In early August this year the FDA put out a MedWatch safety alert (www.fda.gov/Drugs/DrugSafety/ucm363041.htm) noting that acetaminophen has been associated with the risk of rare but serious skin reactions including Stevens-Johnson Syndrome (SJS), toxic epidermal necrolysis (TEN), and acute generalized exanthematous pustulosis (AGEP). These reactions can be fatal. FDA will require that a warning be added to the labels of prescription drugs containing acetaminophen and also is requesting that OTC manufacturers add a warning about serious skin reactions to their products. Acetaminophen (also called paracetamol) is a very common anti-pyretic, anti-pain medication sold OTC as well as in combination with other drugs both OTC and by prescription for colds, pain, fever and other indications.
Other recent changes in labeling to old drugs announced in June 2013 include valproate sodium, divalproex sodium, spironolactone, ursodiol, methylphenidate, terbinafine HCl and others (https://www.fda.gov/Safety/MedWatch/SafetyInformation/ucm359843.htm). Dozens of label changes can occur every month though not all are safety related.
The point here is not that acetaminophen can cause this problem but rather that this drug has been marketed since about 1951 in the US and here we have a new safety finding.
So this raises several questions:
- Why did it take over 60 years to recognize this problem and change the drug labeling?
- Is the reaction so rare that it was not seen or reported till recently?
- Is the system to pick up rare adverse events really so insensitive or is this an outlier?
- What else is out there that we don’t know about?
- Are there other drugs where bad things are discovered years or decades after approval?
The easiest of these questions to answer is the last one. The other questions we can speculate about but not easily answer.
The following chart from FDA gives a striking illustration of how common late safety findings are and their time course. The data show 2600 safety-related label changes made in FDA-approved labeling during the period of October 2002 to August 2005 (2 years and 10 months) for 1600 NDA/BLA products. We see over 1000 safety labeling changes in this short time period for drugs that have been on the market for up to 60 years! The bulk of the changes are found within 10 or 15 years of approval but others take much longer.

That is, many old and familiar drugs that are widely used (warfarin, digoxin, simvastatin, clopidogrel) have changes made ten, twenty, even fifty or more years after approval. This includes not only new adverse reactions (ARs) but also warnings, precautions, contraindications and even black box warnings!
The lessons from this dramatic slide include:
- New safety problems are continuously being picked up.
- Many are of critical importance in patient care.
- The complete safety profile of a drug is NEVER fully known.
- New drugs coming on the market, new uses, new diseases, changes in the population of patients being treated and other factors can change the safety profile of old drugs.
- Continued drug surveillance must be done.
This chart also raises other questions:
- Does this chart represent a success of the surveillance programs in place now or a failure?
That is, it is clearly a good thing that these problems have been picked up and that the product labeling has been changed. However, could these changes have been picked up years or decades earlier thus saving significant morbidity and mortality?
The answer is that surveillance programs are not really a full success. FDA, EMA and multiple other academic and governmental groups (and perhaps pharmaceutical companies too) are actively trying to find new ways to pick up safety data more quickly and with greater sensitivity and specificity. The FDA has multiple initiatives ongoing including Sentinel and Mini-Sentinel (https://www.fda.gov/Safety/FDAsSentinelInitiative/default.htm), including the creation of a database with over 100 million lives that is being used for data mining to identify new signals.
Other initiatives include REMS in the US and RMPs in the EU and elsewhere looking to discover and minimize safety problems using specified targeted tools including education, training, surveys, registries, clinical and epidemiologic trials, limited drug distribution, signing of informed consent etc. Whether these programs will improve drug surveillance remains to be seen.
I’ve discussed previously in these postings other aspects of drug safety that may contribute to better early warnings including the use of social media, crowd-sourcing, big data, collection of consumer and patient AEs and more.
- Is labeling, particularly paper labeling given to the patients and health care professionals, the best way to note and communicate safety changes or any other important drug changes?
Drug labels are now, in many cases, long and complicated making it hard to know what safety issues are in play for a particular drug. For example, have a look at the Zyprexa ® (olanzapine) labeling at https://pi.lilly.com/us/zyprexa-pi.pdf. This label is 33 pages long and the safety section runs approximately 18 pages. It is an excellent, complete and thorough label giving great detail. This is totally appropriate as this drug has significant safety issues and here is all the data laid out before you.
However, suppose a patient has, say, epistaxis while on this drug. There is no quick and easy way to know, looking at the labeling, if epistaxis has been seen and is suspected of being related to the drug. It is not highlighted anywhere but if one does an Adobe search of the pdf of the label one finds two mentions of epistaxis. Once in a listing of infrequent respiratory ARs seen in clinical trials (page 21) and once in a table of treatment emergent AEs (page 22).
So even though we have found two listings of epistaxis, the label really does not indicate whether Zyprexa causes epistaxis (“proving” causality is another exceedingly difficult issue in pharmacovigilance). All we can say is that it seems to be very rare presuming that the reporting of epistaxis by patients and physicians to Lilly and to FDA is being done with some regularity.
So, it is not totally clear that changes to labels are going to be seen and then used meaningfully by practitioners and patients unless the change is dramatic and the notification made “loudly” to the public and health care practitioners. Clearly we need better ways to communicate to the world. Maybe social media will be the answer.
Comments and conclusions: Drug safety surveillance is a critical function that must be done for the public health. A drug’s safety profile is never ever fully known and continued surveillance must be done throughout the life cycle of the drug even though some people in companies might say there are better uses of money and limited resources. We don’t have the optimal methodology to pick up safety issues nor do we have the best means to communicate it. New bad things continuously pop up for old drugs. Even though we don’t yet know best how to do pharmacovigilance, we must keep trying.