48, 95% CI 0.74 to 16.40). MRI was not useful in diagnosing other wrist ligament injuries. The MRI findings need to be interpreted with caution because surgeons who performed the arthroscopies were not blinded to the MRI results. While it is possible that our MRI results may have been better if we had used high resolution rather than low resolution MRI, this would seem unlikely. Faber and colleagues

(2010) reported no difference in the positive predictive values of high and low resolution MRI for diagnosing TFCC injuries, although higher resolution MRI was C59 slightly better for ruling out TFCC injuries. Anderson and colleagues (2008) argued that high resolution MRI was more useful than low resolution MRI for diagnosing wrist ligament injuries, however when we used the authors’ data to derive LRs we found that their results were very similar to our own. MRI combined p38 MAPK inhibitors clinical trials with provocative tests improved the proportion of correct diagnoses of TFCC injuries by 13% and lunate cartilage damage by 8%. That is, eight additional scans would need to be performed to make one more correct diagnosis of the presence or absence of TFCC injury compared to diagnosis by provocative tests alone, and 13 additional scans would need to be performed to make one more correct diagnosis of the presence or absence of

lunate cartilage damage. There was no benefit in performing MRI in addition to provocative wrist tests for diagnosis of SL, LT, arcuate ligament, and DRUJ injuries. The additional

diagnostic benefit of MRI scans needs to be weighed against the cost of 8–13 scans for one more correct diagnosis. The results of the arthroscopies indicated that 63% of wrists had synovitis. Synovitis is often due to an inflammatory reaction following trauma in the absence of arthritis. Perhaps those who had synovitis not had an injury to the joint capsule. This might partly explain the limited value of the provocative tests for diagnosing wrist ligament injuries. This possibility was explored with post hoc exploratory analyses in which any finding of wrist synovitis was cross tabulated with the SS test and then with the TFCC test. The TFCC test did not perform any better. The positive LR associated with an ‘uncertain’ test result (ie, hypermobile or pain different to the primary pain the participant presented with) for the SS test appeared to be moderately useful, but the estimate of diagnostic utility was very imprecise (LR 4.77, 95% CI 0.67 to 34). Further studies could explore the value of provocative tests for diagnosing wrist synovitis or other conditions. Strengths of this study include the recruitment of a consecutive sample of participants suspected of wrist ligament injuries, and that all participants were tested with the reference standard. A limitation of this study was that MRI was conducted at the surgeon’s discretion and performed on only a subgroup of participants.