Professor Sir Aziz Sheikh: Towards next-generation pharmacovigilance capabilities for the UK

Foreword

For decades, the UK has been recognised for its leadership in pharmacovigilance, underpinned by the MHRA’s Yellow Card scheme, which continues to play a vital role in detecting important safety signals early.

As medicines and medical technologies become more complex, used by increasingly diverse populations, and reach patients at greater speed, there are exciting opportunities to further strengthen our safety monitoring systems.

As outlined by Professor Aziz Sheikh in the latest of our strategy blog series, the COVID-19 pandemic demonstrated the potential of a modern, agile safety system. We rapidly digitised the Yellow Card reporting platform, introduced AI to manage large volumes of reports, and combined enhanced passive surveillance with active monitoring of vaccine safety using national usage data and anonymised electronic health records.

With the launch of the Health Data Research Service on the horizon, we now have the opportunity to build on these achievements and apply the same approach across all medicines and medical products – enhancing the UK’s already world-leading, real-time surveillance capabilities and further safeguarding patient health.

Professor Sir Aziz Sheikh is Pro-Vice-Chancellor and Head of the Nuffield Department of Primary Care Health Sciences at the University of Oxford. A leading global authority on primary care, public health and the use of real-world data, he has played a central role in advancing the UK’s capabilities in drug safety, digital health and evidence-based policy.

Guest Blog: Professor Sir Aziz Sheikh

Our current approach to pharmacovigilance has proven invaluable, but it remains largely passive and could be greatly enhanced.  Whilst we’ve moved on from relying on clinicians to tear out a Yellow Card from the back of the British National Formulary (BNF) to sophisticated reporting platforms, the voluntary nature of reporting of cases falls far short of what we should expect, meaning that detecting and assessing safety signals takes longer than we want.– We need a more proactive and systematic effort to identify medication (and related products) associated harm to inform approaches to reducing such risks at the earliest possible opportunity. With our digitised health infrastructure now covering virtually the entire UK population, we have a major opportunity to emerge as a world-leader in next-generation pharmacovigilance capabilities.

The process of digitising the NHS has been long and arduous, but is – thankfully – now nearing completion.  This means that the real-world data emerging out of the back-end of these digitised health record systems can now be linked across the entire care continuum using the unique NHS number (or equivalent in the UK’s devolved administrations) providing an unprecedented foundation from which to build next generation pharmacovigilance systems.  Such systems would allow, for example, the routine running of analyses to estimate the incidence of known adverse events identified during drug development or licensing processes, identification of sub-populations who may be at particularly high risk of drug adverse events, and running formal epidemiological analyses to investigate potential adverse events reported through the Yellow Card scheme or other routes.  An example of such an analysis involved assessing the risk of varenicline, a potent smoking cessation aid, after small-scale analyses suggested that this may increase the risk of cardiovascular and neuropsychiatric risks resulting in a black box warning being issued by the US Food and Drug Administration (FDA) .  Our national retrospective cohort study undertaken in the large, representative QRisk database however failed to confirm these heightened risks when compared with other pharmacological smoking cessation thereby supporting the continued use of varenicline.    

The FDA Sentinel System is now one the largest electronic health record-based database in the world undertaking federated drug safety analyses on up to 170 million patients from around 70 health systems across the US.  Whilst we would not be able to compete with the FDA Sentinel System in terms of numbers, with our UK population of 68 million people we would nonetheless be very well placed to run analyses on a scale that could detect most rare safety signals.  Unlike the US, we also have the distinct advantage of being able to run such safety analyses on the total population thus reducing the risk of bias and maximising generalisability.

Once data pipelines have been built, and data curation and analyses processes have been standardised, there is the potential to automate the most common analyses such that the system of searching for drug related adverse events is in effect “always on” working away in the background.  

Whilst such aspirations might seem something of a pipe-dream at the moment, there has in recent years been considerable progress on a number of the essential elements, including: systematic use of the NHS number, developments in cloud-based dataset hosting, progress in deterministic and probabilistic data linkage approaches, and proof-of principle established during the COVID-19 pandemic that the UK can deliver high quality vaccine and drug safety work – at pace – including on the entire UK population.

The imminent launch of the Health Data Research Service  (HDRS), with substantial political and financial support from the UK Government, offers an unprecedented opportunity to build on the UK’s outstanding strengths in pharmacovigilance to create near real-time whole population capability to support the safe and effective use of drugs to advance human health.