A definition tells you what an abstract is supposed to do. It doesn’t show you what that looks like in a real, published abstract, sentence by sentence. This guide does the second thing: it takes one real structured abstract from a well-known, peer-reviewed study and walks through what each labeled section is doing and why, so the pattern is visible in practice rather than only in the abstract description of it.
Why structure is the point
An abstract is not a teaser or a table of contents — it is a standalone summary that a reader, or an indexing database, uses to decide whether the full paper is relevant without reading it. That function is why journals following ICMJE’s Recommendations require a structured abstract for original research: labeled sections force the author to state the study’s context, what was done, what was found, and what it means, in a fixed order a reader can scan quickly. ICMJE leaves the exact subheadings and word count to the individual journal — commonly some variant of Background, Methods, Results, and Conclusions — which is also the ordering pattern behind the broader manuscript-level IMRaD structure (Introduction, Methods, Results, and Discussion) that the abstract mirrors in miniature.
Worked example: dissecting a real published abstract
The example below is Turner et al., “Selective Publication of Antidepressant Trials and Its Influence on Apparent Efficacy,” New England Journal of Medicine 2008;358(3):252-260 (PubMed). It’s a useful teaching example for two reasons: it uses a clean four-part structured format, and its own subject — how selective publication distorts what the literature shows — makes the discipline of stating results plainly, without spin, especially visible in its own abstract.
Background: one problem, stated in two sentences
The abstract opens by naming the general problem (evidence-based medicine depends on a complete, unbiased evidence base) and the specific mechanism that threatens it (selective publication of trials, and of specific outcomes within trials, can distort the apparent effectiveness of a treatment). Notice what the Background section is not doing: it is not reviewing the literature at length, and it is not previewing the results. Its only job is to make the reader understand, in the shortest space possible, what gap in knowledge the study addresses and why it matters.
Methods: what was compared, and against what
The Methods section states the design in operational terms: the authors obtained FDA reviews of antidepressant trials — FDA review is a useful comparison point precisely because it is largely blind to a trial’s outcome, unlike a decision to submit or publish a manuscript — and matched those trials against the corresponding published literature to see how the same body of research looked from each vantage point. A reader does not need every methodological detail here (sample-selection criteria, statistical tests) to understand what kind of comparison is being made; that level of detail belongs in the paper’s own Methods section, not its abstract.
Results: numbers, not adjectives
This is where the abstract earns its keep as a standalone summary, because it states the finding as a quantified comparison rather than a qualitative claim like “publication bias was substantial.” Of 74 FDA-registered antidepressant trials, the FDA itself classified 38 as positive and 36 as negative or questionable. Of the 38 positive trials, 37 were published. Of the 36 negative-or-questionable trials, only 3 were published as negative; 22 were never published at all, and 11 were published in a way that conveyed a positive result. The net effect: a reader of the published literature alone would conclude that roughly 94% of trials were positive, when the FDA’s own regulatory data showed the true figure was closer to 51%. Pooling only the published trials also inflated the apparent effect size by about 32% overall (ranging from 11% to 69% depending on the individual drug). None of this requires reading the full paper to understand — which is exactly the standard a Results section is supposed to meet.
Conclusions: what follows, stated with appropriate hedging
The Conclusions section states the practical consequence — that selective reporting can produce misleadingly favorable impressions of a treatment’s efficacy for researchers, participants, clinicians, and patients — without overclaiming causation the data can’t support. In this case the authors are explicit that their analysis cannot determine why studies went unpublished or were reported selectively (author decision, journal editorial decision, or some mix), and the abstract reflects that limitation rather than glossing over it. A well-written Conclusions section states what the results support, not what the authors merely hope is true.
What to notice, transferable to any abstract
- Background is short and general-to-specific. One or two sentences establishing why the topic matters, then the specific gap or question — not a literature review.
- Methods states the comparison, not the procedure. A reader needs to know what kind of study this is and what was compared against what; full procedural detail belongs in the body.
- Results lead with numbers. Effect sizes, proportions, and counts read as evidence; adjectives like “significant” or “substantial” without a number attached do not.
- Conclusions match the evidence’s actual strength. State what the data support, and say plainly when the data can’t settle a related question (as this abstract does about the cause of non-publication).
- No citations, no forward references. The abstract never points to a numbered reference or says “as discussed below” — it has to be intelligible read completely on its own.
Common mistakes when drafting your own abstract
- Front-loading background at the expense of results. A common failure mode is spending half the word budget motivating the study and leaving barely a sentence for what was actually found — the opposite of what a reader scanning search results needs.
- Reporting findings without numbers. “The intervention significantly improved outcomes” tells a reader almost nothing measurable; state the effect size, the sample, or the proportion.
- Overclaiming in the conclusion. Stating a causal or generalizable claim the study design doesn’t support is one of the more common reasons an abstract misrepresents its own paper.
- Ignoring the target journal’s actual format. Some journals, including PLOS ONE, require an unstructured single-paragraph abstract rather than labeled subheadings — see CASRAI’s Abstract entry for the structured-vs-unstructured distinction and typical length ranges. The four-part logic walked through above (context, comparison, quantified finding, appropriately hedged conclusion) still applies inside a single paragraph; only the labeled headings disappear.
Related CASRAI resources
For the operational definition, structured-vs-unstructured distinction, and typical word-count ranges, see the Abstract dictionary entry. A graphical abstract covers the same standalone-summary function in a single publisher-displayed image rather than prose, and is a separate, usually optional, submission requirement at journals such as Elsevier and Cell Press. A conference abstract is a related but distinct genre, submitted for presentation selection rather than accompanying a peer-reviewed journal article. For the section-by-section logic of the full manuscript this abstract summarizes, see IMRaD structure, and for CASRAI’s broader coverage of manuscript preparation, see the Scholarly Writing & Publishing hub.
Frequently asked questions
Can I use a real published abstract as a template for my own?
Use one for structure and technique — the sequence of moves (context, comparison, quantified finding, hedged conclusion) — not for wording. Always follow your target journal’s own instructions for authors for the required format, subheadings, and word limit, since these vary by journal even within the same general structured-abstract convention.
Does a structured abstract need to use exactly Background, Methods, Results, Conclusions?
No. ICMJE explicitly leaves the exact subheadings to the individual journal. Clinical-research journals sometimes expand the format with headings such as Design, Setting, Participants, or Main Outcome Measures. Confirm the exact labels required in the target journal’s submission guidelines before drafting.
How long should each section of a structured abstract be?
There’s no universal rule, but in practice a structured abstract’s total length (commonly 150-350 words, journal-dependent) is split so Background gets the least space — often one to two sentences — and Results typically gets the most, since it carries the information a reader most needs to judge relevance.
