ABSTRACT
Randomised controlled trials (RCTs) aim to compare the outcomes of participants in one ormore intervention groups with those in a control groupwho received ‘care as usual’, which might be nothing (Cochrane 1989). Randomisation is used as the method of allocating participants to intervention and control groups because, if enough participants are involved, the different groups closely resemble one another so that any differences in outcomes are unlikely merely to represent pre-existing differences between groups (Oakley 1990). TheRCT is often regarded as the ‘gold standard’ of rigorous evaluation of clinical
(Cochrane 1989) as well as public health interventions (Oakley et al. 1996). However, some criticise the validity of trials, suggesting they are epistemologically ‘positivist’ and thus invalid for investigating the social processes inherent inmost public health interventions (Tones 1997). Others criticise their lack of feasibility, arguing that randomising participants andmaintaining a strict separation between intervention and control groups are usually impossible (Kippax and Van den Ven 1998). This paper does not discuss these matters further and assumes that trials of public health interventions are often feasible and generate evidence regarded as valid by many if not all researchers. Instead, it explores utility: whether trials in practice provide useful evidence. Utility is defined as the ability to inform policy and practice (Walt 1994). To
provide useful evidence, trials should generate evidence about interventions which, according to existing knowledge, are good candidates for implementation but for which we do not have definitive evidence about effectiveness (Oakley et al. 1996). Clinical interventions are subject to trials only after earlier evaluation of biological effects in laboratories and clinical effects in case studies. Proponents of trials of public health interventions also recommend that evaluation is phased (Oakley et al. 1996), withRCTs being done only where there is prior evidence of acceptability and appropriateness inmeeting population needs. Such evidencemight be derived from formative research, employing mostly qualitative data to examine intervention provision and receipt (Ramos et al. 1995), or from evidence of prior successful deployment to meet similar needs elsewhere.
