Adverse Drug Reactions— A Computer-assisted Application of Correspondence Analysis for Automatic Causality Assessment
Ch. Bastin, P. Wertheimer, Jan Venulet · Methods of Information in Medicine · 1984
Summary An assessment forra comprising 27 questions relating to adverse drug reactions and causal relationship was completed for 204 subjects. Questions providing no information were disregarded and answers to the remaining questions recorded and subjected to a correspondence analysis. The correspondence analysis returned a set of mappings together with new quantitative variables to which discriminant analysis was applied aiming at a future allocation of individuals to causality groups. Results of the analysis, based on data from the 204 subjects, demonstrated that in 61.3% of cases assessment of causality obtained with discriminant analysis was identical with the assessment made by a medical evaluator. The allocation rule was subsequently applied to data from 100 new case reports and, in a similar percentage (60%) of cases, allocation to the causality groups coincided with the evaluator’s assessment.