Can you critically appraise a paper from its abstract?

An abstract can support an initial reading of the question, design and reported finding. It usually cannot support a complete appraisal of the methods. Treat an abstract-only review as a map of reported information and open questions. The absence of a detail in an abstract is not evidence that the authors did not perform the method.

Start your review notes with the material you actually have: abstract, main article, supplement, registration or protocol. Consider a fictional abstract that says participants were randomized but gives no allocation details. You can report that randomization is claimed. You cannot invent the sequence-generation or concealment procedure. Mark those details as unavailable in your current source.

A PubMed record may link to the publisher and to a free version in PubMed Central. Follow the available links and check that the title, authors and version match your citation. Library access may provide additional material. A model connection does not grant access to a subscription article, and an appraisal should not imply it read material that was never retrieved.

MedLit labels the retrieved source coverage and retains retrieval warnings. Its evidence checks distinguish matched quotations from unsupported model text. Matching a quotation establishes that the words occur in the source; it does not prove the interpretation. Review the context, missing information and uncertainty before using the output in a journal club or research note.

End with a short list of questions that require the full paper. That list can guide a library search or a discussion with your supervisor. If the full text becomes available later, revisit the appraisal rather than treating an earlier abstract-only conclusion as final. Keep both the source and your own interpretation visible in the review trail.

Explore the teaching example without an account, or bring a paper and connect your own AI provider.