter such as femoral flexion/extension and tibial displacement showed less agreement.
Mechanical alignment technique is considered well performed when the overall limb alignment is within 3° of neutral. Varus- and valgus angles for both femur and tibia showed modifications with a maximum of 0.5° and 2° respectively. Given that the maximum difference of varus/valgus angles is 2° within the same case, it is supposed that these changes are of no clinical importance. Moreover, adjustments to varus/valgus alignment are known to be dependent on the surgeon’s philosophy for an anatomical-, (adjusted) mechanical-, or (restricted) kinematic alignment technique [2, 11]. A patient’s specifications, such as findings from a physical examination (for example preoperative leg axis, body mass index, and laxity) and previous medical history, can be determinative in the decision for a certain alignment.
Patient-specific characteristics uncontrollable by planning software, namely, ligamentous balancing and lower limb alignment, can require intraoperative changes. Therefore, correct matching of the pre-operative plan and intraoperative observations is a crucial factor in PSI-assisted TKA. In case of a mismatch, it is the surgeon’s responsibility to consider a switch to conventional instrumentation. In previous literature intra-operative modifications were made to the pre-operative plan in 23% up to 36% of MRI-based and CT-based PSI-assisted TKA, respectively. Most of these changes occurred due to a poor match between the pre-operative plan and intra-operative observations for the tibial component [3, 4, 15].
Furthermore, each surgeon has a ‘personal touch’ not only in planning, but also intraoperatively with his or her own preferences of additional releases, the decision whether or not recuts are needed, or the consideration to select a thicker insert in patients with a high BMI. Nonetheless, excellent agreement for implant sizes between surgeons and within surgeons was found in this study. Also, the agreement of implant size did not differ between the surgeons who made multiple changes to the proposed plans compared to the surgeon who made very little changes to the proposed plan.
This study has some limitations. Firstly, no power analysis for the number of surgeons, the number of patient cases, and repetitive measurements have been conducted. However, Koo et al. suggested as a rule of thumb that researchers should obtain at least 30 heterogeneous samples and involve at least 3 observers whenever possible when conducting a reliability study . Therefore, in the present study 40 patients were included and planned by 4 different orthopedic surgeons. Secondly, only one type of PSI was evaluated in this study. Other PSI planning systems may perform differently. Thus, these results may not be representative of all PSI technologies available. No comparison to the intraoperative results was made during this study; this study focused on the agreement of repeated planning by different surgeons—the correlation between the planned and placed implant has already been addressed in previous studies.
This study represents the first assessment of intra- and inter-observer reliability in PSI TKA. The study showed an excellent intra- and inter-observer reliability, among which implant sizes. This may contribute to more optimal and potentially effective preoperative planning of TKA surgery in the future. Therefore, this topic can be of interest for further research. Future research, with larger dataset measurements and different types of both MRI- and CT-based PSI, is necessary to further evaluate these results.