Clinical diagnostic evaluation for scaphoid fractures: a systematic review and meta-analysis.

PURPOSE: To provide an overview of available clinical evaluation tests for scaphoid fractures and to compare their diagnostic accuracies. METHODS: PWe performed a systematic review of all studies assessing diagnostic characteristics of clinical evaluation in scaphoid fractures by searching MEDLINE, EMBASE, Cochrane, and CINAHL databases. Only studies on clinical testing prior to radiographic evaluation and with acceptable reference standard for occult fractures were included. Thirteen relevant articles were analyzed that described a total of 25 tests. Diagnostic characteristics of the tests were used to construct contingency tables. If possible, data were pooled and summary receiver operating characteristic curves were fitted. RESULTS: Anatomic snuff-box tenderness (ASB, 8 studies, 1,164 patients) and longitudinal thumb compression (LTC, 8 studies, 961 patients) had sufficient data for statistical analyses. Sensitivity for ASB ranged from 0.87 to 1.00; for LTC, 0.48 to 1.00. Specificity of ASB ranged from 0.03 to 0.98; for LTC, 0.22 to 0.97. Owing to considerable heterogeneity, pooled estimate points were not calculated. Other high-sensitivity tests were scaphoid tubercle tenderness, with sensitivity and specificity ranging from 0.82 to 1.00 and 0.17 to 0.57, respectively, and painful ulnar deviation, ranging from 0.67 to 1.00 and 0.17 to 0.60, respectively. Three studies showed that combining tests increased the specificity and post-test fracture probability while maintaining high sensitivity. Quality assessment showed high or unclear risk of bias and applicability concerns in reference standard and patient selection. Twelve study designs were prospective, and 1 was retrospective. CONCLUSIONS: Anatomical snuff box tenderness was the most sensitive clinical test. The low specificity of the clinical tests may result in a considerable number of overtreated patients. Combining tests improved the post-test fracture probability. This can be used to limit unnecessary immobilization, number of hospital visits, and use of imaging. The data presented herein may help to develop clinical prediction rules that could increase specificity without reducing sensitivity. TYPE OF STUDY/LEVEL OF EVIDENCE: Diagnostic II.