The RCT is one of the simplest, most powerful and revolutionary tools of research (Jaded 1998). The process of assignment in experimental research design is vitally important. In an ideal world, representative, randomly selected samples of patients would be assigned in equal numbers to both the experimental and control groups. This process of allocation is of vital importance, as any partiality may confound the study and devalue the usefulness of the findings. Routinely, in an RCT a proposed new treatment option is evaluated against the best standard treatment currently available. Patients have a 50:50 chance of receiving either the standard treatment or the experimental arm.

To avoid the potential of selection bias it is important that the process of random allocation should be concealed from the individual recruiting for a study. Selection bias relates to potential biases that may be introduced into a study by the selection of different types of people into treatment and comparison groups. As a result, the outcome differences may potentially be explained as a result of pre-existing differences between the groups, as opposed to the treatment itself. Nelson et al (2006) suggested blinded or masked allocation methods to minimise selection bias. These methods include the generation of random number sequences, remote telephone randomisation and randomisation via sealed envelopes. Failure to adequately mask allocation increases the possibility of inflated estimated effectiveness of experimental interventions. Proper randomisation hinges on adequate allocation concealment (Schulz & Grimes 2002).

The randomised control trial is rooted in ‘positivist’ science. It is concerned with events that can be observed, requires a stable environment, should be quantifiable and aims to establish causal relationships (Proctor 1998). The RCT is recognised as the principal method for obtaining a reliable assessment of treatment effects (Richardson 2000).

Cooke et al (2005) employed an RCT design to examine the effect of music on preoperative anxiety in day surgery. One hundred and eighty subjects were randomly allocated to an intervention, placebo or control group in a study which supported the use of music as an independent nursing intervention for preoperative anxiety in surgical patients. Chan et al (2005) designed an RCT to evaluate the effectiveness of an osteoporosis education programme for women in Hong Kong. Pre-, post- and follow-up education outcome measures compared attitudes and consumption frequency before and after the education programme. Using an RCT they concluded that an educational programme can act as simple but effective nursing intervention to promote women’s attitudinal and behavioural intentions towards osteoporosis prevention.

Randomised controlled trials are designed to use large numbers of subjects to test the effect of a treatment or an intervention and to compare the results with a control group who have not received the treatment or intervention. Randomised controlled trials have been used in medical science since the 1940s and they now form the basis for strength of evidence in clinical guidelines (National Institute for Health and Clinical Excellence 2007) often viewed as a ‘gold standard’ approach. These types of clinical trials (e.g. drug trials) are usually conducted in ideal circumstances and subjects are selected according to a narrow set of inclusion/exclusion criteria. They are a popular type of experiment for testing the effectiveness and cost efficiency of treatments and interventions in health care.

Randomised controlled trials are designed specifically to minimise selection bias. However, the researcher should also be aware of other potential forms of bias which may lead to systematic errors in an experiment. These include performance bias and attrition bias.

Performance bias relates to the experimental intervention, or exposure to other factors apart from the intervention of interest. For example, in a study examining the impact of massage therapy in the management of postoperative pain, patients who received psychological support from the nurse researcher in addition to their massage therapy may fare better than others in the study. This improvement may be related to the effect of the additional support rather than the massage therapy and, as such, would introduce bias into the study. Attrition bias relates to withdrawals or dropouts of participants from a study. The way the researcher deals with withdrawals or dropouts from a study has the potential to bias the results of an RCT, as the withdrawal may be related to the intervention or outcome.

Larson et al (2005) used a longitudinal, randomised controlled trial to examine the impact of a nurse-led support and education programme for 100 spouses of stroke victims. Participants were randomly assigned to either an intervention or control group for a 12-month period. The intervention group met six times for support during six months, which had a positive effect on their well-being.

In the absence of an intervention, there is no experiment. The researcher must introduce an intervention to produce an outcome or effect. In nursing research, the term ‘treatment’ may be used instead of ‘intervention’. In conducting an experimental study, the nurse researcher attempts to make things happen. Parahoo (1997) stated that many nurse researchers do not state their hypotheses or questions in studies. This, he argued, does not facilitate the task of the reader, who has to piece together the information in the paper, including the results, to ascertain the researcher’s hypotheses, questions or objectives.

Chang et al (2002) used an RCT to investigate the effects of massage on pain reaction and anxiety during labour. The experimental group received massage intervention whereas the control group did not. The nurse-rated present behavioural intensity as a measure of labour pain and anxiety was measured with the visual analogue scale for anxiety. The intensity of pain and anxiety between the two groups was compared in the latent phase (cervix dilated 3–4 cm), active phase (5–7 cm) and transitional phase (8–10 cm). In both groups, there was a relatively steady increase in pain intensity and anxiety level as labour progressed. Chang et al demonstrated that the experimental group had significantly lower pain reactions in the latent, active and transitional phases. Anxiety levels were only significantly different between the two groups in the latent phase. Twenty-six of the 30 (87%) experimental group subjects reported that massage was helpful, providing pain relief and psychological support during labour.

The researcher should devise strategies to control extraneous variables to ensure that the intervention (treatment) is the only variable responsible for outcome in the study. Extraneous variables may influence the outcome of an experiment. For example, a nurse researcher may wish to use a randomised controlled trial to measure the impact of stress management techniques in patients with a chronic illness who display high levels of anxiety (Smith et al 2002). To confirm that stress management therapy did have a therapeutic impact, the researcher could compare the patients with a group of patients who did not receive relaxation therapy. The group of patients who received relaxation therapy is called the experimental/intervention group and the group to which comparison is made is called the control group. Patients should have a 50 : 50 chance of receiving either the standard treatment or the intervention.