Caldwell and Popenoe's analysis is just as relevant to case reports for publication as to oral presentations. Efficient and effective scientific prose proceeds with "precision and brevity," a phrase coined in another field [3] but relevant to medical writing. Readers of case reports wish to have only what they need to know and no more. They need a presentation of the facts that were efficiently relevant to diagnostic analysis of the case or to therapeutic decisions. They do not need details that simply represent old habits in the ritualistic presenting of cases. A case report properly prepared for publication will not open with a "racial" or ethnic term unless a fact represented by that term was crucial to decisions in the case. Facts representing apparent influences on the patient's problems from genetic or cultural factors related to the patient's ethnic background will be stated as specific facts and not left to be possibly inferred from an ethnic term.

You may be asking, why quotation marks around race and racial and not around ethnicity and ethnic? The concept "race" was derived from taxonomic concepts long applied in botany and zoology. As one medical dictionary [4] puts it, a "race" is "a subspecies or other division of a species." But taxonomic assignments require precise and detailed descriptions that justify creating taxonomic categories. Three longstanding traditional categories of "race" [4] have been "Caucasoid (white), Negroid (black), and Mongoloid (yellow)." Unfortunately, as the same dictionary points out, "a limited range of visible characteristics (of "races") tends to oversimplify and distort the picture of human variation." The concept of "race" implied a degree of genetic homogeneity among persons designated as members of a "race." Modern genetics has trashed that view [5]. As one detailed discussion [6] concludes, "Although estimates of (the amount of genetic variation in human populations) vary, all agree that the amount of variation within any ethnic group is much greater than the difference between groups."

From these and related considerations, thoughtful judgments [5-8] have concluded that "race" has little or no utility in careful medical thinking. In contrast, ethnicity represents a concept that makes no claims to biological precision and that reflects a broader view of factors that may influence the susceptibility of individuals to etiologic agents and the ways in which they may respond to those agents. Senior and Bhopal [7] summarize the concept thus: "... (ethnicity) implies one or more of the following: shared origins or social background; shared culture and traditions that are distinctive, maintained between generations, and lead to a sense of identity and group; and a common language or religious tradition." But for reasons advanced by Caldwell and Popenoe, ethnic identifications in single-case reports are not likely to be any more useful for diagnostic analysis than "racial" identifications.

The question of the utility of ethnic identifications is more complex for reports of epidemiologic research [7-10] or clinical trials. Genetic and cultural characteristics of ethnic groups could be determinants of disease or its severity [7] in at least some members. The finding of a higher prevalence or incidence of a disease in one ethnic group than in another could be a valuable clue to an etiologic agent. The finding of a different response to, or a different frequency of an adverse effect from, a drug in one ethnic group compared with another [11] could be a valuable guide to decisions on dosage or a clue to mechanisms affecting a drug's mechanism of action or metabolism. The practical difficulties in ascertaining all potentially relevant genetic and cultural characteristics of individual subjects in epidemiologic research and clinical trials may force investigators to rely simply on ethnic-group designations of persons for assessing how those characteristics might account for findings. Hence, identifying the ethnicity of persons covered by an epidemiologic study or treated in a clinical trial may be a legitimate part of study designs for such research.

Designers of an epidemiologic study or clinical trial should know explicitly, however, why they include ethnic identifications in their research. Some antecedent findings may form a reasonable basis for identifying ethnicity to test a hypothesis about an ethnic influence on the variables to be measured. The other possible use of ethnic identifications is in heuristic efforts (also known as "fishing expeditions") to find factors that could account for variances in measured values. But with either approach, investigators should be clear in their minds and in published reports about how they justified identifying ethnicities.

The problem of defining an ethnicity is not an easy one, whether in medical research or for much wider applications. In the United States, the most conspicuous example is the classifications for the decennial national census. As preparations for the year 2000 census begin, pressures [12] on the Census Bureau for reclassifications and new categories of ethnic groups are mounting. These pressures may arise in part from emotional needs for ethnic identity, but even stronger pressures are those arising from the economic advantages for ethnic groups defined in various government programs. In the far smaller field of clinical trials, there are pressures to adequately represent ethnic groups that can legitimately wish to know whether trial results found with persons from one ethnic group are more widely applicable. At least one analysis [13] has led to doubts that important ethnic (and gender) differences in the way the body handles drugs are more than rare, but the questions of possible differences may have to be considered for new drugs or reexamined for drugs studied mainly in a single ethnic (or gender) group. If answers to such questions are going to be sought by designers of clinical trials, how should ethnic groups be designated?

Substantial effort [7-10] has already gone into considering this and related questions for epidemiologic research. One clear conclusion [7] is that epidemiologic reports "should state explicitly how (ethnic group) classifications were made." These and other conclusions [7-10] are also relevant to the design and reporting of clinical trials. Clinical-trials research has in common with epidemiology the sampling of populations that may be diverse in geographic origin and cultural customs. Hence, subjects in clinical trials [14] may differ in their genetic determinants of responses to external agents and in their dietary patterns and other cultural influences on responses to drugs. Designers of clinical trials, as well as epidemiologists, must, however, keep clearly in mind that even within properly defined ethnic groups social, cultural, and economic characteristics may differ widely. One Euro-American may be living in poverty and malnourished; another may be a highly paid, well-fed executive in a major corporation. Hence, designers of trials must not rely simply on ethnic identifications if more specific variables covered by ethnicity can be defined for the research design.

These tentative conclusions on ethnic identifications and how they may be properly applied in clinical trials probably also have implications for editors' and readers' expectations of reports on clinical trials. A few months ago, this journal published a position paper [15] proposing the kinds of information that should be included in the introductions, methods and results sections, and discussions of papers reporting trial results. This paper calls for details on "planned study population," "inclusion and exclusion criteria," "planned subgroup analyses," and "demographics ... of the study population." For reports of trials that identify the ethnicities of persons, those details could be expected to include statements on whether the investigators categorized persons according to ethnicity and on the definitions that were used, or on whether persons categorized themselves using their own terms or terms supplied by the investigators. Such information could be crucial for scientifically sound judgments about apparent ethnic differences and the generalizability of results from clinical trials.