Reflections on randomised controlled trials in surgery

Abstract

The stereotype of surgeons as unthinking and uncaring technologists persists. For example, in the Lancet Lifeline on Jan 30, 1999, Peter Kandela, a general practitioner, explains that his most influential teacher was Tom Maine, a psychoanalyst, who introduced him to the Balint technique and converted him a surgeon into a real doctor. However, most surgeons spend more time in clinics than in the operating theatre, and physicians and general practitioners do not have a monopoly on contemplative practice. A component of the role of today's contemplative surgeon is to recognise limitation of knowledge, to reject complacency and dogma, and to take part in clinical research. Another popular misunderstanding concerns the nature of research. An uneducated outsider might assume that surgical research is about techniques and technology, whereas in practice most of it relates to the diseases for which patients are conventionally referred to surgeons. Certainly there is much research into techniques and technology, not the least of which has been the development of minimally invasive procedures. Although important, such research addresses small (how to do it) questions, not the large (why to do it) ones that challenge the biology of disease (figure). Apart from that of emergency surgery for lifethreatening disorders, the benefit from most surgical interventions is on quality rather than length of life. As such, an important advance in surgical research has been the development of psychometric tools for the adequate assessment of quality of life. The purpose of this paper is to highlight the role of randomised controlled trials in the development of surgery. However, randomised comparisons are unnecessary when the natural history of the disease is well established and the effect of the operation is spectacular. For example, with total hip replacement for severe osterarthritis, the striking difference between being crippled by the disorder and being able to walk unaided and free of pain did not need to be proved by a randomised trial or even to be assessed by a formal measurement of quality of life. Unfortunately, these rare successes have been more than outweighed by failures. 2 When the natural history of the disease is unpredictable, random error or systematic bias is more than likely to confound the results. The effect of surgery must then be assessed in randomised controlled trials. Such trials are necessary when the effect is modest, such as the difference between two techniques of hip replacement, or the ability of an operation to extend the life of a patient with cancer by a year or so.