How to summarize a medical research paper (without missing the limitations)

Most bad paper summaries fail the same way: they capture the conclusion and lose the caveats. The abstract says the intervention worked; page seven says the trial was underpowered, the follow-up was short, and the confidence interval nearly touches one. A summary that only carries the headline is worse than no summary at all, because it gets cited.

A good summary is not a shorter version of the paper. It is the same set of fields, extracted the same way, every time — so that forty papers later you can line them up and actually compare them.

The fields that matter

• Study type — RCT, cohort, meta-analysis, preprint, case report. Everything else is read through this lens.
• Population, intervention, comparator, primary endpoint — the PICO frame, taken verbatim where the paper allows it.
• Primary result — effect size, confidence interval, and p-value, exactly as reported. Not paraphrased.
• Key findings with page numbers — so every claim can be checked against the source in seconds.
• Limitations — the ones the authors state, not the ones you infer.
• Funding, conflicts of interest, registration ID — the provenance fields that decide how much weight the paper can carry.

Notice what is not on the list: a verdict. Whether the evidence is strong or weak is the reviewer's call, made across papers — not a property of any single summary.

Doing it by hand vs. letting a tool do the transcription

Extracting these fields by hand is good practice and slow — fifteen to thirty minutes per paper if you are careful, multiplied by every paper in your screening pile. Symmathy's Scan a PDF feature does the transcription: you drop in the PDF, and it returns the fields above as a structured record, with the page number attached to every key finding.

Two things about how it works matter for trust. First, the PDF never leaves your device — the text is extracted in your browser, and only that text is analysed, after a patient-information scrub. Second, the page references are the point: a summary you cannot verify is just an opinion with formatting. Every finding links back to the page it came from, so checking a number takes seconds, not a re-read.

Papers that aren't in English

A large share of the world's medical literature is published in Japanese, Portuguese, Thai, Korean, Spanish, Chinese, French, and Arabic. The scan reads the paper natively in its source language and writes the summary in the language you choose, keeping the original title verbatim so it stays citable. That is how a regional trial ends up in your evidence table instead of being silently dropped at the language filter.

From summary to review

1. Scan each paper and read the structured fields, not just the summary line.
2. Spot-check the page-cited findings against the PDF — especially the primary result.
3. Send two or more papers to Compare when their endpoints line up.
4. Download the print-ready PDF of any scan for journal clubs or supervisor meetings.
5. Keep notes on each record so your reasoning survives until the write-up.

Symmathy reports what each paper states and where it states it. It does not grade, rank, or recommend studies — appraisal stays with the reviewer, where it belongs.

Questions people ask