[HOPIPRAC: Pharmacovigilance interface for detecting cases of adverse events in a hospital data warehouse].

Introduction: Spontaneous reporting is the reference method for collecting data on adverse drug reactions (ADRs). However, it remains insufficient. Text search in the electronic medical record has opened up a new way of collecting ADRs. However, these methods have certain limitations. Data warehouses constitute a rich digital environment bringing together all the informations in an electronic medical record. The Clinical Data Centre (CDC) at Brest University Hospital has developed a pharmacovigilance tool, the HOPIPRAC interface, which provides autonomy in querying the entire hospital database to detect cases of suspected ADRs from computerised patient records.

Method: The data source used was the CHU de Brest data warehouse. The IT tool was developed and tested between 2015 and 2018. A validation test of the tool was proposed, based on the choice of a positive signal from Brest's previous experience (valvulopathy with Benfluorex). The hospital data warehouse query tool can also be used to carry out other forms of interrogation to identify new risks, particularly with new drugs on the market, such as Nivolumab. The tool was tested to identify and analyse bullous toxidermia, described as the most severe and very often of a drug origin.

Results: The validation test identified 98% of the historical cases recorded in the national pharmacovigilance database, as well as potential new cases not reported to the Brest CRPV. The nivolumab cohort identified 34 ADR cases. Of these cases, 82% were serious and some were unexpected at the time of extraction. The toxidermia query identified 137 cases, 56% of which had not been reported to the Brest CRPV. With regard to Stevens Johnson Syndrome (SJS), 28% of cases were associated with antineoplastics (7 cases), in particular tyrosine kinase inhibitors (sorafenib, vemurafenib, regorafenib) derived from advanced therapies, although this risk is described as rare in the summaries of product characteristics for these 3 drugs.

Discussion: The HOPIPRAC interface is an innovative IT tool for assessing drug safety. It gives pharmacovigilants independent access to electronic hospital medical data. Its applications are: 1- amplification of a pharmacovigilance signal obtained by spontaneous notification; 2- monitoring of adverse reactions in cohorts of patients exposed to a specific drug; and 3- mapping of certain drug toxicities within hospital data. However, a number of limitations has been identified: the difficulty of identifying rare adverse events due to the low population representativeness of the surveyed warehouse, and the absence of certain data relevant to the exploration of adverse drug reactions.

Conclusion: The HOPIPRAC interface gives pharmacovigilants autonomy in monitoring drug risk in healthcare institutions, in addition to spontaneous notification. Querying a more advanced warehouse that collects multimodal data, and deploying it on a wider regional or even national scale, will help to overcome some of the limitations of the current warehouse.