Patients receiving placebo do not merely receive an inert substance but also receive support, concern, and reassurance that assists the therapeutic alliance and encourages the positive attitude that forms the basis of cognitive treatment. Response to non-specific factors is seen in all fields of medicine but is particularly potent in psychiatry. The placebo response is variable across settings and across time and is unpredictable. Historical data cannot therefore provide an adequate control for treatment effects in studies of new drugs. The scientific position is clear that a comparison against placebo is required for the unequivocal demonstration of the efficacy of a treatment. Evidence from at least two positive well designed and conducted placebo-controlled studies is generally accepted as appropriate to establish the efficacy of a drug. Attention to diagnosis, severity of illness, and to possible comorbid conditions is needed in the design and conduct of placebo-controlled studies in order to optimise the chance of obtaining valid data. The use of all data obtained, including dropouts due to lack of efficacy, should be maximised. The use of placebo may not be possible in some conditions that represent medical emergencies or may be difficult to justify in serious disorders where an effective treatment has already been established. Alternative designs to placebo-controlled studies can be considered. Consistent superior efficacy compared with a well accepted, effective treatment, given in an easily defended dose, is considered to be good evidence of efficacy provided that the studies are well designed and well conducted. Evidence of superior efficacy to an established effective comparator treatment may be regarded as evidence of efficacy. The demonstration of a dose-response relationship where one dose is found to be significantly better than another, can be taken as evidence for efficacy, particularly where there is already placebo-controlled evidence of the efficacy of the identified dose. Where a new treatment is found, under controlled conditions, to be equivalent to an existing well accepted comparator treatment, given at a clearly effective dose, this may be taken as evidence of efficacy, but only if the comparator is consistently superior to placebo and if equivalence has been defined beforehand. The claims for efficacy based on results from equivalence studies are less easily sustained than the evidence from placebo-controlled studies or studies demonstrating superior efficacy, due to the fact that those studies have no internal validation. (C) 1999 Elsevier Science BN./ECNP. All rights reserved.
