Continuous outcomes of robotic-assisted total knee arthroplasty trials are statistically fragile.

Abstract

The purpose of this study was to calculate the continuous fragility index (CFI) of randomized controlled trials (RCTs) assessing outcomes following robotic-assisted total knee arthroplasty (RA-TKA). MEDLINE, EMBASE, and CENTRAL were searched from inception to April 9, 2024, for RCTs assessing outcomes following RA-TKA compared to conventional total knee arthroplasty (C-TKA). The CFI was calculated for statistically significant outcomes. Multivariable linear regression was performed across the study-level median CFI and various study characteristics. Seventeen studies (2,636 knees) were included in the final analysis. A total of 110 outcomes were eligible for analysis with a median CFI of 4.5 (IQR, 2.0-8.9). Radiographic outcomes were the most common significant outcome reported (33%, median CFI 7.6 [IQR 4.6-13.2]) followed by inflammatory markers (32%, median CFI 2.4 [IQR 1.0-3.0), operative characteristics (21%, median CFI 7.0 [IQR 4.2-13.7]), subjective outcomes (9%, median CFI 5.2 [IQR 3.1-8.3), and objective outcomes (5%, median CFI 3.1 [IQR 2.2-4.4]). There was a statistically significant difference across the aforementioned categories (p < 0.05). Multivariable linear regression demonstrated that larger sample size ( ß = 0.496, p < 0.0001) and inclusion of inflammatory markers (ß = 0.472, p = 0.015) were correlated with higher study-level median CFI. Continuous outcomes across trials assessing RA-TKA versus C-TKA are relatively fragile compared to the orthopaedic literature to date. Radiographic outcomes were the most statistically robust category. The statistical robustness of findings may be limited by RCTs of a small sample size highlighting the importance of future collaborative, multicentre trials. LOE: I.