Consensus development for producing diagnostic procedure guidelines: SPECT brain perfusion imaging with exametazime.

Abstract

UNLABELLED: A method for developing diagnostic practice guidelines is presented in which a team of experts used a semiquantitative scoring system to reach consensus on a standard procedure for SPECT cerebral perfusion imaging. METHODS: An expert panel generated a list of elements that they thought were important for the optimal performance of cerebral perfusion imaging as the first phase of a modified Delphi panel technique. Panel members then scored each statement to indicate the importance of that statement for the performance of cerebral perfusion imaging. The scores were recorded for each statement and the average score, s.d. and variance for each statement were determined for each successive panel round. A total of three panel rounds were conducted. The change in average s.d. between scoring rounds was analyzed for significance using both parametric and nonparametric tests. RESULTS: The average s.d. decreased by 35% from 2.1 to 1.32 between the first and final panel round. This change in average s.d., which indicated enhanced consensus, was significant at p < 0.0001. Following consensus, all statements were grouped into four categories based on average score: critical elements, important elements, less important elements and elements of uncertain importance. This grouping formed the basis for a guideline summary narrative. Results were generated in 3 mo, at low cost and with clear documentation of rationale. CONCLUSION: Through simple adaptations of this methodology, expert panels that develop practice guidelines can replace informal discussion with systematic scoring methods to rate the quality of evidence, generalizability to practice conditions, appropriate indications and strength of recommendations.