Is it really safety first?
The idea of ensuring that drugs are adequately safe is not very old. Legislation that covers drug safety to the standards we have come to expect today, only goes back some 40 years in the UK, with similar histories in other developed countries.
Although 40 years may sound a long time, it isn’t when we consider the ancient dictum ‘Primum non nocere’ - first, do no harm, said to originate 2,400 years ago with Hippocrates (but not in Greek, rather than its usual Latin form).
Today, no clinician can guarantee an absence of risk. But to judge Hippocrates by the standards of his time, and in an age when almost every ‘treatment’ available to a physician was of no benefit at all, I suppose he was talking sense.
Nothing is totally safe
He would make less sense today. We now have medicines that actually treat diseases, and drugs that have effects necessarily have side effects. I have heard it said (by a clinician) that a drug should be withdrawn from the market if even one patient is damaged by it.
My response to that was that it would make the practice of medicine impossible. We are now in an era of balancing risks versus benefits, but risk is a difficult concept to convey.
One problem is that the clinical trials that the regulators require us to do in order to obtain marketing approval, are designed to test efficacy not safety. To have a reasonably clear safety picture, trials would have to be several times the size. Having said that, recent data suggest that there is room for improvement regarding trial size, which is not good news for holders of clinical research budgets.
Hence the specialism of pharmacovigilance (PV) has emerged, and grown into a devourer of money and effort. The lay public probably have little perception of just how much effort is involved. Those with an interest in attacking the pharmaceutical industry frequently cite the US Institutes of Medicine study (2000) which apparently showed that prescription medicines are killing 250,000 Americans every year.
If this were true, it would of course be unacceptable, but the statistic does need to be placed in the context of how many people were helped by their medicines. To put it another way, if none of those people had placed themselves at risk by taking their medication, what would have happened? Well of course many would have died of their diseases, in millions.
Waiting for the brickbats
While difficult to convey as I have said, the industry still doesn’t try hard enough to explain it. Then, when something goes very badly wrong with pharmacovigilance, the companies concerned can easily look as if they are standing up against the wall waiting to be shot.
Two such companies right now are Roche and GlaxoSmithKline. To summarise, Roche underwent a PV inspection by the UK’s Medicines and Healthcare products Regulatory Agency (MHRA), which in May 2012 found that about 80,000 incidents arising in the US (SAEs) had not been evaluated for possible reporting to the regulators.
The vast majority of these originated in marketed products, outside clinical trials, but nevertheless some 600 SAEs from clinical trials had apparently not been reported. The GSK story is a bit more complex, and goes back many years.
The major reason for the huge fine which the company has sustained is for off-label promotion of Paxil, but there is also the matter of failing to report adverse events for Avandia.
Taking adverse events seriously
Now I am aware that I have some readers from outside the industry, so I will quickly explain that an SAE in a clinical trial has a very clear set of definitions. These include unscheduled hospitalisation, threat to life, disability, and death from any cause.
After deciding whether we actually have an SAE, the investigator and the sponsor have to decide whether it is related to the study treatment. In practice treatment-related SAEs are very much in the minority.
But what if the statisticians look at the unrelated events and find a trend that does link them to the treatment? This is the difference between looking at individual patient cases and evaluating the body of data from a population.
If it happened, report it
Thus we have a highly rigorous process for reporting all unwanted events, irrespective of whether the reporter thinks they had anything to do with the treatment. The 80,000 events at issue in the Roche case included over 15,000 deaths. Of course, in a large population a number of people are going to die during a period under study, for all sorts of reasons. But hitherto (as far as I can tell) we don’t know whether any of these was drug-related.
It appears that Roche has a very big job on its hands, to go through all these events and categorise them properly. For the minority that arose in clinical trials, there will be even more work for each event. If the proportion of deaths in the clinical trials population is similar to that overall, then we are looking at over 100 unreported fatal serious adverse events. Of course, I don’t have access to the data so that guess might be completely wrong.
But if I am even slightly right, there is the big problem of trying to retrieve data long after the event. SAEs are subject to expedited reporting, for this very reason. Strict time limits exist for the initial and follow-up reports.
If there are deaths among the events, there would be regulatory breaches for not reporting them in a timely fashion. The European Medicines Agency (EMA) acts as the co-ordinator for an EU-wide inspection programme, and is thus working with Roche on remedial action.
Heads in the sand?
Surprisingly, I can find nothing on the Roche website about all this. Searches using key words relevant to the story return zeroes.
There are of course lots of media releases containing upbeat messages about new products, which is absolutely fine, but surely some sort of reassurance for the public might be appropriate? Roche has responded to the EMA with assurances as to how seriously it is taking the matter, which I am sure is genuine.
Indeed, from my experience of drug safety personnel, they are among the most rigorous and conscientious of pharmaceutical professionals. I have no doubt that the people at Roche are as good as anywhere in the world, and that their fevered brows are locked into finding out what went wrong.
It’s not all bad
As yet none of us know what did go wrong, it would be a sin on my part to speculate. But the fact that it happened is agreed by all parties, and that is very bad news for an industry that increasingly comes under fire from consumer organisations.
High quality research has repeatedly shown that the risk:benefit ratio of prescription drugs is good, and moreover that better treatment compliance - actually getting patients to take their medicines - reduces healthcare costs. We have a great deal of which we can be proud, and handing the critics such ammunition as this is not going to encourage patients to manage their health effectively.
The world is increasingly open, to a large extent driven by the internet (I am not referring to Wikileaks!). The big debate in science now is open access, so that new research is not hidden behind a paywall. The open source movement in IT is unstoppable.
Pharma companies are moving (did I say kicking and screaming?) towards full publication of clinical trial results. Such openness applies when things go wrong as well as when they go right, and with websites the main shop window for all businesses, that’s where Roche should to my mind be posting some appropriate words.
Oh what a tangled web we weave
Let me turn briefly to the GSK case, not because there isn’t a wide choice of abuses to examine, but because it’s topical. Again, the major part of the breaches for which the company has been fined $3 billion fall into the sales and marketing category. And as with the Roche case, the scandal is a US one.
GSK has admitted that over several years, off-label promotion was routine, and American physicians were complicit in it. Although my offerings here always have an R&D theme, the translation of research data into marketing claims is a key issue and not to be separated out.
Initially it might have been thought that GSK reps were citing valid published research that simply hadn’t been included in the regulatory submissions, but it has also emerged that study results were most likely massaged by ghost writers. Even worse, there is the matter of under-reporting of safety data, particularly for Avandia, which gets us back to the main message of this piece.
Well, there are those more expert than I in evaluating the safety issues that currently face Roche and GSK - among others. At least GSK has addressed the matter on its website, with a media release on 2 July 2012. The tragedy is that the lay media can widely conclude that such cases invalidate most if not all of the good that pharmaceutical companies do.
So there may be two main lessons to be learned: hold individual employees accountable for ethical and regulatory compliance (nobody in either Roche or GSK is being prosecuted - yet), and when things do go wrong, face up to them publicly and present the true picture of risk versus benefit.
Les Rose is a freelance clinical scientist, specialising in project management consulting.