Pancreatitis During Drug Treatment: a Comparative Evaluation of Three Algorithms in Assessing the Causal Role of Drugs

Abstract

Background. None of the methods proposed for assessing the possible role of a drug in causing adverse events (AEs) has been accepted as the gold standard because of differences in the criteria used and in the weighing of each criterion. Accordingly, each AE has certain characteristics that differ from the others' such as background incidence, clinical features, etiologies, prognosis, etc. We felt that each method should be appropriate for use in evaluating each major type of AEs. Aim. We wanted to compare three widely used causality algorthms (Naranjo's, Karch and Lasagna's, and the French Imputability system) and to find out which of them is more appriate for assessing the causal role of a drug in cases of acute pancreatitis. Materials and Methods. 100 consecutive  cases of pancreatitis in Jansen's AE darbase were assessed using the three algorthms; a Global estimate made by trained experts was used as a reference. The outcomes from the three algorthms were statistically compared using Wilcoxon's matched-pairs Signed Rank Test, Spearman's Rank Correlation Coefficient Test P-values of less than 0.05 were considered significant. Results. We found significant differences and low correlation between each pair of algorithms (r=0.27-0.59, p < 0.05). The highest correlation was obtained from the comparison between the actual scores on Naranjo's algorithm and the Global estimate and the poorest correlation was obtained from the comparison between the French system and the latter. The results from Karch and Lasagna's algorithm were bound to be unrelated or conditional. Naranjo's algorithm yielded overall higher values than the Global estimate (1-level higher) suggesting a non-appropnate grouping of results by this method. Conclusion. Regarding the usefulness of the algorithms, it is conceivable Naranjo's algorithm is the best method in assessing pancreatitis cases given that the grouping of the scores into four qualitative categories are more revised.