Rationale Use of clinical decision rules can have wide-ranging benefits. Some rules (e.g. the OAR) are reportedly widely used, yet considerable work suggests that physician self-report on issues like this can be unreliable. Our vignette-based study examined whether radiography decisions for realistic ankle injury scenarios are based on clinical findings that are part of the OAR (Rule-based findings) or other (Non-Rule-based) clinical findings. We tested two hypotheses: 1) Physicians should put greater weight on Rule-based findings than Non-Rule findings; 2) Findings used in decision making will differ between identifiable subgroups of physicians (e.g. older vs. younger).

Methods Our postal survey to 240 Canadian emergency physicians included 20 case vignettes of patients with ankle pain. The cases contained 8 clinical findings known to affect the decision to order ankle radiography. In order to reduce the number of cases, the 8 findings were varied according to a fractional factorial design. We asked respondents to estimate the likelihood they would order a ankle x-ray in each case. Using this estimate, we inferred – for each physician – the weight they placed on each of the 8 findings in making their decision. Two findings were based on the rule, 3 were predictive of fracture but not in the rule, and 3 were non-predictive.

Results The survey yielded 116 responses (48% response rate). 1) Weights for Rule-based findings differed from 0 for 91.6% of respondents; for Non-Rule predictive findings, 12.7%, and for Non-Rule, Non-Predictive findings, 5% (Rule vs. others: p < .001). 2) Younger physicians (<7 years post-graduation) differed from the oldest physicians (19+ years) in that they placed less emphasis on one Rule indicator(Bear Weight; p=.013), and greater emphasis on one Non-rule predictive indicator(Age; p=.015).

Conclusions Consistent with previous self-report work, our study shows physicians placed greatest emphasis on the two clinical findings that are part of the Ottawa Ankle Rules. Younger physicians differed from more experienced physicians in their use of both Rule indicators and Non-Rule indicators, suggesting areas for possible targeted training. This methodology is useful for understanding how CDRs fit into the decision making processes of individual physicians.