Head of Health Services Research Unit, NHS Quality Improvement Scotland
Correspondence: Jim.Slattery{at}emea.eu.int
(Received 10 July 2004)
The ARES study (Kiritzé-Topor et al., 2004) addresses an important issue of the use of acamprosate in general practice patients with alcohol dependence. The paper gives a strongly argued advocacy for pragmatic trials in alcohol dependence. The authors have tried to ensure that the study closely mimics clinical practice in almost all possible ways, thus increasing the chance that results will generalize to the non-research setting. Care and effort were expended on ensuring rational treatment of outcome data from subjects who discontinued. Few would argue with the merit of these measures but some other features of the design may provoke more debate.
The outcome measure used was supplied by the ARPQ. This amalgamates several alcohol-related effects into a single score. The authors state that this gives a measure of the impact of treatment on patients' lives. However, this statement appears to be based on a perceived face validity rather than direct questioning of the patients concerning their aim in seeking treatment. Moreover, heavy drinking is associated with long-term ill effects which can be irreversible by the time they are apparent to the drinker. Hence direct measurement of consumption, despite being in some respects a surrogate outcome, may have equal importance to immediate patient perceptions of their problem. However, alcohol consumption data were also presented as secondary outcome measures, and corroborated the ARPQ findings.
This was a study without placebo control. The purpose of a placebo, in addition to maintaining physician blinding, is to ensure that any non-pharmacological effects of a drugthe psychological encouragement of just knowing that you are receiving a treatmentis equal between the groups allocated to different treatments. The disadvantage of not having a placebo is that it is not possible to prove that any treatment effect is not partially, or even entirely, due to this placebo effect.
Does it matter even if the effect is not pharmacological? Certainly the patients are better off in that they have been helped to achieve their goal and there is some chance that this effect, pharmacological or not, will be generalizable to non-trial situations. However, little is known about variations in the placebo effect with setting (in particular whether a research setting may be more conducive to a placebo effect than clinical practice) and the potential for becoming refractory through habituation. Economically the ground is still more shaky. The same placebo effect may be obtained with either cheap or expensive medicines.
Thus we are left with some act of faith in interpreting such a trial. Whilst we may not be certain that any treatment can be attributed to the pharmacology of the drug, it is still reasonable to ask how likely we feel it is that a well established and quite unobtrusive treatment such as acamprosate would induce a measurable placebo effect. On the other hand, the weight of placebo controlled explanatory studies in specialist settings support a pharmacological effect, albeit with some unexplained heterogeneity between studies.
The ARES study investigators have clearly thought long and hard about the design of the study and produced an elegant trial with many good features. However, the principles of pragmatic trials can never be fully achieved. For instance, the ethical requirement for informed consent precludes exact replication of the usual clinical setting. Thus the choice must always be made between pragmatism and conflicting but desirable characteristics. In this case, one may be left wondering whether the uncertainty about the role of placebo effects was not too big a price to pay for operational simplicity and adherence to an attractive but debatable principle.
![]() |
REFERENCE |
---|
![]() ![]() |
---|
|