Objective: To evaluate the reproducibility of US and to compare its efficacy with that of MRI and conventional radiography (CR) for the detection of bone erosion in RA patients.
Methods: A systematic literature search was performed in the Medline, Embase and Cochrane databases up to August 2009. Trials evaluating the reproducibility of US for bone erosion detection or comparing the number of erosions detected by the three imaging methods at patient and/or joint level were included. This last parameter was assessed using the odds ratio (OR) and 95% CI with the Mantel-Haenszel method (OR < 1 favours US). We explored the heterogeneity between the studies by subgroup analysis.
Results: Twenty-one studies including 913 patients were included in this meta-analysis. Intraobserver and interobserver reproducibility of US for erosion detection was good. US and MRI efficacies were comparable at both joint (OR = 1.19, P = 0.45; seven studies, 869 joints) and patient (OR = 1.76, P = 0.22; nine studies, 338 patients) levels. US detected significantly more erosion than CR at both joint (OR = 0.30, P < 0.00001; 4047 joints studied) and patient (OR = 0.31, P < 0.00001; 592 studied patients) levels. The number of patients to screen in order to detect an additional patient with an erosion in comparison with CR was 4, 95% CI (2.4, 5.9).
Conclusion: US is more effective for erosion detection than CR and has a comparable efficacy to MRI with good reproducibility.