There is a lively controversy underway in France right now regarding drug safety. To a certain degree this parallels what happened in the United States about 10 years ago when the FDA’s role and effectiveness in tracking drug safety was thrown into question. This led to investigations and changes in the law starting in 2007 bringing us to where we are today.
France has had two large controversies (or scandals as many call them) erupt in the last couple of years. And now there is a third, though so far smaller issue.
The first was in regard to breast implants in which a company used non-medical grade silicone in manufacturing its breast implants. The French government warned 30,000 women who had this company’s implants to consider having them removed for fear that they were more likely to rupture than other implants. The implants were sold in other countries outside France. The head of the company was arrested and this continues to play out. If you are interested in more information (in English) have a look at the UK website of The Guardian (the UK newspaper) (https://www.guardian.co.uk/world/french-breast-implant-scandal). If you read French there is an enormous amount of information available if you do a google search.
The second controversy involves the drug Mediator which was marketed for use in diabetics who were overweight but was used by many non-diabetics as an appetite suppressant for weight loss. The Guardian (https://www.guardian.co.uk/world/2013/jan/06/france-scandal-weight-loss-drug), citing the French Health Ministry, notes that at least 500 people had fatal heart valve damage and that other studies put the number at about 2,000 deaths. Five million people received the drug over 13 years and other non-fatal adverse reactions have also been noted. It was withdrawn from the market in Spain, Italy and later France. It was never sold in the US or UK. The head of the French company that made the drug is now under investigator on suspicion of manslaughter and other investigations and trials are underway. The French drug agency was , shaken up, reconstituted and renamed with new personnel. This has “rocked” the French medical community according to The Lancet (https://www.thelancet.com/journals/lancet/article/PIIS0140-6736%2811%2960334-6/fulltext). As with the implants this scandal is continuing to play out in the press, in the government, in the courts and with the public.
The third issue is a report just out (https://mail.google.com/mail/u/0/?shva=1#inbox/13c3c1ff8c1739f8) in French in the newspaper Le Monde raising questions about the quality of pharmacovigilance data collection in France. The report notes that in a University Hospital in France there were 47 serious events (pulmonary emboli, deep vein thrombosis or cerebrovascular accidents) over 14 years in women taking “the pill” and who were 15-25 years of age. There were two deaths. These cases should have been reported to the French Drug Agency but apparently only three were so reported and none of the deaths were. According to Le Monde, the French Drug Agency (ANSM) confirmed the receipt of the report and the existence of very significant underreporting. The newspaper extrapolates these figures to indicate that over this 14 year period there would have been 3,000-3,500 serious AEs and about 150 deaths in women 15-25 years of age on the pill. This is part of another, bigger controversy in France and in Europe about whether third and fourth generation oral contraceptives should be on the market in the first place. See (in French):
So how does this play into the title of this posting about underreporting?
The point here is not to get into a discussion about the state of drug safety and PV in France (which, according to some sources is not very health as the Drug Agency has lost its credibility and physicians are not convinced that reporting AEs to the Agency is useful at all). Rather the point here is that there is significant underreporting in France and, surely, everywhere else.
There does indeed appear to be underreporting in spontaneous reporting systems in general. There is a small but noteworthy and consistent literature to this effect.
The Drug Safety Research Unit in Southampton, UK reported in 2006 on a systematic review they did of underreporting (Drug Saf. 2006;29(5):385-96). The abstract is available at: https://www.ncbi.nlm.nih.gov/pubmed/16689555?dopt=Citation. In this article the authors identified and included studies from many sources (hospital, general practice etc.). In 37 studies from 12 countries they were able to generate 43 numerical estimates of underreporting which they found to be 94% (median) with an interquartile range of 82-98%. In some studies the underreporting for all AEs was 95% and for serious/severe AEs 80% suggesting that physicians were more prone to report serious AEs but still relatively few were reported at all. Other literature is consistent with this conclusion.
So what can we say about underreporting?
Firstly, why is this so? Why underreport? Prof. Pierre Biron of the University of Montreal has cited several reasons for underreporting by clinicians (in Pharmacovigilance from A to Z by Cobert & Biron) including:
- The clinician is not be reimbursed or paid.
- Lack of interest or perhaps more accurately and less politely, “lethargy”.
- Uncertainty about causality.
- Belief in the innocence of the drug.
- Prescriber guilt on giving a patient a drug which hurt him or her.
- Fear of appearing foolish.
- Fear of revealing an off-label (unapproved) use of the drug.
- Social factors– interestingly in an old study from Europe, Denmark had a higher reporting rate than Italy. In Montreal, English speaking parents reported more AEs than French speaking parents.
There are probably many other factors but this gives a taste of the issues involved.
On the other side of the coin the recipients of the AEs (in particular, non-serious AEs) do not want to be bothered by cases that “don’t matter” – that is, those that don’t impact on the benefit-risk balance.
For example, FDA in its recent Guidance on SAE reporting in IND trials (see my earlier blog posts on this for a detailed analysis), has “requested” that companies not be too zealous in reporting cases that are clearly due to the underlying disease, frequent or common problems seen in the study population when not taking the study drug and study endpoints. FDA feels that these are “uninformative when reported as single events (i.e., without a comparison of the incidence of the event in treated and untreated subjects), they do not contribute meaningfully to the developing safety profile of an investigational drug or to human subject protection.” They are not a good use of FDA’s limited resources.
So although health agencies want to receive AEs they don’t want the well known reactions (e.g. penicillin skin rashes) or those AEs that don’t matter. What this really boils down to then is that the agencies want SAEs and the rare non-serious AEs that are new, different or perhaps harbingers of SAEs. On another level, this arguably implies that individual, isolated cases may not be worth capturing at all as causality is usually very hard to ascertain.
The final question to be asked is whether more reporting really matters. Will more complete data capture improve public health by whatever measures one wishes to use: hospital admissions for drug related problems, deaths on drugs, insurance payments for drug related issues, outcomes etc. etc.
We all presume that more is better and that more complete and better data capture will lead to better outcomes. Perhaps, but if we are only getting 5% of all drug related events now, will 100% capture be 20 times better. Probably not. There is surely a point where a good sample size is representative enough of what is really happening to allow the agencies, pharma companies and, of course, patients and prescribers to make good decisions.
Given that some of the major PV players (the US, EU) are having major financial difficulties and with an aging population using more and more drugs, perhaps we need a good rethink of how to collect post-marketing spontaneous safety data to best capture the useful data. There is much hope that when the US and EU really do have every inhabitant’s health data in a database, one can mine the drug safety information directly without relying on individual reports. Perhaps one day but until then we should figure out the most efficient way to handle spontaneous reporting. This is a good system which we can make better.