How to do it. Organise a multicentre trial.

This is the authors' abstract. We don't add key points for this paper.

Randomised or non-randomised treatment comparison * Blind or open treatment allocation and outcome assessment * Intention to treat or on-treatment analysis, or both * Sample size and duration of follow up * Subgroup analysis * Defined entry and exclusion criteria * Amount of data to collect on randomised and non-randomised but eligible patients * Defined outcomes * Widespread applicability of trial results; what to do about non-randomised but eligible patients * Interim analysis and the role of an independent data monitoring committee * How intensively to look for adverse effects of treatment * How intensively to monitor compliance * Computerisation of data * Sharing of original data with overview groups looking at similar treatments (meta-analysis) * Involvement with sponsoring company, if any.