Before asking whether a result is impressive, ask what question the study was designed to answer.
Start with the people and comparison
Who took part, and who was excluded? Was the comparison with placebo, another treatment or an earlier period in the same person's life? The answers determine which conclusions are reasonable.
Check the time frame
Distinguish measurements collected during follow-up from outcomes predicted by a model. A longer projected time frame does not create additional observed data.
Read the denominator
Percentages need a reference point. Losing 10% of starting body weight and regaining 10% of previously lost weight describe different amounts. A difference between two percentages is usually expressed in percentage points.
Look for uncertainty and missing outcomes
Confidence intervals, missing data, withdrawals and side effects all matter. A headline benefit is only part of the result.
Check funding and publication status
Read declared funding and interests. Distinguish a complete peer-reviewed paper from an abstract, preprint or press release. None is a substitute for examining the methods.
Ask what remains unanswered
A study may be useful without settling a personal treatment choice. Avoid turning a result from one population, dose or formulation into a claim about every available option.
Read the source
Worked example · Budini et al., EClinicalMedicine
The linked material provides source context. Practical checklists and explanations are this publication’s editorial guidance.
This publication provides general information, not personalised medical advice. Discuss treatment decisions with an appropriately qualified clinician.