Background. Clinical practice guidelines quickly become outdated. One reason they might not be updated as often as needed is the expense of collecting expert judgment regarding the evidence. The RAND-UCLA Appropriateness Method is one commonly used method for collecting expert opinion. We tested whether a less expensive, mail-only process could substitute for the standard in-person process normally used.
Methods. We performed a 4-way replication of the appropriateness panel process for coronary revascularization and hysterectomy, conducting 3 panels using the conventional in-person method and 1 panel entirely by mail. All indications were classified as inappropriate or not (to evaluate overuse), and coronary revascularization indications were classified as necessary or not (to evaluate underuse). Kappa statistics were calculated for the comparison in ratings from the 2 methods.
Results. Agreement beyond chance between the 2 panel methods ranged from moderate to substantial. The kappa statistic to detect overuse was 0.57 for coronary revascularization and 0.70 for hysterectomy. The kappa statistic to detect coronary revascularization underuse was 0.76. There were no cases in which coronary revascularization was considered inappropriate by 1 method, but necessary or appropriate by the other. Three of 636 (0.5%) hysterectomy cases were categorized as inappropriate by 1 method but appropriate by the other.
Conclusions. The reproducibility of the overuse and underuse assessments from the mail-only compared with the conventional in-person conduct of expert panels in this application was similar to the underlying reproducibility of the process. This suggests a potential role for updating guidelines using an expert judgment process conducted entirely through the mail.
Clinical practice guidelines are important and widely used tools for synthesizing clinical information, promoting desired practice patterns, and reducing healthcare costs. 1,2 They often accomplish this by offering explicit recommendations for reducing overuse and underuse of medical interventions, both of which can contribute to poor health outcomes. Evidence suggests, however, that clinical practice guidelines could quickly become outdated. 3 One reason they might not be updated as frequently as needed is that they are time-consuming and expensive to produce. To facilitate the development or updating of guidelines, it would be desirable to simplify the process and reduce its cost.
Most guideline development processes use mechanisms of some sort to incorporate expert opinion. 4 One commonly used mechanism for collecting and synthesizing expert opinion is the RAND-UCLA Appropriateness Method. 5,6 This method has been used to evaluate the appropriateness of performing a variety of medical and surgical interventions, 7 assign priority to patients requiring coronary revascularization, 8 and determine which low-acuity emergency department patients could have their care safely deferred. 9,10 This method incorporates an in-person group discussion about areas of clinical uncertainty and disagreement.
Convening a national or international panel of experts for face-to-face discussions is logistically complex, and increases the time and cost of producing criteria. One method for addressing these concerns is to conduct the process entirely through the mail. However, the impact of this modification is unknown. Our goal was to conduct a formal test of the reliability (reproducibility) of this method comparing feedback of round 1 ratings through mail with the standard in-person feedback of ratings and discussion.