Medical Law International - Social Inclusivity Vs Analytical Acuity? A Qualitative Study Of UK Researchers Regarding The Inclusion Of Minority Ethnic Groups In Biobanks

SOCIAL INCLUSIVITY VS ANALYTICAL ACUITY?
A QUALITATIVE STUDY OF UK RESEARCHERS REGARDING THE INCLUSION OF MINORITY ETHNIC GROUPS IN BIOBANKS
ANDREW SMART
Bath Spa University

RICHARD TUTTON
Lancaster University

RICHARD ASHCROFT
Queen Mary, University of London

PAUL MARTIN, ANDREW BALMER, RICHARD ELLIOT
University of Nottingham

GEORGE T.H. ELLISON
St George’s, University of London

D FINDINGS

We have discussed elsewhere (Ellison et al., 2007; Martin et al., 2007; Smart et al., 2008) how most of the researchers working in the biobanks we examined used the term ethnicity (rather than ‘race’) when classifying their study participants, how they conceptualised ethnicity as having both socio-cultural and genetic components and how they tended to adopt the categories developed for the UK censuses (or something very similar) for the purpose of classification. Nevertheless, as we mentioned earlier, some of the categories and labels used to identify ethnicity can overlap with those used in more traditional ‘racial’ classifications. This is notable because, despite the researchers’ ostensible rejection of ‘race’, some of the terminology used in the UK censuses categories which they adopted is based on just such geographical and phenotypic characteristics (e.g., ‘Asian’ and ‘Black’). As such, the researchers’ preference for using ‘ethnicity’ does not necessarily mean that their work was devoid of any ‘racial’ connotations. Indeed, there is some evidence to suggest that biomedical researchers have adopted ethnicity and ethnic classifications as sanitised versions of ‘race’ (Outram and Ellison, 2006). In our study, one of the researchers’ key justifications for adopting the categories used in UK censuses was that they enabled them to check the extent to which their study cohorts were ‘representative’ of the wider UK population—ie, the extent to which their study contained similar proportions of participants from different ethnic groups to those recorded in the general population. However, as we will see, the issue of having study cohorts that were ‘representative’ involved a complex interplay between the twin imperatives of social inclusivity and analytical acuity.

1. The imperative of social inclusivity
When asked to discuss the inclusion of participants from minority ethnic groups in their studies, the researchers we interviewed responded in three different ways. Inclusion was discussed as: a fundamental ethical obligation; an aspect of research governance; and a practical challenge. The following three extracts, from researchers involved in the same study of gene-environment interactions in common complex diseases, all reflect their opinion that there might be a fundamental right to participation on equal grounds within biomedical research, and how ignoring this right might ‘disenfranchise’ the group(s) concerned:

[It is] totally unacceptable from an ethical point of view, to exclude a substantial proportion of the population from this national study. (PGD 15)

[We’re] having other languages represented so that we don’t, that no one is disenfranchised as such from participation. (PGD 04)

… as a principle we mustn’t exclude any part of the community that is the UK population from research. [We aim to] create the resource in a way that that part of the UK population can basically be served… given that by definition they are going to be a minority component of [the study]. (PGD 10)

The final comment illustrates how, for this respondent, the desire for inclusivity was grounded in the idea of producing a ‘resource’ to ‘serve’ the whole of the UK population. However, a respondent working in a biobank that focused on exploring the aetiology of coronary heart disease noted that the ideal of equality of participation in research could create challenges in terms of recruitment policies and practices:

… we’ve just sought to deal with people as being equally eligible and equally, you know, then if anything trying to over-compensate to encourage ethnic minorities by, when we’ve had families who are from that, you know, using them to represent to other families, go on the radio to their own community (PGD 08)

In this project, the desire to promote equality became a practical matter that required allocating extra resources to recruitment.
Inclusion was also discussed in terms of being an external pressure—a requirement of research sponsors and ethical agencies. One researcher examining gene-environment interactions in common complex diseases reflected critically on funders’ requests to include participants from minority ethnic groups in the research project on which they were working:

… you get the feeling they just throw these things in just hoping that someone is going to pick out certain bits of it. You do really wonder what the [funder] had in mind when they put out these proposals to be honest. (PGD 14)

Other interviewees revealed the potential influence of ethics committees on the design and conduct of research. The following extracts are from a respondent who undertook studies of health and development, and from a researcher exploring renal disease, respectively:

… the pressure comes more from ethics than some funding groups… (PGD 03)

… where we come a-cropper are the bloody ethics committees. Absolutely a pain in the butt (PGD 12).

   In both instances, the respondents were discussing the requirements placed on them by ethics committees to ensure that research documentation and processes were accessible to those whose did not speak English.
   From these examples, we can suggest that the rationale for promoting the inclusion of participants from minority ethnic groups, and judgments about its feasibility, were subject to different pressures: researchers’ normative judgments about ‘fairness’; the processes of research governance; and practical concerns about the conduct of research. However, as shown below, such notions of inclusivity were accompanied by concerns about the perceived impact of ‘heterogeneous’ study samples.

2. The imperative of analytical acuity
   When asked to discuss whether ‘race’ or ethnicity were relevant to the sampling, analysis and/or interpretation of their research, six interviewees voiced concerns about analytical acuity (i.e., the precision and clarity) of their study findings. A common concern among these respondents centered on the potential limitations of ‘representative’ samples, given that these generate relatively small sub-samples of participants from minority ethnic groups that might limit the ‘statistical power’ required for robust comparative subgroup analyses. This viewpoint is illustrated in the extracts below which are, first, from a scientist undertaking studies of childhood health and development and, then, from two researchers examining geneenvironment interactions in common complex diseases:

So we’ve got about 5% of our children who are actively taking part who are from non-white background and they, by the time you split them by what background they are, you’ve got very small groups. And I think it’s very unlikely that we’re ever going to split them for analysis at all. (PGD02)

… it’s no good just having a small number, a small percentage. What’s the South Asians, is it 7% of the population, or something like that? So if you only have that sort of equivalent amount of people in your sample there wouldn’t be any point really. (PGD 14)

[The study] was powered to look at gene exposure characteristics… It is impossible to envisage that that same power calculation could be transferred down to fractional populations… in answer to people who complain that maybe this isn’t going to be a gene exposure study in people of ethnic minority groups, it’s because we’ve done the sums and they don’t add up, you know, we can’t make the cohorts large enough to do a meaningful [comparative] study [of ethnic ‘subpopulations’]. (PGD 04)

While ‘representative’ study samples may be one way of promoting inclusivity, none of the interviewees felt that this was a scientifically sound approach, and none felt that adopting ‘representative’ study samples would help them to ensure that their studies’ findings were applicable to the whole population. In fact, it was argued that ‘representative’ study samples inevitably provide less reliable evidence for any minority ethnic group contained in the sample compared to that provided for the ‘majority’ group, simply on the basis of the differences in statistical power provided by the smaller and larger samples of subgroups generated by ‘representative’ sampling. For this reason, it was argued that the drive for inclusivity would only be scientifically credible if it was suitably ‘powered’. The scientists who reported this concern suggested two possible responses: increasing the recruitment of participants from all minority ethnic groups (by over-sampling or ‘boosting’ the sampling of these groups); or restricting studies to equally-powered comparisons of predefined sets of ethnic groups (i.e., selectively sampling a limited number of specific ethnic groups).
A related concern was that including participants from different ethnic groups in study samples could (and would) introduce ‘unhelpful variation’ into analyses. As this scientist, studying hypertension, argued:

… in a sense you wanted to get as homogenous a group as possible and because of knowing that frequencies of variations in genes vary between the different ethnic groups in a sense that could be a confounder for when you’re going testing in case-control association type studies. (PGD 05)

One response to the perceived problem of confounding, due to that component of genetic variation ‘known’ to be associated with ethnicity, was to restrict studies to ‘single populations’. As this respondent suggested:

If you want to do a study on Afro-Caribbeans then you use a different sampling frame really, and similarly for the Indians or Bangladeshis or whoever. (PGD02)

   To this end, two of the biobanks in our sample restricted participation to the ‘majority’ ethnic group within the UK population, variously labelled as ‘white’, Caucasian’, ‘British’ or ‘European’.
   In summary, some of the scientists interviewed aimed to create study designs that would produce sufficiently large samples of one or more ethnic groups to generate robust findings for subsequent comparisons. This suggests that their decisions about the feasibility and value of exploring health differences across and between ethnic groups were (at least in part) judged according to the logic of inferential statistics. In particular, decisions about including participants from minority ethnic groups were made by calculating the minimum sample size required for analysis using ‘power calculations’. As such, inclusivity was conceptualised in terms of how differences between ethnic groups (particularly perceptions about ‘known’ genetic differences, rather than about social differences or social determinants of phenotypic differences) might compromise analytical acuity.

3. Discordance between social inclusivity and analytical acuity
   Our analysis has revealed the existence of discordance between the two potentially competing imperatives of social inclusivity and analytical acuity in the views and experiences of the biobank researchers we interviewed. The three examples below all relate to a biobank examining gene-environment interactions in common complex diseases. These all focus on perceived external influences on decisions taken by the project, and how these might relate to and impact on the scientific research the project would then be able to do:

[The study] is being funded by [a public funding council] and they want to include everybody in it.… to make it fair, as opposed to any genetic kind of justification… There may be some scientific benefit as well… but we don’t know that for certain and there may not be. (PGD 09)

… what is driving many of the actions are the pressures on them. And it’s what is acceptable in society, and what would be deemed acceptable. And what comes out look like perverse decisions. (PGD 11)

[Critics said] what you should really be doing is over-sampling [ie, ‘boosted’ sampling] and that’s when they’re getting ridiculous… it would be impossible to do and would then starting adding other power implications to the rest of the study. (PGD 15)

These extracts all reflect disquiet about how external, ‘societal’ pressures for inclusiveness were introduced into decision-making about sampling for the project. Responses to the perceived benefits and threats posed to scientific validity range from uncertainty about the unclear scientific benefits of ‘inclusive’ sampling, through concerns about the ‘perverse’ irrationality of decision-making, to fears about potential damage to the project’s feasibility (in terms of recruitment) and quality (in terms of ‘statistical power’).
At least one other interviewee shared these concerns about the motives of external drivers for inclusivity, and their impact on the logic of study design. The following scientist, who was mapping the genetic structure of populations to create a resource for future medical research, made a similar point but in a more disparaging tone:

[The funders] said: ‘well we think, you know, you’d better put in something about these ethnic people, otherwise it won’t be politically correct’; so we did and then the scientific evaluation said exactly what we’d said initially: ‘that’s no good you can’t get adequate information’. (PGD 16)

While we recognise that these remarks are potentially offensive, it is also important to note that the interviewee was exhibiting palpable irritation at perceived infringement of their control over scientific decision-making by those whose concerns were primarily social or political. As the researcher viewed it, they felt under pressure to ‘put something in’ to their study about ethnicity that was not scientifically justifiable.
There was, however, one researcher who adopted a more reflexive approach to the perceived discordance between social inclusivity and analytical acuity. This interviewee, who undertook studies of health and development, recognised that a study’s scientific aims, and the social context in which it takes place, are not easily disentangled:

… there is a problem about saying we’ve got to study everyone because they happen to be here, rather than because it’s actually going to address the scientific question that we want to address. But then there is another whole overlay on top of the scientific question about you know, not making people feel outcast or not welcome or not wanted and that’s, that’s the difficulty.… I do think we have an obligation to be inclusive but I do think we need to do it appropriately… we need to be clear about where ethnicity is important and where it perhaps doesn’t matter so much. (PGD 03)

This position appears far more sympathetic to what was described as an ‘obligation to be inclusive’ than that of the previous extract. However, the proposed solution of identifying ‘where ethnicity is important’ offers no clear solution to the conflict between social inclusivity and analytical acuity, because such decisions can be made on both scientific and socio-political grounds. Indeed, this respondent went on to reveal how the practice of being ‘inclusive’ has the potential to create fresh ethical dilemmas:

… this idea of being inclusive [must be] balanced against the waste of public money and resources, in terms of putting the effort to make them inclusive and then end up excluding them [minority ethnic groups] from the analysis. (PGD 03)

The final point made in this extract indicates that there is not only an ethical imperative for inclusiveness, but there may also be an ethical imperative against it, not least when resources (including participants’ time and goodwill) are unnecessarily squandered collecting data on participants from minority ethnic groups that cannot be, and are not, used in subsequent analyses.
This final set of extracts has revealed how the researchers we interviewed felt that it was necessary to protect the analytical acuity of their research. In this context, inclusivity was conceptualised in a range of ways: as a challenge to scientific rationales; as an example of unnecessary/unhelpful socio-political interference in their work; and as a site of fresh ethical quandaries. However, it is notable that the drive for inclusion of minority ethnic groups in research was always presented as a socio-political rather than a scientific imperative; only one of the interviewees (PGD09) made any allusion to its potential scientific benefits and even then this was done with a degree of scepticism.

<< C. Methods | E. Discussion >>