After the success of Bad Science, Ben Goldacre (doctor and debunker*) has now taken on the pharmaceuticals industry (“big pharma”) in his latest book Bad Pharma.
While admitting that “without medicines, there is no medicine”, he shows that the science behind new drugs is consistently distorted to further the interests of the industry.
The main way this is done is to routinely hide much of the data relevant to judging the efficacy and safety of new drugs. Over a quarter of the 365 pages of text is devoted to this missing data.
The gold standard behind any effective medicine is the randomised controlled trial (RCT) in a population of patients with the condition in question**. This follows a long process of preclinical work, testing the drug’s in-vitro activity, testing its activity and safety in animals, and then testing it on healthy human “volunteers”.
Ideally, the RCT would take a large number of patients with the condition and randomly assign them to receive either the proposed new drug or an indistinguishable placebo. Or, in a “head-to-head” RCT, the new drug would be compared with the existing treatment. The desired outcome is that the new drug is significantly better and no less safe than nothing or than the old drug.
It should be noted here that the “desired” outcome has a lot riding on it. If the old drug is about to go out of patent protection, its price will fall drastically. Or if the old drug is produced by a rival, this will be a way to cut into that market. It is very much desired that a replacement be found. The new drug may be only trivially different from the existing one (a “me-again” or “me-too” drug). As long as it can get a patent, it will do.
RCTs often do not give the desired results. But it’s not necessarily back to the lab to synthesise a new drug. What happens is that some of the data goes missing: guess which!
Goldacre gives an example from his own experience of prescribing the antidepressant Reboxetine. He had read the published trial data and found it to be positive. It had been approved by the UK’s drug regulator, the Medicines and Healthcare products Regulatory Agency (MHRA).***
However, in 2010, a group of researchers were able to assemble all Reboxetine trial data, published and unpublished. They carried out a systematic review and performed a meta-analysis (see box). Only one trial, the published one, showed Reboxetine to be better than a sugar pill; six more with nearly 10 times as many patients did not. Studies comparing it to other antidepressants were split: the published ones said it was as good as the others; studies on three times as many patients showed it to be worse. There was a similar bias when it came to reporting adverse side effects. This has resulted in a colossal waste of money and disappointment for patients.
There are many more attested examples of publication bias for which the causes are various. Many journals have a bias against research with negative results; some researchers self-censor and do not submit negative results; where research is funded by industry, there are often “gagging clauses” so that researchers are not free to publish their results without the permission of the company. Sometimes, negative results are re-interpreted with positive conclusions!
Attempts to identify unpublished trials rely on there being an accurate register of trials undertaken and punishment for non-publication. Neither of these has worked.
The drug regulators, who should be protecting patients, are ineffective or worse. Companies withhold data from them as well but, even when they don’t, regulators refuse to share unpublished data with bona fide researchers on spurious grounds of commercial secrecy and patient confidentiality.
In a notorious case, Cochrane reviewers wanted to see data on trials of diet drugs. The European Medicines Agency (EMA) refused to do this for three and a half years, despite being ordered to do so by the European Ombudsman.**** In the mean time, one of the drugs had been withdrawn on safety grounds (increased risk of serious psychiatric problems and suicide) by its manufacturer.
Well-run trials are very expensive, but bad ones are not, and there are plenty of examples of these.
One trick is to decide the focus of the trial after you’ve done it. By the laws of chance, the more outcomes you measure, the more likely it is that one will appear positive. You can then say that this is what you were investigating (like seeing where the ball went and putting the goalposts there!).
Choice of healthy volunteers for first-in-human trials is problematic. These are often poor people in the US (formerly prisoners) or, increasingly, in the Third World, for whom taking experimental drugs is an important source of income. Whether healthy or not, they are unlikely to be representative of the patient population who will take the drugs. There is also less ethical scrutiny of such trials. and consent is unlikely to be fully informed.
Developing new types of drugs is a lot more trouble than modifying existing ones and, despite their claims, drug companies concentrate on simple modifications of existing drugs. They actually spend much more on marketing their existing drugs than they do on R&D. Goldacre uses over a quarter of the book to show just how insidious companies are in getting to family doctors, consultants, professors, hospital managers and so on, subverting them with gifts, “education”, fees for “authoring” ghost-written articles in journals, not to mention funding patient groups, some of which then campaign for the health services to prescribe the latest cure.
One example of the latter was the breast cancer drug Herceptin. A vocal campaign took place in the press for it to be prescribed, making no mention of its effects on the heart which more or less balanced out its modest effect on breast cancer survival.
Goldacre is rightly indignant, pointing out that patients are suffering and health services are being ripped off. He makes a lot of useful suggestions for individual healthcare workers or members of the public to take action to expose this scandal. But he doesn’t mention the alternative of a nationalised drugs industry with a different set of priorities entirely. It’s not easy to see how to get from here to there but it’s certainly possible and desirable to confront big pharma and its allies and put forward the arguments for a different system.
Read this book!
** The earliest recorded of a controlled (though not randomised) trial is in the Bible (Daniel 1: 1-15).
*** The latest (August 2012) NICE guidance still states that it is similar in efficacy to other antidepressants. It costs about 20% more than the best in the NICE guidelines but, remember, it’s actually a placebo with adverse effects!
**** Reviewers first search systematically for all RCTs in a particular area, published or unpublished. This may involve contacting individual researchers or drug companies, as well as trying to track down articles in an enormous number of journals world wide.
Then they combine the results of all valid RCTs (meta-analysis) to get an overall picture of efficacy and adverse effects.
Systematic reviews, if updated regularly, can spot beneficial treatments as well. Studies on giving steroids to women having premature births, starting in 1972, were inconclusive but meta-analysis of all studies showed a significant advantage in preventing death. Since this knowledge only became available in 1989, it is certain that many preventable deaths occurred.
Many systematic reviews are carried out by the Cochrane Collaboration (http://www.cochrane.org/), a non-profit organisation dedicated to evidence-based medicine (I myself have taken part in a systematic review for Cochrane; see http://onlinelibrary.wiley.com/doi/