Reading a Scientific Paper

#Identify the question and design

Start by identifying the question: who was studied, what exposure or intervention was examined, what comparison was made and which outcomes mattered? The title and abstract provide an overview, but they may omit important qualifications. Treat them as an entry point rather than the whole account.

Next, identify the design. Was this a randomised trial, an observational study, a laboratory experiment or an evidence review? Ask whether that design can support the conclusion being proposed. A study describing an association, for example, needs additional reasoning and evidence before it can support a causal claim.

#Examine how the work was done

Read the methods for participant selection, measurement procedures, follow-up and handling of missing information. Check what the comparison group received and whether outcome assessment could have been influenced by knowledge of treatment. Consider whether the studied population and setting match the question you want answered.

Where available, compare the report with its registration or protocol. Look for the planned primary outcome and any changes to analyses. Changes can be justified, but should be explained. Funding, competing interests and data availability add context; none is a substitute for assessing the methods themselves.

#Separate findings from interpretation

In the results, look for the size of the effect, absolute numbers and confidence intervals, not just statements about statistical significance. Check the denominators, time period, harms and missing outcomes. Tables may reveal qualifications that are easy to miss in a brief summary or discussion.

#Finish with limitations and context

The discussion contains the authors' interpretation, which may go beyond what the data directly show. Compare each main claim with the reported findings. Ask whether chance, bias, confounding or an indirect outcome could explain the result, and whether important alternatives were acknowledged.

Finally, place the paper alongside other relevant research. Peer review does not certify that a study is correct, and one paper rarely settles a broad question. A useful reading can end with a clear account of what was learned and what remains uncertain.

#Common misunderstandings

A published paper is not a final answer. Peer review can improve a study, but it does not guarantee that its methods are sound or its conclusions correct. An abstract is a useful starting point, not a substitute for checking the full results and methods.

“Statistically significant” does not necessarily mean important for health. A small difference may meet a statistical threshold without making a meaningful difference to symptoms, daily life or survival. Equally, a result that is not statistically significant does not prove there is no effect; the study may be too small or its estimates too uncertain.

An association does not, by itself, establish cause and effect. Other differences between groups may explain the finding. Results in animals, laboratory settings or narrowly selected participants may not apply directly to routine care. Finally, one striking study should be weighed against the wider evidence, rather than treated as overturning it automatically.

#Questions worth asking a clinician

• Does this study’s design answer its research question, and do the authors’ conclusions go beyond what the results can support?
• Do the methods and full results support the abstract’s claims, or reveal important limitations it leaves out?
• How large is the reported effect in practical terms, and what does the uncertainty range mean for interpreting it?
• Were harms, missing data and participant withdrawals reported clearly, and could omissions change the study’s conclusions?
• How do these findings compare with systematic reviews and other relevant studies, and what might explain any differences?