Premium
Utility of an Effect Size Analysis for Communicating Treatment Effectiveness: A Case Study of Cholinesterase Inhibitors for A lzheimer's Disease
Author(s) -
Peters Kevin R.
Publication year - 2013
Publication title -
journal of the american geriatrics society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.992
H-Index - 232
eISSN - 1532-5415
pISSN - 0002-8614
DOI - 10.1111/jgs.12308
Subject(s) - rivastigmine , medicine , number needed to harm , donepezil , galantamine , placebo , randomized controlled trial , confidence interval , sample size determination , cognition , adverse effect , number needed to treat , relative risk , psychiatry , dementia , disease , statistics , pathology , alternative medicine , mathematics
Objectives To highlight the utility of using an effect size analysis to communicate the effectiveness of treatment interventions. Design Secondary analysis. Setting Previously published systematic review on cholinesterase inhibitors (Ch EI s) in A lzheimer's disease. Participants Individuals with mild to moderate A lzheimer's disease. Intervention Six‐month randomized controlled trials involving a placebo group and a Ch EI group (donepezil, galantamine, or rivastigmine). Measurements Cognitive function was assessed according to performance on the cognition subscale of the A lzheimer's D isease A ssessment S cale ( ADAS ‐Cog). Global F unction was quantified using the C linician's I nterview‐ B ased I mpression of C hange— P lus ( CIBIC ‐Plus). Harm was defined as withdrawal from a trial because of an adverse event. Several effect size indices were computed based on these domains: the success rate difference ( SRD ), the harm rate difference ( HRD ), the number needed to treat ( NNT ) or harm ( NNH ), and the area under the curve ( AUC ). Harm:benefit ratios were also computed to compare effect size indices across domains of function. Results In terms of benefit, the NNT for cognition ranged from 4 to 14 (corresponding AUC values: 0.64–0.54), and the NNT for global function ranged from 6 to 100 (corresponding AUC 0.59–0.51). In terms of harm, the NNH ranged from 6 to 20 (corresponding AUC 0.58–0.53). Only one of the four studies had favorable harm:benefit ratios in both the cognition and global function domains. Conclusion Effect size indices should be reported in clinical trials because they provide important insight into the clinical meaningfulness of results. Additional benefit is gained by comparing effect size indices across domains of function to reveal harm:benefit ratios.