Database. Two were industry oriented: Pharmaceutical News Index and International Pharmaceutical Abstracts. Two others were primarily agricultural and included international materials: Agricola and Agris International. Seven were either business or general news databases: ABI/INFORM, Trade and Industry Index, Magazine Index, Newspaper and Periodical Abstracts, National Newspaper Index, PTS Prompt, and Newssearch. Four miscellaneous databases included Dissertation Abstracts, the U.S. Government Printing Office database, the National Technical Information Service (NTIS) file, and the Library of Congress' LC-Marc database. Monographs were searched in Books in Print Online, the Online Computer Library Center (OCLC) database, and NLM's Catline file.
Current literature was found in Current Contents Online and through Conference Papers Index. Three databases covered legal and regulatory affairs: Diogenes, Legal Resources Index, and Legi-Slate. In addition, the histories of the reviewed vaccines were searched by using the History of Medicine (Histline) file on the NLM's system and the Legis-Slate database.
Each database was searched in its entirety. Years of coverage vary among databases, beginning no earlier than the mid-1960s, however. Both English- and non-English-language articles were included. When possible, the literature searches were sorted into studies of humans, studies of animals, and review articles.
Much of the retrieved literature came from MEDLINE and EMBASE. The NLM's MEDLINE databases and the Excerpta Medica EMBASE database files have approximately a 40 percent overlap in the journals indexed. MEDLINE's unique content is primarily North American journals published in English, whereas EMBASE's unique content is primarily non-U.S. and non-English-language journals.
Each database has unique indexing structures and thesauri. For example, some databases index Guillain-Barré syndrome as such while others use the term polyradiculoneuritis. Even such overarching terms as adverse events or adverse effects are defined as subject headings in some of the databases but not in others. This inconsistency in classification is very common and presents difficulties. To help ensure that relevant articles would not be missed, search strategies were intentionally broad. In addition to lists of synonyms for the particular adverse events being reviewed for each vaccine, more general terms such as risk, danger, and contraindication were used. A unique search strategy was thus created for each vaccine in each database. Further, the terms were searched for in the title, abstract, and descriptor fields of each database. Limiting a search to the descriptor field might have resulted in missing relevant articles if an indexer happened not to list a particular adverse event.
In addition to the initial searches of vaccine and adverse event combinations (Table B-1), the committee requested searches of the following at