- How generalisable is your sample?
- Has geographical need informed your choice of site(s)?
- Are you excluding anyone, if so is this justified?
- How is data on equality, diversity and inclusion collected?
Selection of participants, sites and samplesSometimes in research you might want to carry out your study on a very homogeneous sample that will show less variation or ‘noise’ to improve the chance of seeing a difference in your analysis. Although this makes sense numerically, and in the realm of personalised medicine where you tailor a drug to a person’s genotype or phenotype it may be justified (Vogenberg et al. 2010), the purpose of most studies is to improve the health and social care of all groups of people within the population being studied.
All research inclusive of trials and qualitative studies should collect demographic data of the sample studied. This is usually presented in a table, or could be in narrative form in qualitative studies. This demographic table should include such characteristics as age, sex, gender, and ethnicity. Collecting this data is now an expectation of NIHR-funded research projects. In addition, researchers should evidence how their recruitment strategies will be inclusive. The NIHR INCLUDE pages have a number of questions that can help with making your research more inclusive.
It is important to consider how well your planned study sample will reflect the population to which the results will be applied and eventually, if the study is successful, implemented on. If the population is heterogenous in terms of demographic characteristics, then your sample should reflect this; otherwise, you could be contributing to health or care inequalities.
Your sample should be relevant to the population and should avoid having unnecessary exclusion criteria. These exclusion criteria sometimes have deep-rooted assumptions; for example:
If exclusions are stated, these need to be justified and supported by compelling evidence. In some disease areas, exclusion based on personal characteristics is required. For example, for asthma diagnosis children less than 5 years old are often excluded due to difficulty in gaining objective tests (NICE 2021). However, if we consider ethnicity as a characteristic, we find that although asthma has a similar prevalence amongst British South Asian people as compared to British white people, South Asian people have a disproportionately higher rate of A&E attendances for asthma attacks (Griffiths et al. 2001). This suggests that they may have different needs or require adjusted logic models when creating an asthma intervention. Without this ethnic diversity included in recruitment, an intervention could fail in this part of the population. It is important to match your sample to the study population, including the groups of people in greatest need.
In some research it may be justified to use oversampling in order to get an adequate sample of a minority group in your research and there are several methods to do this (Vickers et al. 2012; Kalton 2009). In other studies research is carried out in geographical areas where disease or care burden is smaller than other areas that would benefit more from such research (Bower et al. 2020). Research is often carried out on samples located near major research centres and more rural areas are neglected (Smith et al. 2016). NIHR is increasingly keen to see that research is being conducted where the need is, and not only in large cities or leading university hospital trusts.
It is important to demonstrate the diversity of your sample through data collection. If this data capture is carried out in feasibility or pilot work, there is still a chance to tailor the study to accommodate a greater diversity of demographic characteristics which would make the outputs and findings of research more generalisable. NIHR acknowledge that diverse sampling might cost more, and from 27 November 2024 require research inclusion costs to be specified to enable inclusive research with diverse participants.
Key messages
|
A reflection from a researcher on almost missing out on diverse ethnic perspectives in a survey during the Covid-19 pandemic “Involvement and engagement was unusually challenging in 2020, with the tension between tight deadlines and time/resources for involvement heightened by the looming threat of a global pandemic. Asked to assess risk perceptions of attending hospital in a global pandemic for both research and clinical purposes, an 81-item questionnaire was developed and used to collect 402 responses from residents in a Midlands city. 91% of respondents were white. Survey findings were used to inform local and national policy on running research during the pandemic, without including the voices of the very populations most affected by the pandemic at least in terms of morbidity and mortality. There were just enough Black, Asian and minority ethnic responses to tell us there was a significant difference in their opinions. This demonstrates the importance of having access to rapid translation services and good relationships with people and communities from diverse backgrounds who will help you with your research. In the case of this questionnaire, the Asian participants were almost all women from a programme that actively seeks to engage and build relationships with the Asian communities in one area of the city, but the investment in relationship building had been insufficient to achieve decent representation. |
Bower, P. et al. (2020) ‘Is health research undertaken where the burden of disease is greatest? Observational study of geographical inequalities in recruitment to research in England 2013–2018’, BMC Medicine, 18: 133.
Griffiths C. et al. (2001) ‘Influences on hospital admission for asthma in south Asian and white adults: qualitative interview study’, BMJ, 323(7319): 962-966.
Kalton, G. (2009) ‘Methods for oversampling rare subpopulations in social surveys’, Survey Methodology, 35. No 2, 125-141.
NICE (2021) Asthma: diagnosis, monitoring and chronic asthma management, NICE guideline [NG80] (2017) Last updated: 22 March 2021. Available online at: https://www.nice.org.uk/guidance/ng80/resources (accessed 5/05/21).
Smith, T.A. et al. (2016) ‘Selecting, Adapting, and Implementing Evidence-based Interventions in Rural Settings: An Analysis of 79 Community Examples’, Journal of Health Care for the Poor and Underserved, 27(4): 181-193.
Vickers, T., Craig, G. and Atkin, K. (2012) Research with black and minority ethnic people using social care services, London: NIHR School for Social Care Research.
Vogenberg, F.R., Barash, C.I and Pursel, M. (2010) ‘Personalized Medicine: Part 1: Evolution and Development into Theranostics’, Pharmacy and Therapeutics, 35(10): 560-576.