STUDY OBJECTIVE: To develop criteria that optimize clinical decisionmaking in the use of radiography after isolated knee trauma in adults. DESIGN: A prospective survey of emergency department patients over a 7-month period. Standardized data forms were completed by emergency physicians, residents, and certified physician assistants. SETTING: A large suburban community teaching hospital. PARTICIPANTS: Two hundred forty-two patients older than 17 years with isolated knee injuries sustained less than 24 hours previously. RESULTS: We constructed a clinical decision model, calculating sensitivity, specificity, and odds ratios. Twenty-eight patients (11.6%) had fractures, with the patella the most commonly fractured osseous structure. Patients able to walk without limping had not experienced a fracture, nor had patients with twist injuries without effusion. Sensitivity of this model for detecting fracture was 1.0 (99% confidence interval, .97 to 1.0), and specificity was .337 (99% confidence interval, .26 to .42). CONCLUSION: Clinical decision rules are effective in detecting knee fractures with 100% sensitivity and with sufficient specificity to eliminate 29% of knee radiographs in the ED. These findings require prospective validation.
STUDY OBJECTIVE: To develop criteria that optimize clinical decisionmaking in the use of radiography after isolated knee trauma in adults. DESIGN: A prospective survey of emergency department patients over a 7-month period. Standardized data forms were completed by emergency physicians, residents, and certified physician assistants. SETTING: A large suburban community teaching hospital. PARTICIPANTS: Two hundred forty-two patients older than 17 years with isolated knee injuries sustained less than 24 hours previously. RESULTS: We constructed a clinical decision model, calculating sensitivity, specificity, and odds ratios. Twenty-eight patients (11.6%) had fractures, with the patella the most commonly fractured osseous structure. Patients able to walk without limping had not experienced a fracture, nor had patients with twist injuries without effusion. Sensitivity of this model for detecting fracture was 1.0 (99% confidence interval, .97 to 1.0), and specificity was .337 (99% confidence interval, .26 to .42). CONCLUSION: Clinical decision rules are effective in detecting knee fractures with 100% sensitivity and with sufficient specificity to eliminate 29% of knee radiographs in the ED. These findings require prospective validation.