Table 3.

 Optimising the evaluation: example of community based screening for genital infection

Key task	Example
Identify feasible and valid measures of outcomes—both intermediate and final; specify the outcome on which the study will be powered	Key outcomes are tubal infertility and ectopic pregnancy.25 Research evaluating effects on these would take many years or a huge sample. A crucial intermediate outcome is whether individuals at risk of these adverse outcomes will accept screening
Consider randomisation; randomise at the level of the intervention; but cluster randomise only if necessary	A feasibility study used individual randomisation, partly because cluster randomisation required a much larger sample size (about 21 times26). There were, however, practical difficulties with this. Data protection legislation prevented researchers from accessing names and addresses of potential participants, so staff in each general practice had to generate the randomised mailings. This proved prohibitively expensive
Estimate recruitment and retention rates	The feasibility study showed that recruitment rates in the current ethical and legal framework might make a definitive trial difficult27
Calculate sample sizes	A small scale feasibility study in 3 general practices informed sample size calculation for a definitive trial using both cluster and individual randomisation based on rates of acceptance of screening