The most robust evidence for the use of psychiatric drugs is generally agreed to come from randomised controlled trials that compare a particular drug or intervention with a standard or ‘control’ condition, such as a placebo. Randomisation is important because it allows the effects of the intervention being tested to be distinguished from the effects of other things, such as the natural history of the condition and general factors that might produce improvement like seeing a specialist. To further reduce the risk of bias, the investigators and participants may be ‘blinded’, or made unaware of who receives the drug and who the control treatment or placebo.
Combinations of the results of several different trials of the same treatment, called meta-analyses, are also regarded as providing high quality evidence. However, a meta-analysis is only as good or as poor as the trials it combines. A meta-analysis of poorly conducted trials summates their deficiencies or biases and so the result may be more misleading than the original studies.
Randomised controlled trials were developed to test the outcomes of interventions for physical medical conditions. Translating them into the area of mental health is not straightforward and there are various difficulties with interpreting the results.
4.1.1 The validity of measurements
Emotional states and behaviours are properties of living human beings and cannot be described and quantified in the same way that we measure the properties of physical objects. Therefore, the meaning and validity of measurements of mental symptoms is not clear-cut.
4.1.2 Ignoring drug-induced alterations
Since most research is premised on the disease-centred model of drug action, the general alterations that drugs produce on physical and mental functioning are often ignored and interpreted as changes in the underlying ‘disorder’. Yet these alterations may change people’s experience and behaviour without affecting the underlying problem.
4.1.3 ‘Publication bias’
Studies that find positive effects of drugs are more likely to be published than studies that find they have no benefits or cause harm.1 In addition, published reports of studies often emphasise the measures that show the drug in the best light.1 Measures that show no benefit or that indicate harmful effects may not be published or may be concealed in the small print of the article.
Some pharmaceutical companies have been shown to withhold data that do not show their drug in a favourable light.2 But doctors, researchers and editors have also played a part in focusing on research that highlights the positive and plays down the negative effects of drugs. There are extensive financial relationships between these groups, that have been shown to bias the undertaking, interpretation and reporting of research.
The use of a placebo is meant to prevent participants and researchers from knowing whether they are getting the real drug or not. This is why studies using a placebo are referred to as ‘double blind’. However, it is often quite easy for people in trials to tell whether they are taking the drug or the placebo, due to the mental and physical alterations drugs produce independently of any effect they might have on the underlying disorder. The chance that people will detect whether they are taking a drug or a placebo is heightened because people who take part in trials are given detailed information about the ‘side’ effects of the drug being tested.
What this suggests is that many trials that are supposed to be double blind are not. Many of the participants and some of the professionals involved are likely to be able to work out who is taking the real drug and who is on the placebo. Trials in which people are asked to guess what they are taking show that in most cases people can detect the nature of the pill they have been given.3 If people taking part in trials believe that drugs are likely to help them, they may have a heightened expectation of improvement if they suspect they are taking the real drug. Conversely, they may have lowered expectations if they believe they are on the placebo. Any differences in the outcome of treatment may be due to these different expectations, rather than the effects of the drug.
4.1.5 Drug withdrawal effects in trials
Most trials of long-term treatment, and many trials of short-term treatment too, involve people who are already taking the drug that is being tested, or something similar. The people who are randomised to placebo are then taken off their existing treatment and may therefore be vulnerable to adverse effects related to the withdrawal of the prior treatment.4 This is especially problematic because the withdrawal and transfer to placebo is usually done abruptly. Therefore, many studies, particularly those assessing long-term treatment, may assess the effects of withdrawing from prescribed drugs rather than the impact of starting on it in the first place.
This array of potential problems suggests that care must be taken when interpreting research on psychiatric drugs, and the clinical guidelines based upon them.
1. Melander, H., Ahlqvist-Rastad, J., Meijer, G. & Beermann, B. (2003). Evidence b(i)ased medicine: Selective reporting from studies sponsored by pharmaceutical industry: Review of studies in new drug applications. BMJ 326(7400), 1171–3.
2. Jureidini, J.N., McHenry, L.B. & Mansfield, P.R. (2008). Clinical trials and drug promotion: Selective reporting of study 329. International Journal of Risk and Safety in Medicine 20(1–2), 73–81.
3. Fisher, S. & Greenberg, R.P. (1993). How sound is the double-blind design for evaluating psychotropic drugs? The Journal of Nervous and Mental Disease 181(6), 345–50.
4. Moncrieff, J. (2006). Why is it so difficult to stop psychiatric drug treatment? It may be nothing to do with the original problem. Medical Hypotheses 67(3), 517–23.