Detecting scaphoid fractures in wrist injury: a clinical decision rule

Abstract

Abstract Introduction The aim of this study was to develop and validate an easy to use clinical decision rule, applicable in the ED that limits the number of unnecessary cast immobilizations and diagnostic follow-up in suspected scaphoid injury, without increasing the risk of missing fractures. Methods A prospective multicenter study was conducted that consisted of three components: (1) derivation of a clinical prediction model for detecting scaphoid fractures in adult patients following wrist trauma; (2) internal validation of the model; (3) design of a clinical decision rule. The predictors used were: sex, age, swelling of the anatomic snuffbox, tenderness in the anatomic snuffbox, scaphoid tubercle tenderness, painful ulnar deviation and painful axial thumb compression. The outcome measure was the presence of a scaphoid fracture, diagnosed on either initial radiographs or during re-evaluation after 1–2 weeks or on additional imaging (radiographs/MRI/CT). After multivariate logistic regression analysis and bootstrapping, the regression coefficient for each significant predictor was calculated. The effect of the rule was determined by calculating the number of missed scaphoid fractures and reduction of suspected fractures that required a cast. Results A consecutive series of 893 patients with acute wrist injury was included. Sixty-eight patients (7.6%) were diagnosed with a scaphoid fracture. The final prediction rule incorporated sex, swelling of the anatomic snuffbox, tenderness in the anatomic snuffbox, painful ulnar deviation and painful axial thumb compression. Internal validation of the prediction rule showed a sensitivity of 97% and a specificity of 20%. Using this rule, a 15% reduction in unnecessary immobilization and imaging could be achieved with a 50% decreased risk of missing a fracture compared with current clinical practice. Conclusions This dataset provided a simple clinical decision rule for scaphoid fractures following acute wrist injury that limits unnecessary immobilization and imaging with a decreased risk of missing a fracture compared to current clinical practice. Clinical prediction rule 1/(1 + EXP (−(0.649662618 × if man) + (0.51353467826 × if swelling anatomic snuffbox) + (−0.79038263985 × if painful palpation anatomic snuffbox) + (0.57681198857 × if painful ulnar deviation) + (0.66499549728 × if painful thumb compression)−1.685). Trial registration Trial register NTR 2544, www.trialregister.nl.