In recent years the number of adverse events being reported through the FDA Adverse Event Reporting System (FAERS) has generally increased as a result of broader reporting requirements and greater involvement of patients. The question is whether the number of events is increasing as a result of more reporting or because drugs are seemingly creating more adverse events?
I decided to take a look at drugs approved in recent years to see if the average number of adverse event reports for new drugs is rising? I took approvals in 2014, 2015 and 2016 and used a simplistic measure of the average number of reports per drug since launch. Crude? yes, but at least an indicator of potential patterns and certainly a precursor to a deeper analysis of the data.
Recently launched products
So, the simple things first; what was the average number of adverse event reports per product for the three years during the year of approval?
- 2014 – 120
- 2015 – 348
- 2016 – 544
I must immediately say that these figures do NOT mean that drugs approved in 2016 were 4.5 times more dangerous than drugs approved in 2014 (as we shall see in a moment, the opposite may be true). What is true is that there were significantly more reported AEs for the Class of 2016 than either the Class of 2015 or the Class of 2014.
By the end of the second year after launch approval the numbers had shifted:
- 2014 – 1198
- 2015 – 1881
Obviously, we don’t have data for the Class of 2016 yet, but the ratio has shifted from approximately 3:1 to 1.5:1, and this follows a pattern seen throughout the FAERS data. Early peak levels of reporting tend to level off and normalise – it is when products do not normalise that real concerns are raised.
In fact, when we look at the relative severity of reported events both the Classes of 2014 and 2015 show very similar trends in that they start above average (+14% and +13% respectively) but also both fall in year 2 (+13% and +10% – where relative severity uses Proportional Reporting Ratio against the baseline of all drugs).
More reports, but less serious
What is perhaps most interesting is that despite an above average number of reports per drug, the Class of 2016 has a Year 1 relative severity of -17% which indicates a broader spread of AEs being reported rather than a higher concentration in particular categories of event.
At Galileo Analytics we run a wide range of analyses across the FAERS data on a regular basis. Trends by drug, by adverse event, exploration of patterns and trends, co-morbidity and gender effects – all have a role in determining when the data show a real signal and when the data are actually noise.
Adverse events may be on the increase, however, it looks most likely that this is due to increased reporting rather than reduced drug safety. That said, we shall be continuously monitoring to identify what is really happening.