Dr Phil Hammond, GP, lecturer and presenter of BBC2's Trust Me, I'm a Doctor

But will it work, doctor?

Explaining clinical effectiveness to patients isn't easy, which is why most of us don't bother. We say 'You must take the brown inhaler' or 'It's probably a virus but here's some azithromycin just in case'.

However, July 5th marks the start of 'Evidence-based medicine speak week', when all doctors are encouraged to grasp the numeracy nettle and talk like textbooks.

For example, say you're seeing a patient who has had a heart attack and is found to have high cholesterol levels. Now, instead of wading in with 'You must swallow a statin', you have to offer him a menu of risk statements:

  • If you take this drug every day for 5 years, your risk of death will be reduced by nearly 40%
  • If you take this drug every day for 5 years, your risk of death will fall from 8% to 5%
  • If 33 people like you take this drug every day for 5 years, one death will be prevented, but I don't know whether it will be yours
  • If 100 people like you are given no treatment for 5 five years, 92 will live and 8 will die. Whether you are one of the 92 or one of the 8, I don't know. Then, if 100 people like you take this drug every day for 5 years, 95 will live and 5 will die. Again, I do not know whether you will be one of the 95 or the 5.
  • If you swallowed 1825 tablets at a rate of one a day for 5 years and a prescription cost of £400, your risk of death would fall by 0.03. The statistically significant side-effects are muscle damage, headache, abdominal pain, nausea, vomiting, hair loss, anaemia, dizziness, depression, nerve damage, hepatitis, jaundice, pancreatitis and hypersensitivity syndrome.

What's immediately evident from the evidence-based approach is that even when you try to quote 'scientific facts' to patients (or rather probable truths), it's extraordinarily difficult to give unbiased information.

Just quoting relative risk reductions, as newspapers and drug companies are so fond of doing, is clearly misleading. But when you enter absolute risk territory and apply the results of large trials to individuals, most drugs sound much less effective than you'd like them to.

Sure, for a whole population we could save thousands of lives, but for any one patient the odds sound much less favourable.

Very few drugs (pain killers and Viagra excepted) have a 'number needed to treat' of 2 or less, and for primary prevention the NNTs are often ridiculously high.

For example, for every 850 middle-aged men (36–64 years) with mildly raised blood pressure taking antihypertensive drugs, one stroke will be prevented each year.

Or, if you prefer, the number of tablets needed to prevent one stroke in mildly hypertensive middle-aged men would, if laid end to end, go four times round the surgery buildings and into the car park.

My guess is that numeracy in medicine won't catch on. Everyone hates uncertainty, and although science tries to minimise it, patients prefer paternalistic reassurance to rational thought. And although many doctors have at least A level maths, we soon forget it in the consulting room and just pluck figures out of the air.

John Allen Paulos, an American mathematician, recalls a 20-minute conversation with a doctor where he stated that the procedure he was contemplating had:

  • A 1 in a million risk associated with it
  • Was 99% safe
  • Usually went quite well.

Predicting the future is hard enough at the best of times, but in the heat of a 6-minute consultation, it's near impossible. Far better to fall back on the words of George Bernard Shaw: "A medical degree is no substitute for clairvoyance".

  • Dr Phil Hammond is author of the best-seller Trust Me, I'm a Doctor (£9.99, phone orders 0500 418419).

Guidelines in Practice, July 1999, Volume 2
© 1999 MGP Ltd
further information | subscribe