What have we learnt so far?

Research should be carried out prospectively, starting with a hypothesis, clearly defining the elements of the research question and recruiting a cohort of patients and following them forwards in time.

The treatment must be compared against a control group who are treated the same apart from the fact that they don't get the experimental treatment, but get a placebo and standard care instead.

Bias in research designs can distort the results, usually leading to an exaggeration of treatment benefits.

Participants should be randomly allocated to experimental or control groups. The purpose of randomisation is to produce two (or more) groups of patients which are identical at the start of the trial. This ensures that all confounding variables, even the ones we don't know about, will not bias the results.