1. Who, what, compared with what?
Pin the trial down in four parts, often called PICO: the Population studied, the Intervention, the Comparison it was tested against, and the Outcome measured. A treatment compared with a placebo answers a different question from one compared with the current standard of care.
2. Was allocation truly random, and hidden?
Randomisation is what lets a trial attribute a difference to the treatment rather than to who received it. Look for how the random sequence was generated and whether the people enrolling patients could not predict the next assignment (allocation concealment). Then check that the groups looked alike at the start.
3. Who knew who got what?
Blinding the participants, the clinicians and the people assessing the outcome protects against expectation shaping the result. Some treatments cannot be blinded; when they cannot, an outcome that is objective, or assessed by someone who does not know the group, matters more.
4. Is the main outcome the one that was planned?
Trials should name their primary outcome before they start, usually in a public registry. Compare the registered outcome with the one reported. Be cautious when the headline rests on a secondary outcome, a subgroup, or a surrogate measure (a lab value, say) standing in for how patients actually do.
5. How big, and how certain?
Look past whether the result was “significant” to its size and its confidence interval. A small effect with a wide interval says less than it seems to. Where the paper reports it, the absolute difference between the groups tells you more about practical impact than the relative one.
6. Were people lost, and does it apply to yours?
Check how many participants dropped out or were excluded, and whether they were analysed in the group they were assigned to (intention to treat). Then ask whether the people in the trial resemble the people you have in mind, in age, severity and setting.
The CONSORT statement lists what a trial report should contain; if a paper leaves these questions unanswered, that is itself an answer. This guide is about reading trials, not a substitute for clinical judgement or guidance.
