01
What the evidence says
Start before the abstract conclusion. Identify the research question, preregistration or protocol, study design, setting, recruitment, eligibility, intervention or exposure, comparator, primary outcome, duration, and funding. Then ask what would have happened to the same participants without the exposure. Randomization, blinding, measurement, confounding, and selection affect how closely the comparison answers that counterfactual.
Sources for this section: [1]
02
How to use the finding
Read the participant flow and baseline table, then count outcomes and missing data. Look for absolute numbers, uncertainty intervals, adverse events, protocol deviations, and whether the primary outcome changed. Separate prespecified analyses from exploratory subgroups. Statistical significance does not establish clinical importance, absence of a small p-value does not prove equality, and a surrogate outcome is not automatically a patient benefit.
Sources for this section: [1]
03
Limits and open questions
Finish by writing one bounded sentence: in this population, over this time, this comparison changed this outcome by this amount, with these major uncertainties. Compare the authors’ language with that sentence, the protocol, related studies, and systematic reviews. Conflicts do not automatically invalidate work, but they affect which design choices, data access, and claims deserve closer inspection.
Sources for this section: [1]
Sources
- How to read a paperBMJ · 2019