Detecting pharmacovigilance signals: Data mining of benzodiazepines adverse reactions notified to FDA reporting system

• Autores: Beatriz Proy Vega, Anselmo Martínez Blázquez, Julián Solís García del Pozo, Joaquín Jordán Bueso
• Localización: European journal of clinical pharmacy: atención farmacéutica, ISSN 2385-409X, Vol. 20, Nº. 6, 2018, págs. 294-303
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Since patients have easier access to information, thanks to the growing development of technology, their concern about everything related to drugs, especially adverse drug reactions, has increased. Resulting of this, authorities and health agencies are increasingly interested in drug safety and early detection of signals. Quantitative signals detection methods are useful for detecting pharmacovigilance signals, which can be analyzed to establish causality between an adverse event and a drug. Goal of this study was to deepen in signal detection of adverse drug reactions potentially associated to benzodiazepines, through the analysis of spontaneous notifications reported to the United States pharmacovigilance database (FAERS); after a previous work of signals detection of only cutaneous adverse reactions associated to this family of drugs, by applying data mining technique and different algorithms. All this, initially motivated by the market withdrawal of medicines containing tetrazepam in 2013, due to serious adverse skin reactions after the analysis of the suspected adverse reactions reported to the French pharmacovigilance database. The most frequently reported adverse reaction to FAERS associated to benzodiazepines marketed in the United States were analyzed. Our statistical analysis revealed the following signals generated by different algorithmic methods of disproportionality analysis: when Proportional Reporting Ratio applied, only a single signal for estazolam and "Cardiac disorders» was detected. Reporting Odd Ratio algorithm detected 31, the largest number of signals. With Information Component, 24 signals were detected. Finally, using Empirical Bayesian Geometric Mean algorithm, two signals were detected: one for clobazam and "Nervous system disorders», and another for quazepam and "Lesions, intoxications and procedural complications». All these adverse events are described in the Summary of Product Characteristic of these drugs, in different frequencies. A signal is a suggestion that there is a statistical association between a medication and an event. Calculating signals does not provide an estimation of the incidence of adverse reaction or information about whether a statistical association between a drug and an adverse reaction is causal, but highlights potential drug problems that draw the attention of the pharmacovigilance community for investigating in more detail to maximize safety of the drug