Structured Abstracts in MEDLINE: Retrospective Cohort Study and Recent Update
Brown Bag Lecture by Anna Ripple | 5/1/2012 11AM-12PM | 7th Floor Conference Room, Bldg 38A
Abstract: BACKGROUND: A structured abstract is an abstract with distinct, labeled sections (e.g., RESULTS). Since August 2010, they have been reformatted for easier readability in the MEDLINE®/PubMed® database with bolded uppercase section labels followed by a colon, with each section beginning on a new line.
OBJECTIVE: To measure and characterize the growth in structured abstracts in articles indexed in MEDLINE since 1991 and to assess their potential utility in enhancing information retrieval.
METHODS: Identified structured abstracts in MEDLINE records indexed from 1991 to 2006 and analyzed this subset to 1) measure growth in the number of records with structured abstracts and in the number of journals contributing structured abstracts; 2) compare characteristics of MEDLINE records with and without structured abstracts. All unique structured abstract labels were manually grouped using lexical matching techniques and examined to determine if they could be categorized in ways that might aid retrieval.
RESULTS: The percentage of MEDLINE records with structured abstracts rose from 2.5% in 1992 to 25% in 2010. Over 1000 journals continuously publish structured abstracts. Records with structured abstracts are larger than the average MEDLINE record length and have more access points (text words and Medical Subject Headings (MeSH®) terms), following the pattern identified in a 1989-1991 study. Almost all identified section labels were mapped to five broader category labels: BACKGROUND, OBJECTIVE, METHODS, RESULTS, CONCLUSIONS.
CONCLUSIONS: The substantial growth in the individual number of MEDLINE records with structured abstracts and in the number of journals that continuously publish structured abstracts demonstrate widespread adoption of structured abstracts. NLM is exploring their utility in facilitating indexing and enhancing information retrieval, discovery, and display.
Bio: Anna Ripple received a MLS from the University of Missouri-Columbia in 1990. She came to NLM in 1991 as an NLM Associate Fellow. After joining the Lister Hill Center in 1992, she has worked numerous projects, including Coach/UMLS (DOS-based) Metathesaurus Browser, Internet Grateful Med, ClinicalTrials.gov, Semantic MEDLINE, and Genetics Home Reference.