Addictovigilance is a safety monitoring targeted at substances with potential for abuse and dependence. This vigilance was involved during the period of COVID-19 epidemic due to the significant changes in access to drugs and psychological disruption caused by the

Adverse drug reaction (ADR) reporting is a major component of drug safety monitoring; its input will, however, only be optimized if systems can manage to deal with its tremendous flow of information, based primarily on unstructured text fields. The aim of this study was to develop an automated system allowing to code ADRs from patient reports. Our system was based on a knowledge base about drugs, enriched by supervised machine learning (ML) models trained on patients reporting data. To train our models, we selected all cases of ADRs reported by patients to a French Pharmacovigilance Centre through a national web-portal between March 2017 and March 2019 (n = 2,058 reports). We tested both conventional ML models and deep-learning models. We performed an external validation using a dataset constituted of a random sample of ADRs reported to the Marseille Pharmacovigilance Centre over the same period (n = 187). Here, we show that regarding area under the curve (AUC) and F-measure, the best model to identify ADRs was gradient boosting trees (LGBM), with an AUC of 0.93 (0.92-0.94) and F-measure of 0.72 (0.68-0.75). This model was run for external validation showing an AUC of 0.91 and a F-measure of 0.58. We evaluated an artificial intelligence pipeline that was found able to learn how to identify correctly ADRs from unstructured data. This result allowed us to start a new study using more data to further improve our performance and offer a tool that is useful in practice to efficiently manage drug safety information.

Since spontaneous reporting of adverse drug reactions depends on the physician's opinion of the relationship between the drug and the adverse event, we compared physicians' opinions with the scores obtained by the causality assessment method used in France. During a 2 month period, all physicians who reported adverse drug reactions (ADRs) to our pharmacovigilance centre expressed their opinions on the causal link by means of visual analogue scales. ADR reports were then assessed with the French causality assessment method by a clinical pharmacologist who was blind to physicians' opinions. The assessment by both physicians and the standardized method was performed for 75 ADR cases involving 120 drugs. Physicians used a wide range of assessments, with a preponderance of extreme scores, resulting in a U-shaped distribution, while the standardized method gave generally low scores. Scores given by physicians were very high (causality considered very likely or likely) in 60% of cases and very low (causality considered unlikely or dubious/possible) in 32% of cases. Scores obtained using the causality assessment method were low (causality dubious/possible) in 89% of cases and causality considered likely in only 11 cases, essentially in cases with positive rechallenge. Complete agreement occurred in only 6% of cases. Adding complete agreement and minor discrepancies raised the percentage to 49%.

The study was conducted in the context of a national pharmacovigilance follow-up for which the Centre de Pharmacovigilance de Bordeaux was appointed by the French Medicines Agency (Agence Nationale de Sécurité des Médicaments, ANSM). The ANSM had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication. This publication represents the views of the authors and does not necessarily represent the opinion of the ANSM.