Can surgeons accurately estimate loss of threshold alignment (instability) of distal radius fractures? the influence of imaging diagnostic accuracy of radiographs compared with CT

Aims: Almost half of distal radius fractures (DRFs) lose threshold alignment (i.e. instability) after closed reduction and immobilization. This study aimed to investigate surgeons' ability to estimate secondary displacement by addressing three questions: 1) What is the diagnostic accuracy of surgeons to estimate instability of DRFs on pre- and post-reduction radiographs?; 2) What is the diagnostic accuracy of surgeons to estimate instability of DRFs on post-reduction CT imaging?; and 3) What patient factors are associated with estimating instability? Methods: We performed a scenario-based, randomized experiment with two distinct online surveys. In Part I, 116 members of the Science of Variation Group assessed radiographs of 20 initially displaced DRFs (11 'stable', nine 'unstable'), and estimated the loss of threshold alignment after closed reduction. Half viewed pre- and post-reduction radiographs, while half viewed only post-reduction radiographs. In Part II, 115 participants assessed 15 DRFs cases (six 'stable', nine 'unstable') to estimate loss of alignment. Half of the participants evaluated pre- and post-reduction radiographs, and half also received post-reduction CT imaging. Results: In Part I, diagnostic accuracy for estimating loss of threshold alignment on pre- and post-reduction radiographs was 54% (95% CI 51% to 57%), similar to 55% (95% CI 46% to 62%) when only viewing post-reduction radiographs (p = 0.063). In Part II, the accuracy was 70% (95% CI 64% to 77%) with both radiographs and CT, compared with 67% (95% CI 61 to 67) with radiographs alone (p = 0.240). Patient factors associated with estimating instability were female sex and higher age. Conclusion: Surgeons' ability to detect DRF instability on both pre- and post-reduction radiographs, as well as post-reduction CT-scans, was limited, reflecting a restricted value of probability estimates for clinical decision-making. Given suboptimal estimations of alignment loss, it seems prudent to monitor adequately reduced fractures during initial immobilization. Future studies should focus on aids that can overcome this limited accuracy.