Aim: To determine whether follow-up radiography is a valid diagnostic investigation in patients with suspected acute scaphoid fractures and normal initial radiographs.
Materials and methods: Fifty sets of radiographs (initial and follow-up) were independently assessed by four expert observers for the presence or absence of a scaphoid fracture. Magnetic resonance imaging (MRI), performed in all cases, was used as the gold standard to determine the sensitivity, specificity, positive and negative predictive values of the observers' assessment of the follow-up radiograph. In addition, the reliability among observers of the follow-up radiograph was determined by reliability variance analysis.
Results: Of the 50 sets of radiographs, 35 patients had a scaphoid fracture and 15 were normal, as established from MRI report. For individual observer's assessment of the follow-up radiograph, sensitivities of 11, 9, 43 and 49%; specificities of 93, 93, 87 and 80%; positive predictive values of 80, 75, 88 and 85%; and negative predictive values of 31, 30, 39 and 40%. A reliability coefficient of more than 60% is needed for a diagnostic test to be considered reliable. Overall, the inter-observer reliability coefficient was 33%, with pair-wise inter-observer coefficients ranging from 18-53%.
Conclusion: With poor sensitivity, poor negative predictive value and poor reliability, follow-up radiography cannot be considered a valid diagnostic examination for the detection of scaphoid fracture in patients with normal initial radiographs.