How can evidence-based medicine (EBM) methodology support drug withdrawals?

Withdrawing drugs over safety concerns requires careful analysis of the health benefits and risks by regulators.¹ But while data on adverse reactions is rightly used to help with these decisions, there is currently no additional modelling on the positive impact regulatory action could have on public health.

Our research paper, which is being submitted for publication this summer, aims to create an evidence-based medicine (EBM) methodology that can assess this positive impact, to further help regulators determine appropriate risk minimisation measures.

Supporting regulators

Monitoring and studying the side effects of licensed medicines is an essential part of drug development to ensure public safety. Known as pharmacovigilance, our epidemiological research records and investigates adverse drug reactions when medicines are used by the general public. This is usually on a much larger scale than is possible in clinical trials and can uncover less common side effects.²

National regulators use pharmacovigilance research when deciding whether risk minimisation measures should be imposed. The most severe measures are temporary or permanent product withdrawals.³

To aid decisions about potential product withdrawals, we set out to develop a new EBM methodology to show the positive impact this could have; with quantifiable modelling of the estimated number of adverse reactions that would be prevented if a specific drug was withdrawn from use.

Study focus

Our research looked at products that were voluntarily withdrawn, or had their licence revoked or suspended within the EU for a safety reason between 2012 and 2016. It focused on data from France, Germany and the UK.

Six products were withdrawn or revoked for safety reasons during the study period.

The withdrawn drugs were:

• meprobamate – a sedative used to treat anxiety
• nicotinic acid/laropiprant – used to lower cholesterol
• almitrine – a respiratory stimulant
• calcitonin – able to controls calcium levels, it was used for conditions such as osteoporosis.

The revoked drugs were:

• some formulations of metoclopramide – an anti-sickness medication
• some formulations of domperidone – used to prevent sickness and nausea.

Three products were suspended:

• cyproterone acetate/ethinylestradiol (Diane-35 and generics) – birth control
• tetrazepam – a muscle relaxant
• ketoconazole – an antifungal medicine.

Real-world data and results

Real-world data was gathered from IQVIA Medical Research Data (IMRD) electronic health records (eHR) databases for France, Germany and the UK to find out how many patients were using each drug during the 12 months before and after regulatory action. An estimate of how many people were exposed to the drug each year prior to withdrawal or suspension was calculated.

Then European Medicines Agency (EMA) records were used to identify adverse drug reactions, and a systematic review was carried out to quantify the risk of adverse drug reactions in patients exposed to each withdrawn drug.  

The types of adverse drug reactions examined included low blood pressure, diabetes, cancer, bleeding, involuntary movements and weight loss.

An absolute risk for each withdrawn drug was calculated using published literature. This risk, expressed as a percentage of people using the drug, was based on the estimated annual number of people exposed and the number of adverse reactions.

An estimate of the number of adverse reactions prevented as a result of each product withdrawal, revocation or suspension was also calculated based on the exposure and absolute risk figures. The total was an estimated 3,020 per year.

Limitations

There were, of course, limitations to the study. Most notably, we found that usage of the withdrawn drugs did not always drop to zero within 12 months of regulatory action. This could have been because other strengths and formulations that remained on the market were misclassified. In the case of the anxiety medicine meprobamate, we know this was gradually withdrawn over the duration of our study period. Meanwhile, some suspended drugs were returned to the market.

The dataset for Germany included medicine exposures in secondary care (hospital patients), but this was not available for the databases of France and the UK. Therefore, it is possible that exposure estimates in France and the UK are less accurate, and the public health impact of medicine withdrawals and suspensions in these populations may have been underestimated.

Similarly, the data we used does not record over-the-counter medicines used by patients, which could impact outcomes. While pooling data from different sources, as per our methodology for calculating absolute risk of each event, may not always be appropriate.

Wider benefits and future research

One of the key benefits of our research is that the new EBM methodology can be applied to any other drug where usage can be measured. It can be employed with any database where exposure can be estimated and the results can be stratified by demographics and patient subgroups, such as older people, if data allows.

Pharmacovigilance measures are never imposed lightly. But quantifiable evidence of the predicted positive impact of regulatory actions on public health will help regulators make these decisions in future.

This is one of several studies underway at the UK’s independent Drug Safety Research Unit (DSRU) looking at new methodologies to quantify the impact of pharmacovigilance measures. We plan to broaden our research to look at the impact of other marketing authorisation restrictions, such as limiting use to specific age groups, and the public health impact of people switching to alternative drugs when their current treatment is withdrawn.

Our research was carried out under the auspices of the DSRU’s Institute of Pharmacovigilance Sciences and the European Network of Centres of Pharmacoepidemiology and Pharmacovigilance (ENCePP).

DSRU’s Institute focuses on EBM methodology research, including expanding existing benefit-risk evaluation methods, as well as developing new methods, such as the application of artificial intelligence in pharmacovigilance. Researchers also study the long-term effects of medicines and vaccines, as well as the safety of new and advanced products.

References

1. Pharmacovigilance – how the MHRA monitors the safety of medicines [Internet]. MHRA; [cited June 2022]. Available from: https://assets.publishing.service.gov.uk/government/…
2. Boyko E. Observational research — opportunities and limitations. Journal of Diabetes and its Complications. 2013;27(6):642-648.
3. The Human Medicines Regulations 2012, UK statutory instruments, 2012 No 1916, PART 5, Revocation, variation and suspension of marketing authorisation. Available from: https://www.legislation.gov.uk/uksi/2012/1916/part/5/crossheading…