3.5 Evaluating intervention studies
3.5.1 Expectancy versus effect
One of the biggest problems in evaluating psychological interventions is that even if a treatment appears to ‘work’ it can still be difficult to ascertain whether the results were a consequence of the treatment itself. The improvement might have occurred anyway, with or without the treatment, or the apparent benefits might have resulted from other factors, such as being able to discuss the difficulties with a professional who understands. Any treatment can lead to expectations of improvement that can be self-fulfilling. Even a treatment with absolutely no benefits for the condition in question can be followed by substantial improvements if the person believes that it will help. This is the so-called placebo effect.
For biochemical interventions, professionals evaluate treatments via randomised, double-blind, placebo-controlled trials (see Box 17). This kind of trial remains the benchmark of ‘evidence-based medicine’, as the most objective way to find out whether a treatment is effective. It can also be used to evaluate other types of intervention strategy where it is possible to develop a suitable placebo.
Box 17: Randomised, double-blind, placebo-controlled trials – some definitions
Placebo-controlled – the treatment being tested (known as the ‘active' treatment) is compared with another treatment (the ‘placebo’) that is believed to be neutral with regard to the underlying specific basis of the condition in question, but is otherwise indistinguishable from the active treatment. Studies with no placebo are known as open treatment studies.
Double-blind – to eliminate as much bias as possible, it is crucial that neither the participants nor those carrying out the study know which people are receiving the ‘active’ treatment and which the placebo (i.e. everyone involved in the study must be ‘blind’ to treatment status until all data have been collected and analysed). Sometimes it is not possible to achieve this owing to the nature of the treatment, in which case the next best option is the single-blind study, in which the participant does not know what kind of treatment they are receiving but those carrying out the trial do.
Randomised – the kind of treatment that each participant receives must be determined by pure chance, otherwise there is an opportunity for bias (conscious or unconscious) in the allocation of treatments that could influence the results.
Proof of efficacy is required before any new drugs are licensed for prescription, but inevitably this evidence is often obtained under rather limited conditions. Individual differences in the metabolism of many drugs can mean that not everyone reacts biochemically in the same way to a given dosage. For ethical or other reasons, clinical research trials often do not include particular groups such as children or women. Many drug-based interventions also have undesirable side-effects that have to be weighed against the possible benefits. Since psychological problems usually have a wide range of possible interacting causes, in many cases drugs may not be the primary treatment option.
There are difficulties inherent in carrying out placebo-controlled trials, particularly for non-pharmacological treatments, including:
deception – can it be justified?
double-blind trial – can this be achieved when the experimenter is administering a cognitive or behavioural ‘placebo’ therapy?
homogeneity – for some medical problems it may be possible to ensure close similarity of symptoms experienced across all participants. However, for many complex psychological conditions this is much more difficult.