Recruitment into Randomised Trials

Abstract

One of the most important aspects of undertaking an RCT is obtaining sufficient numbers of participants to allow us to demonstrate or exclude an important effect. Health care trials seem particularly vulnerable to under-recruitment, particularly trials that use ‘trickle’ or sequential recruitment, where we cannot identify all those in our sample of participants at the same time. For instance, if we want to recruit to a surgical trial for appendicitis we need to recruit participants as they arrive for surgery for the condition. Similarly, if we are recruiting to a study testing the effect of a court order on children absent from school, we need to recruit as the parents are referred to court. In contrast, other trials, such as a study of a school-based intervention means we can recruit the schools and then the children, achieve our sample and then randomise all the participants at the same time. Recruitment to this sort of trial tends to be easier and quicker as we can identify a list of schools, mail out to the schools and then the parents of the children in the schools, all of which can be undertaken relatively quickly and easily. In contrast, with sequential recruitment we have to estimate the numbers of potential participants arriving in a clinic, estimate how many are likely to be eligible and assume that this will remain constant. All of these estimates are subject to error, and often the numbers are over-estimated, leading to shortfalls in total recruitment and a lower sample size than is ideal.