Trials tend to be focused on effectiveness or efficacy, as they are conducted in the real world of an emergency department. Efficacy refers to how well something works under ideal circumstances, while effectiveness refers to how well something works in real life.

Trials are either pragmatic or explanatory. An explanatory trial is one that seeks clues as to how an intervention may work. Pragmatic trials are more concerned whether an intervention works. Surrogate outcomes are measures that are easily measured and predict poor outcome, but are not an actual indicator of clinical harm or benefit.

Blinding is a key feature of any randomised controlled trial. The more blinding there is, the less biased results are likely to be. It is not always possible to ‘blind’ the physician and the patient. Lack of blinding can potentially cause observer bias.

Try to decide whether the control patients received the same care as you would give. If there are losses to follow-up, bias can result. An intention to treat analysis means that all data on patients allocated to the intervention and control, including all the drop outs, crossovers and protocol violators should be analysed along with all those who completed the protocol. Failure to use an intention to treat analysis usually results in a more favourable outcome for the intervention.