Risky Business

Vioxx, Bextra, Seroxat, Crestor, Baycol, Lotronex, MMR, HRT - hardly a month goes by when drug safety doesn't seem to be in the news. The questions are numerous:

ï Do Cox-2s cause more risk of heart damage than other pain relief drugs - and if so, is this a risk patients are willing to trade for relief from the agony of arthritis?

ï Are some statins more likely to cause muscle wasting than others - and, if so, is this a price patients are willing to accept to cut their chances of a heart attack?

ï Are you more likely to kill yourself as a result of anti-depressants or as a result of depression?

ï How large is the extra risk of breast cancer or heart disease for women taking HRT, and what choice are they allowed in saying they will carry this risk to be free of menopausal symptoms?

The queries focus on two areas - what are the facts (which are themselves often surrounded by uncertainty); and what do those facts mean for the individual patients?

Whether the answers come from industry or activist groups, patient charities or drug regulators, it is usually the case that only one of these halves is addressed. Often some degree of risk is set out, but what this means in the context of patients' lives is seldom clear. Alternatively, the risk of the treatment might be highlighted, but rarely the risk of the disease.

Silence is far from golden

Traditionally, the routine public position of pharmaceutical companies regarding side-effects has been we'd rather not talk about it. This in itself has been part of the problem. The industry has contributed to the very situation it increasingly complains about - a public perception that drugs should be 100 per cent safe for 100 per cent of people for 100 per cent of the time.

Therefore, when news of side-effects emerges in the media, patients feel betrayed and questions about the monitoring of all drugs are raised. Politicians get involved and drug regulators feel something must be done or they will be seen to be toothless. Then the lawyers start circling, sensing lucrative mass actions. How different the political and regulatory response and the litigation climate might be if patients said: We knew that - but it's a risk worth running to be free of the disease.

Honesty would be the best weapon against the charge that medicines have side-effects. It could be said that companies are being honest with the public in listing every possible side-effect on patient leaflets. But unless the frequency and severity of the side-effects are explained, it isn't clear how many patients get them and whether it matters if they do.

How to best communicate risk

Since it is in companies' best interests to honestly communicate risk, how should this be done? One important point to acknowledge at the outset is that a considerable body of work on communicating risk already exists and it seems pointless for individual companies, or individual product managers, to continually reinvent wheels.

For example, the British Medical Journal on 27 September 2003 devoted an entire issue to risk communication - it should be required reading for everyone in the industry on how do it, and how not. For instance, there is a rough kind of consensus on the factors that explain why one activity or treatment is perceived as riskier than another. There is a similar consensus on why these risks may or may not be picked up by the media. These summaries have been set down by Peter Bennett, from the Department of Health, as `fright factors' and `media triggers'. These have been published in a book Risk Communication and Public Health (Oxford Medical Publications 1999).

In the UK, some of the most innovative thinking on this subject has been by Sir Kenneth Calman, the former Chief Medical Officer, now Vice Chancellor of Durham University. In the USA, there are various professionals who do little else but communicate risk, but among these John Paling has written extensively on the subject. Both men have called for, and invented, visual scales to show the degree of risk an activity or treatment will produce. One of Calman's illustrative proposals was the creation of a `risk community' using a house, street, village, town or city to allow people to visualise the threat to them without using a numerical scale. A risk of one in a 100 means one person in a street will be affected; a risk of one in 1,000 equates to one person in a rural village; a risk of one in a million is analogous to one person in large city or an average English county.

This concept could be made even more immediate by using real places. Imagine an aerial shot of Sheffield, population 513,000, with just one red dot superimposed. That would vividly show the true meaning of a one in half a million risk. Likewise, Exeter (population 111,000) could be used to show a risk of one in 100,000, and Birmingham (population 977,000) to show one in a million. One of John Paling's ideas, which he calls the Paling Palette, is a chart showing 1,000 stick people. To illustrate a risk of four in a 1,000, four of the stick people are coloured in, and the remainder left blank.

Confidence intervals don't work

The public is not statistically literate, so doctors need to employ different tools beyond confidence intervals or p values to communicate drug risks.

Presentation is key. It's no good saying the evidence was there if it is not understood. Drug companies are very good at doing this - appearing puzzled and hurt that their data is not understood when they have made no effort to explain it to non-specialists. It's extremely debatable if any patient understands a risk of 0.04 per cent, but most will have a grasp of less than half a person in every 1,000 or one in 2,500.

Using a yardstick of 1,000, as opposed to 100 or 10,000 or 100,000, seems to make sense, as this is a number most people can easily grasp. Two American obstetricians, David Grimes and Gillian Snively, tested patients' understanding of medical risk with regard to genetic testing. They gave the same comparison for risks of Down's Syndrome at ages 35 and 40, in two different formats - 2.6 versus 8.9 per 1,000 women (rates) and one in 384 versus one in 112 (proportions). Three-quarters of the `rates' women understood the risk compared to around half of the `proportions' group.

Professionals also get it wrong

It is not just the public who struggle to conceptualise risk. Those who make purchasing decisions can come to very different conclusions depending on how the evidence is presented.

A study illustrating this was carried out by Thomas Fahey and colleagues, from Oxford's Department of Public Health Medicine. They published a study in the British Medical Journal in 1995 looking at how 183 health authority executives understood risk and benefit. The executives were given the same information on two programmes (mammography and cardiac rehabilitation) but it was presented in four ways - relative risk reduction, absolute risk reduction, proportion of event-free patients and number of patients needed to be treated to prevent an adverse effect.

Despite the data being identical, the way the effectiveness of the programmes was presented made a huge difference to the willingness to purchase them. The most `successful' approach to winning funds was to use relative risk reduction - but as Fahey and colleagues point out, this measure of efficacy gives no idea of the background event rate or the absolute benefits.

One classic example of the consequences of stressing relative risk over absolute was the contraceptive pill scare of 1995. One of us (Chris Mihill, working at that time as medical correspondent on The Guardian) inadvertently contributed to causing hundreds of abortions or unwanted children, and worry and uncertainty for thousands of women.

Journalists reported, accurately, what they were told by the Committee on the Safety of Medicines - that third generation contraceptive pills had double the risk of causing deep vein thrombosis compared to earlier pills.

If the CSM has said the risk had gone up from infinitesimally small to incredibly small, and was hugely outweighed by the risk of pregnancy, there probably wouldn't have been a pill scare.

Corporate reputation

On the face of it, a reluctance to provide cheap AIDS or malaria drugs in Africa, a willingness to use lawyers to spin out patent extensions and an aggressive pricing policy in countries where this is allowed might seem to have little to do with communicating drug safety. But they are inescapably intertwined.

Trust in the speaker is an essential ingredient when communicating risk, as numerous studies have found and common sense suggests. If the giver of the information is seen as untrustworthy, it is not surprising that the message is not believed.

Here is how the respected Lex column in The Financial Times summed up the industry recently: What was once known as the `ethical pharmaceutical industry' is now mentioned in the same breath as tobacco and asbestos stocks. Any manufacturer's claims are likely to be taken sceptically. But when assertions of safety are met with naked disbelief and claims that the industry is, in effect, covering up the poisoning of people to shore up profits, then lack of trust becomes a huge commercial threat.

Conclusions

Risk communication needs to be carried out by the pharmaceutical industry. At the moment this generally falls under the heading of crisis management and is not, on the whole, done well. Much more could be done to explain statistics on the risks and benefits of medicines, in an easily understandable way, to both doctors and patients.

the authors

Chris Mihill, managing director, Clew Communications, and
Stuart Mayell,account director, Clew Communications