SYSTEMATIC REVIEW-BASED FINDINGS TO GUIDE THE DEVELOPMENT AND REPORTING OF CONSENSUS-BASED CRITERIA FOR INAPPROPRIATE PRESCRIBING IN OLDER ADULTS

Abstract Background Formal consensus development methods are widely employed to develop inappropriate medication assessment criteria for older adults. However, standards for conducting and reporting these methods are lacking. This study aims to summarise findings on consensus attainment across published explicit criteria. Methods A systematic literature review of criteria for assessing inappropriate medications in older adults was conducted in PubMed, CINAHL, EMBASE and Scopus (1991 to 2020). Search terms pertained to the concepts of inappropriate prescribing, criteria development/validation, and older patients (aged 65 years or more). Articles were included for the current subanalysis if they represented a consensus based explicit list, assessing appropriateness of medications in older adults. Results From 73 tools identified in the systematic review, 58 tools (79%) had been developed and/or validated by formal consensus methods. 51 criteria sets were reported as having been developed using Delphi (n = 30) and modified Delphi (n = 21) methods. However, most (29/30) of the reported Delphi methodologies had actually been modified by introducing an initial literature-based questionnaire. The criteria for achieving consensus between experts were pre-specified for 42 criteria sets, while the remainder did not report or prespecify consensus definitions. Where available, consensus definitions varied considerably. Arbitrary levels of proximity to a central tendency (e.g. mean +/− 95% confidence interval, median +/− interquartile range) defined consensus in 22 criteria. Seven more criteria followed the RAND/UCLA consensus definition (i.e. three-point region containing the median) (1). Only one set addressed consistency of agreement between rounds. Likert scales or categories for ranking ranged from three-point to 100-point scales. Conclusion The development and reporting of formal consensus-based criteria varies substantially, particularly in distinguishing between Delphi methods and their modifications. This highlights the need for standards for conducting and reporting consensus attainment. 1. Fitch K, Bernstein SJ, Aguilar MD, et al. The RAND/UCLA Appropriateness Method User’s Manual: RAND Corporation, 2001.