Methods

Design

In this feasibility study, we invited 36 orthopaedic surgeons in The Netherlands and one in Sweden to participate. Further, we recruited two additional orthopaedic surgeons to record 19 videos of patients with meniscal tears undergoing diagnostic arthroscopy, and subsequently (in the same surgery) either having meniscal repair or APM. Only anonymized videos of the diagnostic visualization of the joint and meniscus was shown to orthopaedic surgeons asked to review the videos. Orthopaedic surgeons signed informed consent before study participation. The study was approved by the Medical Ethics Committee in The Netherlands (2023.0690).

Videos

Arthroscopic videos of knees in patients aged 19–46 years with an isolated meniscal tear were recorded in 2023. Videos of patients were excluded if the patient had I) fractures of the affected extremity within the previous 6 months, II) had previous knee surgery on the affected knee, or III) had a comorbidity requiring treatment other than meniscal tears (e.g. ACL injury).

The meniscus videos were collected from two orthopaedic centers (one public hospital and one private clinic in the Netherlands). The videos were anonymized and then saved on an online platform that was accessible only with a password. Orthopaedic surgeons were eligible to rate the videos if they had a minimum of one year of experience as an orthopaedic knee surgeon. Orthopaedic surgeons participating in the study received access to the platform to watch the videos. Participating surgeons were aware of the in- and exclusion criteria but did not receive additional clinical information. For each video, the orthopaedic surgeons judged 1) whether they were willing-to-randomize the patient with the meniscal tear shown on the video (yes/no), 2) which meniscal surgery they would perform on that meniscal tear (resection, repair, or no surgery), and 3) whether they preferred conservative (non-surgical) treatment over surgical treatment for this type of meniscal tear (yes/no). They recorded their findings in an Excel file. The surgeons rated the videos independently and were not aware of their colleagues’ responses nor of which treatment the patient actually received. One orthopaedic surgeon reported the type of meniscal tear, its location, and the vascularization of the area based on the Cooper classification, that takes the vascularization of the different meniscus sections into account [16].

Statistical analysis

We based our sample size on the following calculations. We generated up to 600 simulated datasets of artificial data for 18 different scenarios, assuming prevalence of answer “yes” from almost 0 to almost 1 and varying degrees of agreement between the raters. After running all scenarios, we concluded that we needed 30 raters (orthopaedic surgeons) and 20 patients’ videos to estimate an agreement coefficient with acceptable precision (width of 95% confidence intervals (CI) for kappa at most around 0.4).

Our primary outcome was the willingness of orthopaedic surgeons to randomize patients with meniscal tears to either resection or repair. The secondary outcomes were the preference for the type of meniscal surgery to perform, and whether non-surgical treatment was preferred over meniscal surgery. If a video review was missing, we still analyzed all available data from that reviewer. We analyzed agreement using Fleiss Kappa. Fleiss Kappa can be interpreted as the more well-known Cohen's Kappa, where values greater than 0.75 represent excellent agreement, values between 0.40 and 0.75 represent fair to moderate agreement, and values below 0.40 represent poor agreement. We reported the Fleiss Kappa coefficients with their 95% CI.