WEB-RADR @ Royal College of Physicians
We are now half way into the WEB-RADR project and this was the second General Assembly meeting arranged. The agenda was to report on the different work package and to share thoughts and experiences from the first 18 months of the project. The meeting where held in the library of the Royal College of Physicians. The library was a very suitable location with all the knowledge collected but to some extent hidden in all the books, most of them so old that you need permission to read them. In the same way we now try to unlock all the knowledge hidden in Social Media streams on drug related problems!
The day was introduced by outlining the agenda and talking generally about the WEB-RADR project and how far it has come. A new explanation to the WEB-RADR acronym was also introduced. So far it has been recognized as WEB – Recognizing Adverse Drug Reactions but the extended explanation is as follows:
Web-Recognizing Adverse Drug Reactions
Embracing New Technologies
Both Public and Private Partners involved
Reporting via mobile Apps vs Established Reporting Schemas
Algorithms and Analytics
Develop a policy framework
Reshape the Pharmacovigilance world!
The first half of the day the focus was on general governance of the project and the Social Media scanning which is one of two major activities. It was presented what has been done in relation to the scanning of social media streams and how that scanning is being done, from gathering, via filtering and curation to statistics. During the presentation it was demonstrated how data collected from Social Media could be automatically compared in real time with traditional spontaneous reporting data from VigiBase using the VigiAccess API. It was striking to see the differences highlighted in reporting patterns.
Some refined methods used to improve the matching and reducing the number of false positives in social media streams was presented after tea-break. Those methods involve the use of WHODrug for drug name matching and reaction term verbatim being reported to VigiBase. For example event matching scores was improved from 14 to 30% using reaction verbatim in VigiBase compared to using only terms from the MedDRA terminology.
In the discussions a lot of interest where around how industry should handle the reports from Social Media. Should they be ignored, analysed separately, together with the normal spontaneous reports or seen as part of a bigger picture to take into account. Most participants leaned to the latter option. One specific problem with the social media data was raised. There is a requirement to anonymize the collected data and a requirement for a Spontaneous Report is that it must have an identifiable patient.
In the afternoon the focus was shifted to mobile reporting apps. There are now three apps developed in the scope of the project whereof two, the MHRA Yellow Card App and the LAREB Bijwerkingen App is in production. The HALMED app is currently being acceptance tested with a planned release first half of May 2016.
There were lots of interest around the app and then especially the information provided to the user by the app. The reporting possibility is of course there and very important for the WEB-RADR project but to make it interesting for patients and HCPs to install the app, and continue to use it, it must provide useful information that is regularly updated. There was also a great interest in extending the app to a global scale and some concrete pilots for Sub Saharan Africa was discussed.
In the end of the day some more general work was presented:
Assessment of utility of the App
Assessment of the utility of Social Media data
Look at how to integrate ADR data from Multiple Data Sources
An interesting aspect of the last point is how to handle the fact that different sources of ADRs have different origins and types of reporters that have different health profiles and therefore different reporting patterns. One example of this is that to be able to compare data from the mobile app and VigiBase for instance, we must take into account the reporter types. This means that for example the VigiAccess tool should also have the possibility to filter by reporter type to make comparisons meaningful.
A very interesting and fruitful day was wrapped up with the following take home messages:
WEB-RADR IS delivering!
Using Social Media is not a question about IF but HOW!
How can we think international?
Mobile app technologies is perhaps not primarily about reporting but more about communication!
Now we all look forward to the second half of the project where there will be more focus on analysis of the data collected to see what it can bring, or not, to the very important and never ending work on drug safety with the goal of safer patients.