Intended for healthcare professionals


Describing race, ethnicity, and culture in medical research

BMJ 1996; 312 doi: (Published 27 April 1996) Cite this as: BMJ 1996;312:1054
  1. Kwame McKenzie,
  2. N S Crowcroft
  1. Honorary research fellow Department of Psychological Medicine, King's College Hospital Medical School and Institute of Psychiatry, London SE5 8AX
  2. Fellow European Programme for Intervention Epidemiology Training, Institute of Hygiene and Epidemiology, Brussels 1050, Belgium

    Describing the groups studied is better than trying to find a catch all name

    The terminology of race, ethnicity, and culture is a source of continuing debate and will change because of fashion and politics.1 Given our diversity, the fact that ethnicity is now most often self classified, and that both ethnicity and culture are dynamic it seems unlikely that an agreed taxonomy can be achieved. But if researchers want to be able to compare results of studies now and in the future a framework is needed for the classification of ethnic or cultural groups.

    The nearest we have to an agreed classification in the United Kingdom are the categories used by the Office of Population Censuses and Surveys in the 1991 census.2 This is a pragmatic classification that balances ease of collection against a need to produce data on the population.3 It is of limited use as a measure of sociocultural differences.3

    This leaves a void in which there are many different terms used for groups. For instance, a black Baptist born in the UK but whose parents were born in Jamaica might be called Afro-Caribbean, black British, of Caribbean origin UK born, West Indian, and, of course, Jamaican. It also leaves those who wish to research ethnic differences in health with studies that are difficult to compare and in the position of having to decide on an ad hoc basis what to call their ethnic groupings.

    Terminology should reflect the hypothesis under consideration.1 For example, the OPCS classification may be adequate for assessing access to services.3 If cultural differences are thought to be the predominant influence on patterns of disease, however, those cultural differences should be measured and govern the categorisation used. For instance, if it were thought that religion and being born in the UK were important determinants of the risk of committing suicide, then groups for comparison could logically be: UK born practising Muslim, non-UK born practising Muslims, UK born practising Sikh, UK born non-practising Sikh, UK born practising Baptist, and so on. If researchers believe that race underlies differences in disease patterns then the onus is on them to prove that race has a biological correlate relevant to the disease in question, since race is a poor predictor of biological difference or risk.4 5

    When it comes to reporting research the aim is to produce results that are easily compared, now and in the future, by following the basic principle taught to medical students: if you do not know the right name for something then describe it. Names should be as descriptive as possible and should reflect how these groups were demarcated. Authors should describe fully in their methods section the logic behind their ethnic groupings and how these were assigned. In their discussion authors should be careful not to draw conclusions which go beyond their data. For instance, it would be difficult for researchers who had demarcated people on racial grounds to proffer a cultural explanation for their findings.

    The BMJ's new guidelines on this subject (see p 1094) do not try directly to answer the vexed question “What should we call them?” but gives advice on how to go about making such a decision. The terms used should be those which most accurately reflect how people have been assigned to groups.

    In the real world many researchers find it difficult to know whether ethnicity, culture, or race are important and so cannot produce hypothesis led classifications at the start of their research. The simplest solution is to collect a range of information that will help describe the groups being researched. This can be collected by using the OPCS categories and then adding extra information, driven by the hypothesis, such as country of birth, parents' country of birth, mother tongue, special diets, religion practised, and years in the UK. Information on socioeconomic status should also be collected, since it is a confounder that is often neglected in comparisons between ethnic groups.6 The relative contributions of each factor can then be estimated, and when the research is reported the terminology should reflect the nature of the factors that have been evaluated.

    Discussion about research into cultural or ethnic grouping is often reduced to arguments about terminology. In our experience doctors often try to use names that will offend as few people as possible rather than those that impart as much information as possible. They worry not only about the groups they are researching but also about editors and readers. “UK Afro-Caribbean” is easier to read than “black people of Caribbean origin born and living in the UK,” but less informative.

    Nevertheless, some readers will probably be disappointed that this editorial does not contain a simple list of politically correct terms to use. Apart from the fact that fashions in terminology change, the more important objection is that such a list would not curtail the growing number of reports describing work that cannot be interpreted or compared. Until we agree on how culturally or ethnically to demarcate the people of the world—and people stop mixing—the best advice for researchers remains to collect as much information as possible and describe what they have done.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.