Derivation of clinical prediction rules to identify children with fractures after twisting injuries of the ankle

Abstract

Objective The authors sought to derive maximally sensitive prediction rules for identifying children with significant fractures after acute twisting injuries to the ankle. Methods The authors prospectively enrolled a convenience sample of patients younger than 18 years of age who presented to the pediatric emergency department of an urban, tertiary care center after sustaining acute twisting injuries of the ankle. The ankle was defined in two regions: the malleolar zone and the midfoot zone. Clinical findings were documented on standardized data collection sheets by faculty physicians before completion of ankle and/or foot radiograph series. Significant fracture was defined a priori as any fracture other than an avulsion ≤3 mm. The authors conducted binary recursive partitioning with cross-validation to develop models to predict fracture. Results Data from 717 enrolled patients with a median age of 12.9 years were analyzed. Significant malleolar zone and midfoot zone fractures were diagnosed in 81 of 682 (11.9%) and 10 of 173 (5.8%) patients for whom ankle and foot radiographs were obtained. Recursive partitioning identified patients at low risk for malleolar zone fracture if 1) they had no bone tenderness at either malleolus or the region just proximal to the fibula malleolus or 2) they had bone tenderness at either malleolus but were able to walk four steps in the emergency department and had no swelling at either malleolus. The two-part malleolar zone rule had a sensitivity of 100% and specificity of 19.1% on the learning data and 95.1% (95% confidence interval [95% CI]=87.8% to 98.6%) and 20.0% (95% CI=16.8% to 23.4%) on the test data. Tenderness either at the proximal fifth metatarsal or cuboid identified midfoot zone fractures with a sensitivity of 100% and specificity of 31.9% on the learning data and 90.0% (95% CI=55.4% to 99.7%) and 35.0% (95% CI=27.7% to 42.8%) on the test sample. Conclusions A set of sensitive prediction rules was developed to identify children with significant fractures after twisting injuries to the ankle. External validation and refinement of the rules will be needed before recommendation of widespread use.