The role of evidence in the consensus process. Results from a Canadian consensus exercise

Abstract

As part of a consensus conference on cesarean birth, the ten-member consensus panel rated 224 clinical scenarios on their appropriateness for a cesarean section. Ratings were obtained before and immediately after the consensus conference. The level of agreement (consensus) among panelists was assessed separately for scenarios with good research evidence (evidence scenarios) and for those with conflicting, poor, or no evidence (nonevidence scenarios). For each scenario, consensus between panelists was measured as total agreement, partial agreement, or disagreement on the appropriateness of a cesarean section. Before the conference, total or partial agreement existed for a larger percentage of evidence than nonevidence scenarios (85% vs 30%), with the pattern reversed for disagreements (15% vs 70%). After the conference, possible improvement in the level of consensus actually occurred for 71% of the evidence and only 24% of the nonevidence scenarios. Thus, the consensus process, as structured here, was sensitive to the availability of good evidence and suggests that aspects of both expert and public processes can successfully be combined. However, an improvement could be made in the process by grading final recommendations according to the availability of rigorous research evidence.