EFFECTIVE, safe and non-addictive: that is how the new generation of antidepressants was billed in the early days. Now these drugs are the focus of a fierce dispute. Some patients and doctors claim they are of questionable efficacy and can induce suicidal thoughts. Advocates, including the companies that make them, insist these medicines have helped millions of people, and that withholding them would do more harm than good.
It would be easier to judge which side was right if all the relevant information about the drugs were publicly available. But it isn鈥檛. Believe it or not, the law does not oblige companies to disclose the findings of their research on licensed medicines, and scientists, doctors, patients and even public organisations have no legal right to inspect the evidence that led regulators to license drugs.
This problem is serious because under-reporting of clinical research is biased and can be lethal. For years, patients who had suffered a heart attack were prescribed drugs to prevent heart rhythm abnormalities. Yet by 1990, more than a decade after these drugs were introduced, it has been estimated that they were killing more Americans every year than died in action in the Vietnam war. Had some of the early evidence suggesting the drugs were lethal been published, this catastrophe might have been prevented.
Advertisement
Biased under-reporting of clinical trials has been recognised as a problem for a long time. In 1980, the Finnish researcher Elina Hemminki showed that among studies submitted for new drug licences those that had documented side effects were less likely than others to be published subsequently. A quarter of a century later, Hans Melander at the Swedish Medical Products Agency looked at published information about new antidepressants and compared it with information about the same drugs submitted by the companies to the drug regulatory authorities. Yet again, they exposed biased reporting favouring the new drugs. And only last week, Hemminki and Klim McPherson of the University of Oxford revealed biased under-reporting of the effects of hormone replacement therapy (British Medical Journal, vol 328, p 518).
What is even more alarming is that governments and ethics committees have acquiesced in this state of affairs. Consider official reactions to an analysis of the effects of human albumin solution. Although this resuscitation fluid has been widely used for more than half a century, none of the clinical trials has been large enough to yield reliable estimates of its effects on mortality. When the published and unpublished trials were reviewed systematically for the first time in 1998, the data suggested albumin might be killing about 6 per cent of patients treated with it. Moreover, the investigators showed that the increased risk of death could have been spotted a quarter of a century earlier, had the UK鈥檚 Medicines Control Agency (now the Medicines and Healthcare products Regulatory Agency) systematically reviewed the accumulating evidence. Yet the MCA denied the public access to the information upon which they had re-licensed the product in 1993.
Official acquiescence is bad enough, but acquiescence sometimes extends to complicity. In August 1994, I was asked by the UK Department of Health鈥檚 Advisory Committee on NHS Drugs whom to approach to commission a systematic review of the effects of evening primrose oil on eczema. On my advice it invited a professor of dermatology and a professor of pharmacy to take on the task. The review revealed little evidence to justify the cost of the drug to the National Health Service, then more than 拢7 million and rising. The manufacturers responded by asking the DoH not to disclose or discuss its contents without their agreement. Incredibly, officials accepted these demands, and evening primrose oil remained on the NHS drugs list. Five years later, following yet another review, the MCA withdrew the marketing authorisation for the drug. However, as with albumin, the raw results underlying this and their earlier decisions have not been published.
Three lessons seem clear. First, if companies have not studied effects on the key outcomes 鈥� like death 鈥� that matter to patients, regulators should grant only provisional licences. Second, evidence from successive clinical trials must be accumulated and reviewed systematically. Third, biased reporting of clinical trials must be outlawed.
A mere six drug companies have voluntarily adopted the Good Publication Practice guidelines () which have been developed within the industry. Given the importance of these companies to many national economies, there is nothing fundamentally wrong with governments wanting to support them. But permitting companies to keep the effects of licensed drugs secret? Ignoring the evidence that biased reporting of trials harms patients and wastes money? These surely cannot be in the public鈥檚 interest.