Prospective meta-analyses are an attractive option to clinical trialists who, although appreciating the benefits of single, adequately sized trials, are unable to undertake them (Simes 1987, Probstfield 1998). It can be a useful methodology, for example, when large sample sizes are required to ensure adequate power, but single, large-scale trials are not feasible. This could be due to local interests preventing participation in a trial when information is perceived to be ‘lost overseas’. This can also be a particular problem in rare diseases where gaining access to large numbers of trial participants in a timely manner may be difficult.
Hence, an alternative is for investigators to conduct their own study locally, and to collaborate with the investigators of similar studies, arranging for the results to be combined at the completion of each trial. This enables individual investigators to maintain a certain amount of autonomy, and at the same time to plan appropriately for the meta-analysis. Another situation where it may be beneficial, particularly in the absence of mandatory prospective registration of randomized trials, is when two or more trials addressing the same question commence and the investigators are ignorant of the existence of the other trial(s). Once similar trials are identified, investigators can collaborate (adapting data collection if necessary) and plan prospectively to combine their results in a meta-analysis.
What also distinguishes a PMA from a multicentre trial is that there is no requirement in a PMA for the protocols to be identical across studies. Variety in the design of the studies may be viewed by some as a desirable feature of PMA, and thus a degree of expected variation in populations or in aspects of the interventions is considered acceptable. FICSIT (Frailty and Injuries: Cooperative Studies of Intervention Techniques) is an example of a pre-planned meta-analysis of eight studies of exercise-based interventions in a frail elderly population (Schechtman 2001). The eight FICSIT sites defined their own interventions using site-specific endpoints and evaluations and differing entry criteria (except that all participants were elderly). This deliberate introduction of systematic variability in design, known as a ‘meta-experimental design’, is a possible approach to PMA (Cholesterol Treatment Trialists' (CTT) Collaborators 2005).