With the release of the new Eudravigilance system on 22nd November 2017, the European Medicines Agency (EMA) have introduced greater access to adverse event data stored in the central repository. Furthermore, on the same day a new update (Revision 1) to Module IX – Signal Management of the Guideline on good pharmacovigilance practices (GVP) came into effect.
Together these developments will likely have a significant impact on the way in which a Marketing Authorisation Holder (MAH) might approach its signal detection and management obligations.
The requirement to demonstrate a robust approach to signal detection and management is not new, the initial introduction of Module IX to the GVP legislature in 2012 represented a major sea change in drug safety – it placed greater emphasis on the MAH to establish and maintain a comprehensive signal strategy, over and above the underlying obligation for regular single case and aggregate reporting. The MAH is expected to support a risk based approach, using globally available data, across the whole product lifecycle (thus incorporating clinical and spontaneous adverse event data, along with other relevant sources).
Module IX has been actively re-enforced via regulatory inspection since its inception – for instance, in the UK, the Medicines & Healthcare products Regulatory Agency (MHRA) have clearly prioritised inspection of signal detection arrangements. In thirty-four inspections between April 2015 and March 2016, findings relating to signal detection were reported as follows:
- Critical findings – in 9% of inspections
- Major findings – in 12% of inspections
MHRA findings in relation to signal detection, such as no formal procedures, no formal periodic review, and poor documentation of procedures, were the third most common category, after the usual suspects of product labelling and Quality Management System (QMS) issues respectively.
The penalties for failing to comply with pharmacovigilance requirements, including signal detection management, are stringent with fines up to 5% of the previous year’s turnover and personal liability for non-compliance.
What is the impact of Module IX Revision 1?
The new revision is clearly designed to further emphasise the MAH’s role in the continuous identification of validated signals, however, the burden of assessment of these signals appears to move more towards the regulatory agencies.
An obligation is placed upon MAHs to monitor EudraVigilance data and inform EMA and national competent authorities of validated signals. It should be noted that this requirement will initially only impact MAHs of active substances contained in medicinal products included in the ‘List of medicinal products under additional monitoring’ and will only come into effect from 22 February 2018 (after a grace period of three months for implementation).
Other companies will also now be able to access data from Eudravigilance and it seems likely that the scope of responsibility will be extended after the planned one year transition period defined for the enhanced monitoring.
The regularity of monitoring is expected to be at least every six months, but the timeframe should be defined according to risk and the chosen frequency should certainly be justified in signal detection and management documentation.
Revision 1 also includes a revised definition and process for handling emerging safety issues, migrating an expectation previously described in Module VI – Collection, management and submission of reports of suspected adverse reactions to medicinal products. MAHs are obliged to report such issues as soon as possible, and no later than three working days after establishing that a validated signal or safety issue exists. Interestingly, the notation in Module IX Revision 1 removes all specific example sources that should be considered outside of the regular Individual Case Safety Report (ICSR) submission framework. The sources were previously cited explicitly in Module VI.
Again, this seems to be indicative of the regulators’ desire to place broader responsibility on the MAH to consider all available information without providing prescriptive scenarios.
The update to Module IX also focuses the fundamentals, including the roles and responsibilities, of a solid signal management process – to define how validated signals should be processed. NB: It is not the intention of this review to consider the structure of such a procedure.
New addendum to cover statistics
An additional document, Module IX Addendum I – Methodological Aspects of Signal Detection from Spontaneous Reports of Suspected Adverse Reactions, also comes into effect on 22nd November 2017. The addendum extends and updates previous advice relating to the use of, amongst other methods, statistical metrics for the evaluation of adverse event data for the purposes of signal detection.
The role of statistics in signal detection
It is important to take a holistic approach to signal detection. Relevant information is available from many different sources or varying types, and critically there is no single method that can be relied upon to reveal every signal.
The importance of systematic, qualitative review of every ICSR should not be underestimated. A recognition that a signal may exist as just a single reported drug-event combination (DEC) should always be held in mind. It is essential to implement a standard dictionary of terms for consistent evaluation and to consider a watch list of key event signs, such as those designated as always serious. Likewise, expert review of cases can be very effective in spotting changes in severity of signs associated with certain drugs.
That said, when looking for evidence of changes in frequency of those DECs, statistical methods and tools can prove extremely useful, especially as case volumes increase. The basic principle of using statistics in signal detection is to find pairs of specific drugs and events which occur more commonly than would be expected if there was no causal association – this is called a disproportionate relationship. There are various methods and calculations that can be used to evaluate disproportionality and these each have their own merits and limitations, they all require a background, or denominator, dataset.
Background data for statistical analysis
One pre-requisite for a disproportionality analysis is a broad background dataset which is used to define the normal incidence of a given sign with any given drug, such that an abnormal incidence for a specific drug can be identified.
Different statistical measures require different volumes of background data to be effective; for example, for Proportional Reporting Ratio (PRR) calculations it is typically stated that the background dataset should be in excess of 30,000 DECs. For Bayesian statistics an even larger dataset of more than 500,000 DECs is optimal.
For many small and medium sized MAHs this presents an immediate problem, which is compounded by the fact that the product range is typically limited – perhaps just representing a few pharmacological classes or therapeutic areas. So regardless of absolute case volume, the spread of adverse event data held in the company’s own database is unlikely to be sufficient to form a representative background.
The improved access to Eudravigilance and its extensive case history, granted following the November update, offers a great opportunity to all companies – it should now be possible to extract additional adverse event data into statistical tools to provide that essential background.
The role of PV-Analyser in a signal detection strategy
Using an automated approach to signal detection can complement a manual case review methodology to ensure that the MAH doesn’t just meet regulatory obligations, but is able to attain exceptional levels of product insight.
The Ennov PV application PV-Analyser offers great potential for any company to integrate a rich, easy-to-use, statistical analysis platform into its signal detection program. PV-Analyser does not require the core Ennov pharmacovigilance solution to operate, as it can be populated with case data in a standard E2B format. Furthermore the EMA adverse event data may be imported for use as background data (in addition to cases from other sources, such as FDA’s Open FDA initiative).
The application, which is designed for use out-of-the-box by business users, offers a broad array of statistics as well as powerful data cubing and visualisation functionality. By being able to run quick efficient reviews of adverse event data, the MAH can quickly and routinely examine the dataset for potential signals that may require further evaluation.
Now thanks to the greater availability of additional background data (made possible in part by this EMA update) and the features of PV-Analyser, comprehensive statistical analysis becomes a viable option for all MAHs looking to meet an expanding set of obligations in Europe and beyond, regardless of their case volume.
Module IX (Rev 1) : http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2017/10/WC500236408.pdf
Module IX Addendum 1 : http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2017/10/WC500236405.pdf
Public access to adverse event database : http://www.adrreports.eu/en/index.html
Nic has 18 years of experience with pharmacovigilance systems, in the UK and the USA, managing the product range and working with customers to implement and improve operations. As the director of pharmacovigilance products, he ensures that Ennov’s pharmacovigilance offer meets the highest international standards.