A plain language summary is an account of a study written for readers outside the field: patients, carers, policymakers and journalists. It says why the study was done, what the team did, what they found, what it means and what it cannot show, in everyday words. Check the journal's or funder's rules first, because no single standard exists.
A plain language summary (PLS) is not a shorter abstract. The abstract is written for specialists, and our guide to writing a strong abstract covers that version. A PLS answers a lay reader's questions: why should I care, what happened, and how sure are they? Some journals, funders and regulators ask for one; others do not.
Find out whose rules apply before you draft
No single standard governs plain language summaries. A 2022 systematic review of plain language summary guidance covered 90 records, 17 of them guidelines. It states that there is no consensus on what a PLS should contain, and it found no empirical evidence behind most of the criteria in those guidelines. It also notes that guidance is often tied to the purposes of the body that issues it, so name the guidance you follow.
| Source | Who it applies to | What it sets |
|---|---|---|
| Cochrane plain language summary guidance (Version 1, January 2022) | Cochrane Reviews only | 400–850 words including the title; set headings, most phrased as questions; no statistical data; no treatment recommendations in the key messages |
| EU Clinical Trials Regulation 536/2014, Article 37(4) and Annex V | Sponsors of medicinal-product trials, since 31 January 2022 | A lay summary sent with the trial results to the EU database; Annex V lists 10 elements. Not a journal requirement |
| EU Good Lay Summary Practice guidance (2021) | Lay summaries under that Regulation | Guidance, not law; it places journal plain language summaries outside its scope |
| Your target journal or funder | Its own articles or grants | Whether a summary is wanted, its name, length and headings |
Where nobody sets a rule, use the five-part structure below. It echoes Cochrane's question headings, but it is our suggestion, not a standard.
Decide who will read it and what they already know
Pick one main reader first. A patient knows the condition's symptoms and everyday names, but not your methods. A policymaker wants the size of the effect and who it applies to. Write for the least specialist reader you expect: assume intelligence, not knowledge.
Structure a plain language summary in five parts
Build the summary around five questions a lay reader brings. Why was the study done? What did the team do? What did they find? What does it mean? What can it not tell us? Use them as headings if your guidance allows.
The fifth question stops a claim the study cannot carry. Match every verb to the design: "was linked with" for observational work, "reduced" only for a randomised comparison. The guide to matching a conclusion to its study design gives the wording for each design.
Copy the template and fill every slot. Delete any line your guidance does not ask for.
Title: [a question or a plain statement of the finding, with no abbreviations]
Why did we do this study? [the problem, in a patient's words]. [what was not known].
What did we do? We [studied / asked / followed] [number] [who took part] in [where]. [what was compared or measured, and for how long].
What did we find? [the main result in words]. [counts, such as x in every 100 in each group, if your guidance allows numbers].
What does this mean? [what the finding adds, and for whom].
What can this study not tell us? [what was not measured, or who was not included]. [the study that would answer it].
Who paid for the study? [funder, or "no specific funding"].
Where can I read more? [link to the full article].
Choose words a reader outside your field already knows
Replace a technical term with everyday words where you can. Where the reader needs the term, such as the name of their condition, keep it and define it once so they can search for it. Drop abbreviations. Short sentences, the active voice and no filler matter even more here; writing clear scientific English covers those techniques.
| Term in the paper | Plain wording |
|---|---|
| randomised | put into groups by chance, as if by tossing a coin |
| placebo | a dummy treatment that looks the same but has no active ingredient |
| cohort study | a study that followed a group of people over time |
| prevalence | the share of people in a group who have the condition at a given time |
| incidence | how many new cases appear in a group over a set period |
| adverse event | an unwanted medical event during the study, whether or not the treatment caused it |
| statistically significant | leave it out and say how big the difference was; in everyday English "significant" means "important" |
Present numbers so a lay reader can picture them
Guidance conflicts on numbers. The 2022 review reports that some guidelines, among them American Psychological Association guidance, advise leaving numbers out, while Cochrane guidance from 2013 to 2019 stressed specific figures. Cochrane's current template says its summaries report no statistical data. Follow your journal's or funder's rule; where there is none, use this approach.
- Give absolute numbers first. "The risk halved" fits a fall from 2 to 1 in 100 and a fall from 40 to 20 in 100: 1 fewer case per 100 people, or 20 fewer. The Results section guide asks specialists for both the absolute and the relative difference; for a lay reader, put the absolute one first.
- Use one denominator. Write "15 in every 100" each time, not a mix of percentages, "1 in 7" and fractions.
- Say who the numbers apply to: "among the 400 adults in this study".
- Put uncertainty into words, such as "the study was too small to be sure", instead of an interval.
- Do not turn an unclear result into a negative one. "The study could not tell whether the treatment helps" is accurate where "the treatment did not work" may not be; a non-significant result is not evidence of no effect.
Leave out what a lay reader does not need
- Names of statistical tests, p-values, odds ratios and confidence intervals, unless your guidance asks for them.
- The literature review; one sentence on what was not known is enough.
- Secondary outcomes and subgroups, unless the reader's question depends on one.
- Advice on whether to use a treatment. That decision belongs to readers and their clinicians, and Cochrane's template bars treatment recommendations from its key messages.
- Promotional words such as "breakthrough" or "cure".
The same restraint applies when you share the paper, as promoting your published paper honestly explains.
Check the reading level, then test the draft on a real reader
Readability formulas such as Flesch Reading Ease score text from sentence length and syllables per word. They flag long sentences and heavy words; if your guidance sets a reading-level target, use the formula it names. But a score cannot tell whether a short word is unfamiliar, whether ideas follow in order, or whether a number misleads.
So test the draft on a person outside your field, such as a patient partner or a colleague from another discipline. Ask them to retell what the study found and what it cannot show, and which word made them stop. Where their account differs from your finding, rewrite that part.
A plain language summary rewritten, before and after
The study below is invented to illustrate the structure; its numbers are not real data.
Before (written for specialists): In this pragmatic two-arm randomised controlled trial (RCT; n = 400), short message service (SMS) reminders significantly increased uptake of diabetic retinopathy (DR) screening compared with usual care (75.0% vs 60.0%; risk ratio 1.25, 95% confidence interval 1.09 to 1.44). SMS reminders are effective and should be implemented across DR screening programmes.
After (written for a lay reader):
Do text-message reminders help people with diabetes go to eye screening?
Why did we do this study? Diabetes can damage the back of the eye. Screening can find this damage early, but some people who are invited do not go. We did not know whether a text reminder would help.
What did we do? We asked 400 adults with diabetes to take part. A computer split them by chance into two groups of 200. One group got a text reminder as well as the usual letter; the other got the letter only.
What did we find? With a text, 150 of 200 people went to screening: 75 in every 100. With the letter only, 120 of 200 went: 60 in every 100. The reminder meant 15 more people in every 100 attended.
What does this mean? In this study, a text reminder helped more people attend eye screening.
What can this study not tell us? We counted who attended, not whether anyone's sight was protected. Everyone had a mobile phone and the study ran in one region, so the result may differ elsewhere.
Who paid for the study? A public research grant.
Where can I read more? Full article: doi.org/10.xxxx/example.0001.
What changed, and why:
- Every abbreviation went, and "diabetic retinopathy" became "damage to the back of the eye".
- The risk ratio and interval became counts with one denominator and the absolute difference.
- "Significantly" was cut; the size of the difference says more.
- The call to roll out reminders went. One trial in one region cannot carry it, and it measured attendance, not sight.
- If your guidance allows no numbers, keep the sentences and drop the counts.
If you find an error in this guide, report the problem to us and we will look at it.