How to Read a Research Paper Critically

You open a new paper and the title catches your eye. Before you decide whether the conclusions change anything for the patients in front of you, you need a way to test the work itself. A critical read is not about finding flaws for their own sake; it is about knowing how much weight any single study can safely carry.

Clarify the question the authors actually asked

Read the introduction and the final sentence of the abstract until you can state the research question in one plain sentence. Many papers drift between what they set out to measure and what they later claim to have proved. If the question remains vague after two readings, the rest of the paper will be difficult to judge. Note whether the authors chose a question that can actually be answered with the design they used.

Look closely at who was studied

Examine the methods section for the exact inclusion and exclusion criteria and for how patients were recruited. Ask whether the people in the study resemble the patients you see in age, severity, comorbidities and setting. A study that excludes everyone over seventy or anyone with previous surgery may give clean numbers yet leave you uncertain about the patients who actually arrive in clinic. Pay attention to how many eligible patients were never enrolled and why.

Trace the methods for bias and error

Work through each major step of data collection and analysis as if you were doing it yourself. Check whether randomisation was truly concealed, whether outcome assessors knew the treatment allocation, and whether losses to follow-up were handled fairly. Look for any post-hoc changes to the primary outcome or any subgroup analyses presented without a prior plan. Small decisions about measurement or grouping can shift results more than most readers expect.

Compare the results with the claims

Return to the discussion and highlight every sentence that moves beyond the data. Distinguish what the study directly measured from what the authors infer about mechanism, cost or long-term benefit. See whether the confidence intervals are wide enough to include clinically important differences in both directions. If the abstract and the discussion emphasise a finding that appears only in a secondary analysis, treat that emphasis with caution.

Notice what the authors have not said

Every study has boundaries. Ask what the paper does not address: longer-term outcomes, rare but serious harms, performance in different healthcare systems, or the experience of patients who declined to participate. Good authors state these limits plainly. When the gaps are left unmentioned, you must supply them yourself before you decide how far the results can travel.

Decide what, if anything, you will carry forward

After you have worked through the previous steps, ask a single practical question: would this finding change how you talk with the next patient who matches the study population? If the answer is yes, decide what extra information you still need before you act on it. If the answer is no, the paper has still done useful work by sharpening the boundary around what you already do.

A paper earns your trust when it survives this kind of patient reading, not when its conclusions happen to match what you hoped to find.