The long-awaited changes to signal management are finally upon us, but what exactly are they, what we should be doing as marketing authorisation holders (MAHs), and can this enhance pharmacovigilance?
Signal detection is part of routine pharmacovigilance and is essential to ensure MAHs and competent authorities are aware of the latest information on a medicinal product’s benefits and risks. The EudraVigilance database is an important source of signals. Previously, only the regulatory authorities had access to this database, however, on 22 November 2017, the EMA launched the new EudraVigilance system, enabling MAH access via the EudraVigilance Data Analysis System (EVDAS). 1,2
Since access to Eudravigilance by MAHs was announced, the EMA and the European Commission have agreed on some transitional arrangements in order to streamline the implementation of this new process. Arrangements are detailed on the EMA’s signal management page1. In short, there will now be a pilot period of one year, which started on 22 February 2018. During the pilot period, only MAHs of the active substances contained in the List of Medicines under Additional Monitoring (fixed as of 25 October 2017 (Rev. 49)) are required to monitor Eudravigilance. This document is named, “List of active substances and combinations involved in the pilot on signal detection in EudraVigilance by marketing authorisation holders” and is also available on the EMA’s signal management page. The EMA clarifies in the aforementioned document that all medicinal products containing active substances or combinations included in the present list will be involved in the pilot, regardless of whether or not the medicinal products themselves are subject to additional monitoring. This means that, for example, even though the levofloxacin product under additional monitoring is an inhaled nebuliser solution, all MAHs with medicinal products that have levofloxacin as the active substance, regardless of formulation, will be required to partake in the pilot phase. Subsequently, this means that generic companies, as well as innovator companies, are involved in the pilot phase. Although generic companies might not be too eager to participate, this should be beneficial in the long run, as any difficulties faced by generic companies, as well as innovators, during the pilot phase, should be considered prior to full implementation.
Although only certain MAHs are required to participate in the pilot phase, all MAHs will have access to Eudravigilance data during this time. MAHs not involved in the pilot phase may choose to integrate the monitoring of Eudravigilance data into their own signal management processes. In fact, it is encouraged that this is done if MAHs are validating a signal from their own databases (in order to make use of all available data). However, whilst MAHs involved in the pilot phase are required to notify the EMA and competent authorities of signals detected in Eudravigilance via the new system of standalone notifications, MAHs who are not involved in the pilot phase, and are monitoring Eudravigilance by choice, should not notify the EMA of signals via this route.
After one year, the EMA will base the next phase of implementation on experience gained through the pilot phase. It is expected that either the pilot phase will be extended, or, mandatory monitoring of Eudravigilance by all MAHs will be implemented.
The EVDAS data
For those of you who have not yet initiated the use of EVDAS, the basic principles are3:
As you can see from the above, there is a lot of new information for MAHs to digest and it is unsurprising that there was industry wide panic and confusion over how to review this new and extensive dataset when it was first unveiled. However, the EMA have supported MAHs by providing webinars and information days in order to try and address these concerns.
In some aspects, clarity has been provided; the EMA confirmed that they will send the lists of all signals notified by MAHs, confirmed and non-confirmed, to all QPPVs the week prior to each Pharmacovigilance Risk Assessment Committee (PRAC) meeting. This list should be reviewed by MAHs to prevent sending a further signal notification for a previously identified signal, thereby preventing some duplication of work.
However, there is one aspect in which the EMA refuses to commit an unambiguous response; how should MAHs decide what DECs require further review? The response, which is included in the relevant guidelines4, is that scientific judgement should be used. Whilst initially the guidance (or lack thereof) may seem frustrating, after further consideration, the response is understandable and somewhat refreshing, as it gives MAHs the scope to decide how they want to analyse the data. Bearing in mind that any decisions made should be justified, and not outside of legislative requirements, and this justification should be documented.
Based on the above basic principles, the most sensible approach seems to be to export the eRMR only and access any further information required through the links in this document. But that is the easy part, the difficult part is how to analyse the output. There appear to be multiple alternative approaches, which could all be considered correct, provided the rationale is sound.
Although the eRMR has a large number of columns, it seems to be possible to initially focus on a select few. One significant column is the “SDR All” column – denoting which DEC are considered a signal of disproportionate reporting (SDR). However, it is important to understand the reasons behind why a DEC is being considered an SDR. As mentioned above, the eRMR provides us with a statistical measure, the ROR (a disproportionality measure). This is one of the criteria for a DEC to be considered a signal of disproportionate reporting (SDR). In simple terms, as explained in the Appendix of “Screening for adverse reactions in EudraVigilance2”, the ROR is the odds of a certain event occurring with your medicinal product, compared to the odds of the same event occurring with all other medicinal products in the database. If the ROR point estimate (result) is greater than one, it suggests that the product is causing an event. For example, if the ROR is equal to three, the odds of reports of this event with your medicinal product are three times higher than the odds of reports of this event among all other reports in the database.
However, this ROR point estimate is not actually included in the eRMR. Instead, the lower bound of the 95% confidence interval of the ROR is provided (ROR (-) on the eRMR). The reasoning behind this is explained fully in the EMA’s “Screening for adverse reactions in EudraVigilance2” guidelines. In short, the 95% confidence interval gives an indication of the precision of the estimate of the ROR. Therefore, when the statistic is based on few reports, this figure falls further below the point estimate and makes an SDR less likely. Using the ROR (-) results in a smaller and more precise set of SDRs.
The full set of criteria a DEC to be an SDR, as detailed in the EVDAS user manual3, are as below.
The following criteria must be met in at least one of the regions:
To the untrained eye, it might seem that the EMA have done all the work for us, we simply need to filter by SDR ‘yes’ and we have our signals. But this is not the case. As duly noted by the EMA “an SDR is not the same as a validated signal and on the other hand there could be real safety signals that do not show as SDRs at a certain point in time”3. So firstly, MAHs need to identify those SDRs which do not require validation, and then they need to identify any other DECs which do require validation.
Whilst it can be time consuming, it is not too difficult to identify SDRs which do not require validation. Criteria to consider are:
Finding DECs that are not SDRs but require validation, without looking at every single DEC is more of a minefield. Suggestions are to review:
This is without considering the paediatric/geriatric specific SDRs and running targeted eRMRs for specific System Organ Classes (SOCs) or Standardised MedDRA Queries (SMQs). It is down to the individual MAH to decide, aside from the SDRs, what criteria and threshold they are comfortable with dismissing DECs at, in order not to miss true signals.
As demonstrated, the sheer amount of data available from EVDAS poses challenges to MAHs when deciding how to pragmatically review the data. Diligent MAHs are striving to act correctly in order to safeguard public health, but require the process to be efficient in order not to place undue burden on their resources.
Improvement or Confusion – The Future
The EMA predict that their enhanced signal-detection and data-analysis tools to support safety monitoring directly by Member States and MAHs will enable better detection of new or changing safety issues, enabling rapid action to protect public health. EVDAS enables users to analyse safety data collected in EudraVigilance so that better-informed decisions can be made about the safety profile of medicinal products3.
The pilot phase has only just begun, so the above statement remains to be proven. It is hoped that continued close collaboration between the EMA and MAHs during the pilot phase will ensure the optimisation of EVDAS when it is rolled out for all active substances.
For more information, or to request a proposal, please e-mail: firstname.lastname@example.org