Consistency and accuracy of the Medical Subject Headings thesaurus for electronic indexing and retrieval of chronobiologic references.

The aim of this study was to investigate the reliability of the National Library of Medicine (NLM)'s Medical Subject Headings (MeSH) thesaurus for electronic indexing and retrieval of published chronobiologic papers. A sample set of 228 recent chronobiologic references was downloaded from the MEDLINE's database together with all MeSH entries associated with them. The following descriptors were analyzed among the headings of obvious chronobiologic relevance: chronobiology, chronobiology disorders, biological clocks, circadian rhythm, chronotherapy, drug administration schedule, periodicity, seasons, sleep disorders/circadian rhythm, and time factors. A comparison was made between the number of references identified by each heading and the number of articles actually pertinent to the same heading (as ascertained after reading each article of the sample set). This made possible an assessment of consistency (retrieved number not less than actual number) and accuracy (retrieved number not greater than actual number) of the usage of each MeSH entry. By reading each article, it was also possible to identify common chronobiologic concepts not yet associated with specific MeSH headings. In the preselected set of chronobiologic references, seasons identified all articles pertinent to seasonal variations and rhythms. However, chronobiology disorders missed 97.6% of its pertinent articles; periodicity, 95.2%; chronobiology, 87.7%; chronotherapy, 70%; time factors, 62.3%; and sleep disorders/circadian rhythm, 47.4%. Drug administration schedule missed 40% of the chronotherapeutic articles and identified 15% of the chronopharmacologic articles; biological clocks missed 24.1% of its pertinent articles and wrongly identified 8.3% of the retrieved articles; and circadian rhythm missed 2.7% of all circadian studies and wrongly identified 8.2% of the articles it retrieved. When used to search chronobiologic articles in the entire MEDLINE database, drug administration schedule, seasons, and time factors appeared to lack sufficient specificity to produce accurate results. Some common chronobiologic concepts were found not to be associated with any specific MeSH heading, namely, chronoepidemiology, chronopharmacology, chronotoxicology, chronotype, entrainment, and masking. For common chronobiologic concepts and definitions, the use of available MeSH headings appears to often yield inconsistent and inaccurate results; moreover, the MeSH thesaurus remains incomplete.