Validity of ultrasound in knee osteoarthritis - A systematic review and meta-analysis

Objective The study aims were to systematically review the literature on ultrasound validity in knee osteoarthritis (OA) and conduct meta-analysis where appropriate. Methods Databases were searched, including PubMed, Embase (Ovid Medline), and EBSCOhost (Medline), from their inception to December 2024. The quality of articles was critically appraised with the modified Downs and Black checklist. Result initial search identified 3112 records that were narrowed down to 106 for final analysis, including a total of 9786 patients (and 11756 joints with a mean age of 62.95 years. Ultrasound pathologies were defined according to OMERACT (17, 16.0%), European Alliance of Associations for Rheumatology (EULAR) (10, 9.4%), EULAR-OMERACT (4, 3.8%), and ZAGAZIG (named according to the ZAGAZIG University, Egypt) (2, 1.9%) guidelines. Pooled analysis demonstrated weak positive correlations between ultrasound pathologies and pain (r = 0.04-0.22, I 2 =0-95%) and serum biomarkers (r = 0.05, I 2 =0-32%), weak-moderate positive for X-ray (r = 0.13-0.52, I 2 =0-77%), moderate for MRI-synovitis (r = 0.52, I 2 =32%), and strong positive for histology (r = 0.64-0.66, I 2 = 0-8%). Ultrasound discriminative validity for Baker's cyst (BC), effusion, and anserine bursitis was strongly positive (OR = 3.26-5.96, I 2 =0-49%). Moreover, the ultrasound external responsiveness of synovial effusion regarding pain was demonstrated weak (SMD =0.30, I 2 =1%], while internal responsiveness for BC and effusion was revealed strong (r = 0.91 - 2.50) [I 2 = 0-11%]. Conclusion Ultrasound demonstrated heterogenous criterion, construct, and discriminative validity in knee OA. Given that the data pooling yield from high-quality primary research useful in clinical decision-making and could be done.