Safety and efficiency of the Ottawa Ankle Rule in a Swiss population with ankle sprains.

Abstract Clinical decision aids aim to guide diagnosisor treatment by integrating a small set of easilyaccessible clinical information. The prototype ex-ample for such a decision aid is the Ottawa AnkleRule (OAR) [1] to reduce the number of unneces-sary x-rays in patients presenting with acute anklesprain. The OAR states that ankle series radi-ographs are only indicated if there is any pain inthe ankle region, and if there is either bone ten-derness along the distal 6 cm of the posterior edgeor the tip of either malleolus or inability to bearweight (4 steps) immediately after the injury andin the Emergency Department (ED).The OAR is calibrated towards high sensitiv-ity at optimal specificity. High sensitivity min-imises false negative results (fracture is present al-though the OAR is negative). On the other handspecificity correlates with the usefulness of therule in helping avoid unnecessary x-rays and asso-ciated costs. A systematic review assessing the di-agnostic value of the OAR revealed substantialheterogeneity of specificity ranging from 11 to 67percent that could not be explained from reportedstudy characteristics [2]. This indicates that it isfairly unclear what we can expect from an imple-mentation of the OAR in terms of cost savings inpractice. More understanding of the circum-stances which affected specificity in the reviewedstudies could increase our understanding of theoptimal use of the OAR in practice. Therefore, in this validation study of theOAR we had two aims: a) to validate the OAR in aSwiss population and, b) to study variability of therule’s efficiency in practice using a set of co-regis-tered details about the examination itself, the as-sessor of the rule and the patients. Since interpre-tation of the bony tenderness item appears to bethe most challenging task when applying the rule,we hypothesised that medical staff that completeda surgical training would perform better than sur-gically inexperienced assessors even if bothgroups were trained to perform the rule as sug-gested by the developers [1, 3].