MEDLINE definitions of race and ethnicity and their application to genetic research

To the editor

Over the last five years, the use of MEDLINE has increased more than ten-fold, attesting to the importance of the database in the scientific community (see http://www.nlm.nih.gov/bsd/medline _growth_508.html). MEDLINE must organize thousands of topics into an indexing system that strikes a balance between chronological consistency and current interests. It is an invaluable resource to scientific research. But race- and ethnic group–related terms in the Medical Subject Headings (MeSH) used to index articles in MEDLINE are decades old and inconsistent. In light of recent debates concerning how best to describe human genetic variation, this valued resource requires attention so that it can continue to support high quality research^1,2,3.

A search for 'race' in MeSH returns the phrase 'racial stocks,' which MeSH defines as “major living subspecies of man differentiated by genetic and physical characteristics” (see http://www.nlm.nih.gov/mesh/MBrowser.html). MeSH lists four races: Caucasoid, Mongoloid, Negroid and Australoid. It defines each of the first three as a “major racial group” distinguished “according to physical features.” The definition also lists each stock's geographic regions and some of its populations, which for Negroid include “Andamanese, Afr [sic] Bushmen, Half-Hamites, Hottentots, Melanesians, Negrillos, Negritos, Papuans, Pygmies, Semangs” (see http://www.nlm.nih.gov/mesh/MBrowser.html). Several of these outdated terms appear in the same 1950 publication that provides the original source of this definition for Australoid (ref. 4 as cited in ref. 5). This definition omits reference to genetic and physical characteristics and lists only the names of some Australoid populations, such as the Veddahs of Ceylon. The Ainu of Japan are assigned to both Mongoloid and Caucasoid racial stocks (see http://www.nlm.nih.gov/mesh/MBrowser.html).

MeSH defines ethnic group as “a group of people with a common cultural heritage that sets them apart from others in a variety of social relationships.” MeSH lists 13 such groups, drawn primarily from United States populations. Ethnic group definitions range from one phrase to several sentences. The characteristics listed include, in order of frequency cited, geographic location (nine), racial classification (six), ancestry (three) and history (two). Three characteristics are used only once: religion to define Jews; social organization (“itinerant life and tribal organization”) to define gypsies; and language group (Semitic) to define Arabs. 'Aborigine' is defined as a kind of population, “indigenous inhabitants,” rather than a specific population “with a common cultural heritage,” contradicting MeSH's own ethnic group definition (see http://www.nlm.nih.gov/mesh/MBrowser.html).

MeSH distinguishes between racial stocks and ethnic group. The former is designated for indexing articles that concern a population's “physical characteristics, genetic characteristics, anthropometric measurements, physiology,” whereas the latter designates articles about “cultural, psychological, social, sociological or ethnological aspects.” MeSH instructs indexers to override an author's choice of race and ethnicity terms if it conflicts with MeSH's definitions (see http://www.nlm.nih.gov/mesh/indman/ chapter_30.html). Searches conducted in October 2002 suggest, however, that indexers apply this distinction inconsistently and that in some categories, as many as 30% of articles conforming to racial stock criteria are instead indexed to ethnic group.

Additional searches, also conducted in October 2002, showed little need for several outdated and sometimes offensive⁶ terms used in racial and ethnic group definitions. For example, 'Hamites' is listed as a Caucasoid population, but as a keyword search, the term returns only one article⁷. 'Hamitic-Semitic' subjects are referred to in two articles^8,9. From the Negroid racial stock definition, 'Hottentots' returns a handful of articles, mostly historical. 'Negrillos' and 'Half-Hamites' each return the message that MEDLINE is “[u]nable to map your term to a subject heading.”

MeSH should conform with standard language and abandon nineteenth century colonialist terms, such as Hottentots, Hamites and Half-Hamites. Furthermore, in addition to a general review and standardization of racial and ethnic group terms, MeSH needs to reconsider its assertion in the definition of racial stocks that there are sub-species among humans. For historical accuracy, MEDLINE might devise a way to identify articles that ascribe to the sub-species argument. But the implication that this long-repudiated notion is current and legitimate is unacceptable^10,11.

See "Reply to MEDLINE definitions of race and ethnicity and their application to genetic research" by Stuart J. Nelson