How to synthesize evidence for imaging guidelines

AIM: To provide guidance on how to gather and evaluate evidence from the literature on the efficacy of imaging, using as an example the assessment of the role of magnetic resonance imaging (MRI) in the diagnosis of osteomyelitis. This method was adopted for evaluating evidence for the musculoskeletal section of the 5th edition of the Royal College of Radiologists' (RCR) guidelines. MATERIALS AND METHODS: A systematic review of the literature published between 1966 and July 2001 was carried out. Eligible articles described studies in patients with suspected osteomyelitis and who were diagnosed using MRI. Search strategies were developed to identify relevant imaging studies. Studies included in the systematic review were selected using predefined criteria. The sensitivity, specificity, accuracy and likelihood ratios for MRI reported in the studies were used to evaluate the value of the procedure in osteomyelitis. Where the above were not reported, they were calculated by the reviewers. RESULTS: The average sensitivity of MRI in osteomyelitis was 91% (range 76-100%), the average specificity was 82% (range 65-96%), average accuracy was 88% (range 71-97%), and the average positive likelihood ratio was 7.8 (range 2.3-21.1). Four studies evaluated the use of MRI in the diagnosis of osteomyelitis in the diabetic foot, two in osteomyelitis of the lower extremities, while four each evaluated the use of MRI in vertebral osteomyelitis, in the diagnosis of any form of osteomyelitis, osteomyelitis in spinal cord-injured patients and in cranial osteomyelitis. CONCLUSION: Systematic reviews of literature can be used to obtain evidence on the value of imaging procedures. The quality of the studies included in the review should always be considered when selecting studies to limit bias. In our example, MRI appears sensitive, specific and accurate in the diagnosis of osteomyelitis at different sites