Home / Evidence & Research Literacy
Reading a Scientific Paper
Reading a scientific paper critically means connecting its question and methods to its actual results, limitations and wider evidence.
#State the question before reading the conclusion
Identify the people studied, the intervention or exposure, the comparison and the outcomes. Then identify how the study assembled that comparison. A randomised trial and an observational study require different questions about assignment, selection and confounding.
Use the abstract as an overview, then read the methods and results. STROBE and CONSORT both call for enough detail to show what was planned, done and found. A headline or a broad design label cannot supply that account.
Evidence: STROBE: observational reporting explanation and elaboration / CONSORT 2025 explanation and elaboration / Cochrane Handbook, Chapter 25: bias in non-randomized intervention studies
#Trace the methods and the missing information
Read eligibility, setting, measurement procedures, follow-up and who contributed to the analysis. For a trial, check the comparator and how randomisation and blinding were handled. For observational treatment research, check how important confounders were chosen and measured.
Where available, compare the report with the protocol and analysis plan. Look for the planned outcomes and explanations of changes. CONSORT also asks for information about registration, funding, conflicts of interest and access to data and code. These disclosures add context; they do not replace reading the methods.
Evidence: Cochrane Handbook, Chapter 8: risk of bias in a randomized trial / Cochrane Handbook, Chapter 25: bias in non-randomized intervention studies / CONSORT 2025 explanation and elaboration
#Read the estimate and its uncertainty
Look for the actual numbers in each group, the effect estimate and its uncertainty, rather than only whether a result was called statistically significant. STROBE recommends reporting estimates with confidence intervals rather than only a significance label. Read the time period and outcome definition as well.
Check harms, missing outcomes and who was included in each analysis. STROBE asks authors to distinguish unadjusted from adjusted estimates and explain which confounders were included. A precise estimate can still be biased, and an adjusted estimate still depends on the assumptions of its analysis.
Evidence: CONSORT 2025 explanation and elaboration / STROBE: observational reporting explanation and elaboration / Cochrane Handbook, Chapter 25: bias in non-randomized intervention studies
#Keep the conclusion within the evidence
Compare the authors' interpretation with the comparison and outcomes actually studied. Ask what bias, confounding, missing information or limited applicability could change the conclusion. For intervention studies, Cochrane distinguishes risk of bias from whether the result transfers to people outside the study.
Use reporting guidance to find these details, not as a score that proves the paper is good. STROBE explicitly says it is not a quality-assessment instrument. A useful reading ends with a bounded account of what was found, what remains uncertain and what other evidence is needed. It is not a treatment decision from one paper.
Evidence: STROBE: aims and limits of reporting guidance / Cochrane Handbook, Chapter 25: bias in non-randomized intervention studies / CONSORT 2025 explanation and elaboration
Source note
The sections above were checked against the linked sources. No clinical review has been performed. This is general research education, not a clinical guideline.