It is true that where they can be carried out, double blind trials are usually the most effective way of performing a clinical trial. But it is wrong to think that such a test is foolproof or that it is the only effective scientific test. One cannot eliminate every possible effect and it is misleading of statisticians to follow Sir Ronald Fisher in suggesting that randomisation can achieve that. Randomisation will do no better than, and will often do worse than, an informed attempt to eliminate or take account of conceivable distortions.
Nor should we think that double blind testing is all that is available. There are techniques used throughout science which show that this is not the case. The problem Ms Toynbee raises is that of patients who want to know what treatment they are receiving, typically demanding that they receive the treatment under test. In many cases it is perfectly possible to carry out a trial in which they do so and the results are compared to historical data; where necessary a scientific estimate of the placebo effect can be taken into account. This may not be statistically convenient, but it should be remembered that statistics is not the ruler of science but its servant.
Dr ALEXANDER BIRD
Department of Philosophy,
University of Edinburgh