Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Book Reviews

Stephen Senn. Dicing with death: chance, risk and health
Dylan Evans. Placebo The Belief Effect

Stephen Senn. Dicing with death: chance, risk and health. Cambridge University Press, Cambridge. ISBN 0 521 83259 4 (paperback). Pp251. £14.99.

"I am going to try and convince you that when it comes to making decisions and scientific inferences, if you can't count you don't count". That is how Stephen Senn prefaces his book. It is partly about statistics, partly about statisticians, but mostly about thinking and how to think.

While many books on statistics deal with numbers, Senn deals with people. The stars are the men and women who, over the past several hundred years, have developed our understanding of numbers and chance, what today we call statistics. Almost none of them started out as statisticians, but as doctors, chemists, churchmen or some other profession, but who were drawn inexorably to play with numbers as a way of understanding the truth, whatever that is.

For those of us to whom the word Bernoulli means a (now old fashioned) form of computer disk, it is fascinating to learn about the six generations of Swiss Bernoullis, most of whom were shooed away from mathematics but couldn't keep away. Or Poisson of the distribution that actually wasn't his. Or Gene Glass of Nebraska, who credited "Eysenck with the invention of meta-analysis by anti-thesis". Statisticians are people with a particular interest and talent, and we should give thanks to them for helping us understand evidence when we see it.

Equations do not figure a lot, and when they do Senn helpfully tells us when we can afford to skip them. His medium is words, not numbers, and when numbers appear they are as examples to help the flow of the words, rather than obstruct the flow of ideas. A text book this is not. It probably won't help in passing exams, but boy will it help dealing with problems when they come along.

Few readers will walk away and recall the minutiae of arguments about probability. You will understand it when you read it, but five minutes later only the framework will be left. But it is the framework that is the important bit. Bandolier despairs at the number of times people fail to realise the lack of importance of trivial chance findings compared with veritable mountains of evidence. MMR is a good example, and the discussion on MMR in the final chapter is great reading.

Senn is urbane, charming, and often funny. He uses quotes a lot, and is not averse to using them against his own values. A favourite is the following: "Meta-analyst: one who thinks that if manure is piled high enough it will smell like roses". Useful for teaching.

So who is this book for? Anyone seeking understanding, teacher or student, professional or public. You might need a dictionary occasionally, but otherwise you can read this in bed or on the beach, and it won't be out of place. Bandolier loved it.


Dylan Evans Placebo. The Belief Effect. Harper Collins London. ISBN 0 00 712612 3 £16.99 pp204

This book is about the placebo effect, the intriguing phenomenon itself and about the underlying mechanism. The structure is first to look at placebos in the context of medical trials, as (negative) controls to help scientists establish the difference between therapeutic effect and a control intervention. The control intervention is there to provide a comparison, and while one might think that a placebo pill or injection should do very little, in fact it can have effect. Evans then proposes a mechanism by which placebo has its effect, and follows with a chapter on the foundation of the belief that triggers physiological events through which placebo might work. The evolutionary context follows, posing questions as to how and why a placebo effect may have evolved. Later chapters focus on alternative medicine and psychotherapy, asking whether they have any greater effect than placebo. Lastly it discusses ethical issues in the context of clinical trials and in clinical practice.

Henry Beecher was, as Evans suggests, the most influential writer on placebo in the mid twentieth century, and the spin on his 1955 paper was that a third of us respond to placebo and the extent of that response was a third of the maximum possible effect. We know now that the proportion of us who respond in any given context varies widely, and that the extent of the response also varies.

These points are vital to the genesis of better clinical trials, and the bottom line is that much of the variation is due to chance, and is minimised by making trials as large as possible. This point does not come through nearly strongly enough. While Evans rightly stresses the importance of statistics in making sense of placebo effects, he often digresses into intriguing results from individual small trials. This is a major hazard of placebo. People have wasted years trying to work out why the placebo effect in a particular trial might be greater in one context than another. Reality is that there is probably a ‘true' underlying value for the proportion responding and the extent of the response will only be established with data from many thousands of patients. Individual trials may produce high or low values for placebo response due to chance.

This does not help us determine how placebo has its effect. Evans argues that placebo works by suppressing acute phase responses, our response to injury. He invokes tenuous biochemistry and immunology. Work on whether reversing the body's endogenous opioids, endorphins, with the opioid antagonist naloxone, could reverse the placebo effect has not been satisfactorily replicated. An acute phase response theory is testable, but we may be too simplistic in attempting to discover the (a) mechanism of placebo.

There is also the belief explanation. Beecher tried to categorise placebo responders, who tended to be older, female, believers in authority, and church-attending if not God-believing. Charisma surrounding many alternative therapies and therapists clearly should work to enhance the belief. The consistent failure to show credible therapeutic effect with many of them stands, but it needs to be said that if a therapy makes people feel better that is no bad thing. Just don't badge it as bona fide therapeutic effect. The charisma, the belief, the placebo response, may make you feel better.

Evans defends the legitimacy of placebos in clinical trials in non life threatening conditions to show whether or not treatment is better than placebo. Patients are free to withdraw and have other treatment at any time. It is, after all, unethical to recruit patients into studies that cannot deliver an answer, and without placebo controls that would be the case in many contexts.

Placebo response is intriguing, and this book should bring that interest to a wider audience. There are caveats, but as a one-stop well thought out compilation this is the place to start.

previous story