Evidence in Medicine. Iain K. Crombie

of the UK Medical Research Council, used random numbers to allocate patients to treatments. This methodological advance was proposed by the medical statistician Professor (later Sir) Austin Bradford Hill [40]. It was first used in a large field trial that assessed the effectiveness of a vaccine for whooping cough [41]. Although this study began in 1944, it was not published until 1951. The second trial, of streptomycin for pulmonary tuberculosis, became the most highly acclaimed study in the history of treatment evaluation. It began in 1946, but was the first to be published, in 1948 [42]. The third trial involved a large‐scale field trial of an antihistaminic drug (thonzylamine) for the prevention of the common cold [43].

As well as being published first, the streptomycin trial provided a major advance in the treatment of a feared disease, tuberculosis: it reduced the fatality rate at six months from 27% to 7% and also reduced the severity of disease among survivors. An editorial that accompanied the paper identified the advantage of individual randomisation over alternate allocation: it prevented a patient being included or rejected, based on whether the next treatment was to be antibiotic or control [44]. For example, if the doctor thought that the drug would not be effective in seriously ill patients, they might not be included in the study if they were scheduled to receive the active treatment. This would only need to happen a few times to bias the results of the study.

      In addition to the use of random numbers to allocate patients to treatments, these three trials stand out for two other reasons. Patients were recruited from multiple centres to provide sufficient participants to be able to draw firm conclusions. The researchers also made considerable efforts to ensure that the participants, and the clinicians measuring the outcomes, were unaware of which treatment the patients received. This prevented bias in the reporting of symptoms by participants, and by those recording the outcomes: in modern terminology, it was double blind.

      The landmark streptomycin trial in tuberculosis was followed by another study on pulmonary tuberculosis, published two years later [45]. It compared three treatments: streptomycin, another drug, para‐amino‐salicylic acid (PAS), and a combination of these two drugs. The same methodology was used as in the first streptomycin trial. The combination therapy had the best outcome, with streptomycin coming second. More importantly the combined treatment led to a much lower frequency of bacterial resistance to streptomycin. This study has been credited with leading to the maxim ‘never treat active tuberculosis with a single agent’, which is now the standard for managing this disease [46]. The clinical benefits apart, this set of four rigorous studies supported by the Medical Research Council inaugurated the era of high quality clinical trials.

THE NEED FOR BLINDING

A major concern of several trials in the middle of the twentieth century was to ensure that the patients and clinical observers were unaware of how treatments were allocated [39, 41, 43]. This would prevent knowledge of who received which treatment from influencing the outcome of the study. The idea was not new; it featured in studies to evaluate a treatment called animal magnetism. This treatment was championed in the late 1700s by Franz Anton Mesmer, who believed he could impart magnetic energy and thereby cure a wide range of illnesses [47]. Mesmer achieved great fame, and a lucrative medical practice in Paris. This popularity was of such concern to other doctors, and to the government, that they persuaded Louis XVI to establish a Royal Commission to evaluate the claims of cures and dramatic effects [47, 48]. The Commission conducted a series of studies in which participants either thought they were being magnetised (when they were not), or thought they were not subjected to magnetism (when they were). The findings were convincing. Participants only reported benefits when they (falsely) believed they were being treated: ‘the imagination is the real cause of the effects attributed to magnetism’ [47]. Following publication of the Commission's report, Mesmer was ridiculed, and animal magnetism was abandoned in France.

      A more recent example of the importance of blinding is the evaluation of a surgical technique, internal‐mammary artery‐ligation, for the relief of angina symptoms. Several reports in the 1950s had claimed that the operation provided considerable relief of symptoms [49, 50]. This prompted two groups of researchers to carry out controlled trials to evaluate the effectiveness of the surgery. Patients were randomly allocated to have artery ligation, or to a control group which received a sham operation involving only a skin incision. The patients, and the cardiologists who evaluated the outcomes, were blind to treatment group. The ligation operation provided no benefit, as most patients in both the treatment group and in the control group reported significant improvement in symptoms [49, 50]. The authors concluded that these claims were most likely a psychological response to undergoing surgery.

      The response to a sham treatment is known as a placebo effect. Understanding of the psychological and physiological factors underlying the placebo response has advanced greatly in recent years [51, 52]. A consistent finding is that patients who have high expectations of their treatment usually experience improvements in symptoms. If patients were aware of their treatment allocation, only those in the active group would have the high expectations. Concealment of treatment allocation could prevent this bias from creating difference between the groups.

CONCLUSION

      Obtaining evidence on treatment effectiveness is a challenging business. As Passamani remarked in 1991, ‘The history of medicine is richly endowed with therapies that were widely used and then shown to be ineffective or frankly toxic’ [53]. A similar view was expressed by the celebrated American physician, Oliver Wendell Holmes in 1860, ‘if the whole materia medica, as now used, could be sunk to the bottom of the sea, it would be all the better for mankind – and all the worse for the fishes’ [54]. These may seem somewhat jaundiced views, but they reflect the large proportion of ineffective and possibly harmful treatments that were once used. Even in the early years of the twentieth century many ineffective treatments were widely used [55], and some treatments of little value continue to be used today [56]. Concern about this has led to a recent international campaign, ‘Choosing Wisely’, to reduce the use of ineffective or harmful treatments [57].

      This chapter has presented examples of different approaches used to identify potentially effective treatments. Reliance on theories of disease processes is often unreliable and can result in harmful treatments being used. Careful observation of treatment outcomes in a series of patients can, if the benefits are immediate and substantial, identify effective treatments. Comparisons of groups of patients given different treatments are often more insightful, but are vulnerable to the criticism that the groups might not be similar at baseline. As the eminent French physician P C A Louis pointed out in 1834, ‘it is necessary to account for differences of age, sex, temperament, physical condition, natural history of the disease’ [58]. The use of groups constructed to be similar on some factors at baseline is a definite improvement, but leaves open the question that they differ on other (unmeasured) factors. Allocation of individual patients to treatments using random numbers overcomes two problems: clinician bias in assigning patients to groups, and differences in unmeasured factors.

The sequence of methods presented in this chapter could be taken to imply that there was a steady progression to increased robustness of study design. However, as the dates for the individual studies show, there is little evidence for continuous improvement in methods: rather there was substantial overlap in the use of these methods. The major advances in trial methodology occurred in studies conducted in the middle of the twentieth century. They used three techniques that are now hallmarks of high quality trials: randomisation, blinding of the investigators and patients to the randomisation process, and objective outcome measures.

      In summary, this chapter has reviewed the development of methods to evaluate treatments up to the middle of the twentieth century. It has highlighted pitfalls of many of the earlier methods and concluded with an outline of the advantages of the double blind randomised controlled trial. This method is now used around the world to identify the benefits of treatments. Medicine now has the tools to ensure that only effective treatments are used. The next chapter explores whether the benefits of the randomised controlled trial have been realised.

REFERENCES