The alluring nature of technology is a constant presence within 21st century healthcare, frequently resulting in an unquestioning faith that if we only invest in it sufficiently it will cure all our ills. All too often, however, our enthusiasm turns a blind eye to two significant stumbling blocks to the realisation of these ambitions – the irritating habit of a true understanding of statistics to dampen proceedings, and the unerring tendency for human nature to stuff things up when people don’t do as they are told. This is no clearer than in the arena of genomic sequencing.
There is no doubt that gene sequencing has come a long way. Building on the lofty ambitions of the Human Genome Project, companies are on the verge of offering individual whole genome sequencing to patients. At a mere $1,000 a time, this promises delivery of the ultimate in personalised medicine – the knowledge of your entire genetic profile and all the inherent health risks and benefits within your individual blueprint. Aside from the ethical problems this introduces – such as whether you really want to know that you have an increased risk of an untreatable disease; or whether you have to declare the results to your insurance company; or what if it throws up the risk of a serious illness which you never develop in your lifetime – we should also question the value of the information it will reveal.
Proponents of genomic testing often state the positive impact such a test might have on lifestyle, for instance – let us just examine that one claim. The argument goes that if you find out that your genes give you an increased risk of, say, heart disease, then you would be more likely to look at cardiovascular risk factors such as stopping smoking, eating healthily and taking up regular exercise, and by this means the test would save lives. There are three major steps that would have to be taken before a life is saved, however, and they are worth a bit of detailed scrutiny.
The first is the chance that the genetic test will throw up information on cardiovascular risk over and above what is already currently available. Anyone undertaking the test will have one of three potential results. The first would be a low risk outcome – probably the majority, since none of us will consider that we have an increased risk unless our risk is significantly above average. These patients may feel reassured, but they will not benefit from the test beyond this, and may even increase their risk by becoming complacent about their lifestyle choices. The second will be those who have an increased risk, but where the information gleaned has added no new information. We already have several indicators of cardiovascular risk – family history, ethnicity, cholesterol, blood sugar and blood pressure are all readily accounted for and far cheaper than genomic testing. If genetic tests simply confirm the risk we already know about, then there will be no added benefit. The third group will have learned something new about their risk; assuming this information is reliable then they could potentially benefit from lifestyle changes. We have no way of knowing how large this group will be, but can be sure that it will not be that big – 10% seems to me to be a generous estimate.
Of these 10%, the next question is whether or not respondents actually achieve a sustained change in lifestyle. Again people will fall into three camps – and this is where human nature starts to play a significant part. The first will not benefit, as they will be those who do not need to change their lifestyle, because they are already health conscious and adhere to healthy living advice. Since those requesting genomic testing will be a subset of our society who are particularly interested in their health, this is likely to be a large group; on the other hand, many of those who would most benefit from lifestyle changes will be some of the last to undertake such a test. A second group will receive advice to change their lifestyle, but not be able or willing to adhere to it. Anyone who has tried to achieve behavioural change will know that this will also be a sizeable group; changes in behaviour are hard to achieve, and a simple knowledge of risk is frequently a poor motivator. Most people with newly diagnosed diabetes or hypertension fail to lose weight or quit smoking, despite the new information on risk that they have acquired. The third camp will use the test to stimulate change, and may actually benefit – let us be generous and estimate this group as 20%. It would probably be far fewer.
20% of 10% means that only 2% of those undertaking the test may make a change to their lifestyle – what we now need to know is the absolute reduction in mortality by making such a change. Absolute risk reductions are where the statistics really bite, for they are some of the least sexy numbers in medicine as they are usually embarrassingly small. If we assume that the actual risk of those making a change to their lifestyle is at the level where other cardiovascular interventions are usually recommended – which is a 20% risk of developing cardiovascular disease over 10 years (a far higher figure than the risk of actually dying from the disease), and we generously estimate the relative risk reduction of the lifestyle measures to be in the order of 25%, then we are left with an absolute risk reduction of 5%.
5% of 2% is only 0.1%, or 1 in 1,000 of those ordering the $1,000 test avoiding developing cardiovascular disease as a result, while the other 999 do not benefit from these claims about improved lifestyle outcomes. I may have got my estimates wrong, of course, but experience tells us that when you have to divide a number three times over, you usually end up with a small number.
Of course those in the industry will protest at such figures and argue the many other benefits of knowing your own gene sequence – but if you were being paid $1,000 a time, wouldn’t you?