Evidence Synthesis

Preserve Structured Abstract Sections for Screening

Decide which text reviewers should see

Open the article record and identify each abstract, its language, its type when stated, and its section headings. A scientific abstract, a lay summary, and a translated summary can coexist. Preserve these identities when preparing a screening export. Concatenating every available block into one unlabeled paragraph can repeat the study description and blur which result came from which source.

Within a structured abstract, retain the sequence of headings and paragraphs. Check the output for embedded symbols, lists, and paragraph boundaries. An abstract that loses a minus sign or joins the last word of one section to the first word of another can materially change the reading of a result.

Pilot a hypothetical export

Ten articles each have one scientific abstract. Four also include a lay summary and two include a translated scientific abstract. The export contains 16 abstract blocks across ten records. If screening requires one scientific abstract per article in its original language, ten blocks populate the primary view and six remain available as supplementary views. The number of blocks is unrelated to the number of independent studies.

Sample the screening view against the publisher page and record the handling of missing or unlabelled abstracts. Retain a link to the full record when the abstract is insufficient for a decision. An exporter's failure to find a particular element should produce a visible missing-text state.

JATS abstract element reference

Related worksheet: Collection pilot worksheet.

Source checked on 5 November 2026.

Related reading

Reading path: Prepare data. Guide 8 of 50. Choose any guide to suit your project.

Evidence Synthesis

Preserve Identifier Namespaces in Article XML

Build article keys from identifier type and value when more than one identifier occurs in a record.

Research Workflow

Preserve Leading Zeros in Research Identifiers

Import identifiers as text and detect collisions before a spreadsheet or parser converts them to numbers.

Browse this reading pathTools for this stageGuides and downloads for this stage