This meta-analysis was performed using methods described by the Cochrane Methods Group on Systematic Reviews of Screening and Diagnostic Tests. We included original, English-language papers published between January 1975 to October 2004 that examined contrast-enhanced arthrography with or without subtraction for diagnosis of loosening of total hip prostheses. A qualitative and quantitative analysis was performed by two investigators.
With regard to the acetabular component, pooled sensitivity and specificity for contrast arthrography was 70 % (95 % confidence interval, 52–84) and 74 % (95 % CI, 53–87), respectively. Subtraction arthrography had a significantly higher sensitivity of 89 % (95 % CI, 84–93) (p = 0.01), with a similar specificity of 76 % (95 % CI, 68-82). For the femoral component, pooled sensitivity and specificity for contrast arthrography were 63 % (95 % CI, 53–72) and 78 % (95 % CI, 68–86). Pooled estimates for subtraction arthrography revealed a significantly higher sensitivity of 86 % (95 % CI, 74–93) (p = 0.003). Specificity was 85 % (95 % CI, 77–91) and was similar to the data of contrast arthrography (p = 0.23).
Using the present data we found that the subtraction arthrography is a sensitive technique for detection of loosening of total hip prostheses, offering added value over contrast arthrography, especially for evaluation of the femoral component.