How We Write Study Explainers and Research Updates
The fixed template we use to summarize a study, what we always report, how publication dates differ from our update dates, and why we avoid the word "latest".
Content last updated . This page has not been medically reviewed.
Why a fixed template
A news headline can make one small study sound like a settled fact. To keep that from happening here, every study explainer and research update on this site answers the same questions in the same order. If a question cannot be answered from the paper or record we read, the page says so instead of filling the gap. This page describes the method. For the labels we use, see How We Label Evidence in Specialty Topics.The template
| Part | What it must say |
|---|---|
| The claim | The exact question the study tested, in plain words |
| The model | Cells, animals or people. Findings are never carried from one to the next |
| The intervention | The specific product, device, procedure or exposure, not just the field it belongs to |
| Design and comparator | Randomized or observational, blinded or open, and what the comparison group received |
| Size and length | Number of participants and how long they were followed |
| What was measured | A clinical outcome, a biomarker, or a brain or lab signal, kept apart |
| Results | The effect size with its confidence interval, and absolute numbers as well as relative ones |
| Harms | Adverse events reported, including none found |
| Limits and funding | What the study could not show, who paid, and any stated conflicts |
| Corrections check | Whether the paper has an erratum, expression of concern or retraction |
| Status | Established, investigational or commercially speculative, with the regulator and product named if approval is claimed |
Rules behind the template
- Cells and animals stay in their lane. Early laboratory work is not a substitute for studies in people. [1] Our pages label this evidence preclinical.
- A biomarker change is not a clinical result. A surrogate endpoint is a stand-in for how a person feels, functions or survives, and it only counts as reliable after extensive evidence that it predicts those outcomes. [2]
- Absolute and relative numbers together. A 25% relative drop and a 1 percentage point drop can describe the same change, from 4% to 3%. [3]
- Uncertainty is part of the result. The confidence interval shows how uncertain an estimate is. [3]
- Count trials, not papers. Several papers from one trial are one body of evidence, not several confirmations.
- Null and mixed results go in. Findings that fail to hold up are common enough that one result is never treated as final. In one set of highly cited clinical studies, 16% were later contradicted and another 16% had found stronger effects than later work. [4] In a large replication project in psychology, replication effects were half the size of the originals. [5]
- Check the record. We look for correction and retraction notices before using a paper. See Corrections and Retractions in Health Research.
Two different dates
Every study has a publication date. Every page here has a separate editorial date for when we last changed the text. They are not the same thing, and we never let one stand in for the other.Databases keep several dates too. In PubMed, the date of publication is separate from the date a record was created and from the entry date used for processing. [6] In ClinicalTrials.gov, the first-submitted date is when the sponsor or investigator first sent the record in, which can be well before any results exist. [7] When we write a date, we say which one it is.
Why we avoid "latest"
"Latest" is a promise that nobody has checked anything newer. We cannot keep that promise by hand for every topic. So we do not label anything latest, new or breakthrough. Instead:- We write "as of" with the date we read the source.
- We name the search we ran, so a reader can repeat it. See How to Find Emerging Research and Trial Records.
- When something important changes, such as a retraction or a regulatory decision, we update the page and its editorial date rather than adding a news item.
- We do not claim live counts of studies. Any number we give comes from a dated record.
Frequently asked questions
Does "study explainer" mean you reviewed the study for accuracy?
No. We summarize what the paper or record says and what it cannot show. We do not claim a medical review, and we do not re-analyze data.
Why do you give both the publication date and your own update date?
They answer different questions. The publication date tells you how old the evidence is. The editorial date tells you when we last checked the page. Databases also keep separate dates for creation, entry and publication. [6]
Why do you report biomarker results separately from outcomes?
Because a lab or imaging change does not always mean a person does better. A surrogate endpoint is accepted only after extensive evidence that it predicts a real clinical benefit. [2]
Why is there no "newest research" list?
Without a process that is repeated and dated, a list like that goes stale quietly. We prefer pages that say what was read and when.
What do you do if a study we used is later retracted?
We remove or rewrite the claim it supported and change the editorial date. A retraction is a signal to re-check the evidence, not proof that a whole field is wrong.
More details: evidence, limits and cautions
The details behind this page: cautions, limits of the evidence and where sources disagree. Open any note to read it.
A safety note
References
- FDA, Step 3: Clinical Research. Read 2026-10-22.
- FDA, Surrogate Endpoint Resources for Drug and Biologic Development. Read 2026-10-22.
- Cochrane Handbook for Systematic Reviews of Interventions, Chapter 15: Interpreting results and drawing conclusions. Read 2026-10-22.
- Ioannidis JPA, Contradicted and initially stronger effects in highly cited clinical research (JAMA 2005). Read 2026-10-22.
- Open Science Collaboration, Estimating the reproducibility of psychological science (Science 2015). Read 2026-10-22.
- National Library of Medicine, PubMed User Guide. Read 2026-10-22.
- ClinicalTrials.gov, study data field definitions (API metadata). Read 2026-10-22.