Efficacy and effectiveness relate to one of the most important questions in medicine – does a particular intervention work or not? This may also be one of the most interesting questions in medical research, yet it provides an incomplete picture of the usefulness of a given treatment in daily practice. Of equal importance in the evaluation of therapeutic interventions are the issues of the safety and cost of treatment modalities. These aspects are intimately interconnected. Their proper assessment is aimed at ensuring that a treatment does more good than harm. They are viewed as the most fundamental preconditions that ought to be met before considering the integration of any therapy into routine care, regardless of whether this involves complementary or mainstream medicine. The terms efficacy and effectiveness are frequently used in the medical literature. Seemingly similar in meaning, they express distinctly different concepts. It may be timely to elaborate on these terms. Patients in clinical trials are often highly selected; in contrast, patients encountered in real-life situations are not. For example, participants in clinical trials usually have no health problems other than the condition being investigated, whereas in routine clinical practice patients frequently have multiple illnesses. In addition, patients not participating in a clinical trial may fail to follow medical advice, which could have the result that success rates for treatments are different from those reported in clinical trials. In general, one can define two broad questions. Does the treatment work under ideal conditions? Does the treatment work in everyday life? The terms efficacy and effectiveness refer to these two questions. Whether or not an intervention can work under ideal conditions relates to efficacy.1 If the conditions of a trial are optimised, the measurement of the outcome variable may detect even relatively small effects of the treatment. In such a trial, bias is excluded as far as possible, for example by including only patients who are likely to cooperate fully with the medical advice.2 In such a trial, patients should not be treated with concomitant medication and other co-interventions should be avoided. A treatment is efficacious when it proves to be superior to (usually) placebo or another treatment of known efficacy. On the other hand, the pragmatic question of whether an intervention works in routine clinical care relates to effectiveness.1 In this setting, the inclusion criteria for patients are more relaxed. In such a trial, the question is whether the treatment does more good than harm among those to whom it is offered and patients are allowed to accept or reject the treatment, much as would be the case in real-life situations.2 Patients in such trials might be treated with concomitant medication and other co-interventions may also be carried out. If a treatment is found to be ineffective, this may be the result of a lack of efficacy or other factors. The majority of clinical trials addresses the issue of efficacy of a particular intervention. Similarly, systematic reviews or meta-analyses evaluating such trials usually assess the efficacy of a given treatment. These studies should adhere to similar principles as outlined above. From the results of systematic reviews or meta-analyses assessing efficacy it is, however, possible to hypothesise on the potential effectiveness of a treatment. In general, trials testing the efficacy of a particular intervention are of an explanatory nature, the aim being to establish a causal relationship. In trials investigating effectiveness, a more pragmatic approach is taken with the aim of assessing an intervention in routine clinical practice. The distinction between these terms is not purely academic but relates to important concepts in medical care and thus is of clinical relevance. Consider, as a hypothetical example, the evaluation of a mono-preparation of a plant extract for the treatment of the common cold. To determine the efficacy of this extract, i.e. whether it works under ideal conditions, a randomised, double-blind, placebo-controlled trial is carried out. No co-intervention is allowed and patients taking concomitant medication are excluded. The trial reveals beneficial effects of the extract when compared with placebo. However, in such an efficacy study, only patients who comply with the treatment would be analysed. Suppose that the bitter taste of the plant extract results in a high drop-out rate. When the extract is later evaluated in a large-scale randomised trial with the aim of determining its effectiveness in routine clinical care, it is possible that, owing to the high drop-out rate, no beneficial effects would be detected. This (somewhat simplistic) example of a treatment with demonstrated efficacy but no effectiveness shows that the results of trials may vary according according to the question asked. It also demonstrates that it may be wise to establish the efficacy of a treatment before attempting to assess its effectiveness. If one or (better) more efficacy trials fails to detect any beneficial effects, it is obviously unwise to embark on an expensive trial investigating effectiveness. On the other hand, if a treatment is shown to be effective without having been proved to be efficacious, it is impossible to ascertain whether the beneficial effects were caused by the intervention or by other factors. It would be easy, almost tempting, to draw the wrong conclusions. Within the field of complementary medicine, the available data for a large number of therapies are relatively scarce: efficacy studies are needed. Practitioners frequently argue that the constraints of efficacy trials render them unsuitable for the assessment of complementary therapies. For example, treatments are individualised and have multiple components with synergistic effects, the therapeutic relationship is an important component of the therapy, and randomisation creates an artificial environment that affects both the patients and the therapists adversely. We believe that it is both imperative and practicable to design efficacy studies that take into account these special requirements. If a given complementary therapy proves to be efficacious, the above-mentioned factors may render it all the more effective in a real-life situation. It is normally more logical and more economical first to test whether complementary therapies can work under ideal circumstances and subsequently to embark on studies of their effectiveness in routine care. One of the most important reasons for this strategy may be the limited resources available for research into complementary medicine. In conclusion, efficacy and effectiveness are two different concepts that are often confused. A clear distinction and a reasonable research strategy are issues of considerable scientific importance and practical relevance. References 1. Spraycar M (Ed). Stedman’s Medical Dictionary. 26th edn. Baltimore: Williams & Wilkins, 1995. 2. Fletcher RH, Fletcher SW, Wagner EH. Clinical epidemiology – the essentials. Baltimore: Williams & Wilkins, 1982. 3. Max H Pittler are Associate Editors and Research Fellows at the University of Exeter, UK. 4. Adrian R White are Associate Editors and Research Fellows at the University of Exeter, UK.