RESUMEN
AIMS/HYPOTHESIS: Type 2 diabetes mellitus prevalence is increasing globally and the greatest burden is borne by racialised people. However, there are concerns that the enrolment of racialised people into RCTs is limited, resulting in a lack of ethnic and racial diversity. This may differ depending whether an RCT is government funded or industry funded. The aim of this study was to review the proportions of racialised and white participants included in large RCTs of type 2 diabetes pharmacotherapies relative to the disease burden of type 2 diabetes in these groups. METHODS: The Ovid MEDLINE database was searched from 1 January 2000 to 31 December 2020. English language reports of RCTs of type 2 diabetes pharmacotherapies published in select medical journals were included. Studies were included in this review if they had a sample size of at least 100 participants and all participants were adults with type 2 diabetes. Industry-funded trials must have recruited participants from at least two countries. Government-funded trials were not held to the same standard because they are typically conducted in a single country. Data including the numbers and proportions of participants by ethnicity and race were extracted from trial reports. The participation-to-prevalence ratio (PPR) was calculated for each trial by dividing the percentage of white and racialised participants in each trial by the percentage of white and racialised participants with type 2 diabetes, respectively, for the regions of recruitment. A random-effects meta-analysis was used to generate the pooled PPRs and 95% CIs across study types. A PPR <0.80 indicates under-representation and a PPR >1.20 indicates over-representation. Risk of bias assessments were not conducted for this study as the objective was to examine recruitment of racialised and white participants rather than evaluate the trustworthiness of clinical trial outcomes. RESULTS: A total of 83 trials were included, involving 283,122 participants, of which 15 were government-funded and 68 were industry-funded trials. In government-funded trials, the PPR for white participants was 1.11 (95% CI 0.99, 1.24) and the PPR for racialised participants was 0.72 (95% CI 0.60, 0.86). In industry-funded trials, the PPR for white participants was 1.95 (95% CI 1.74, 2.18) and the PPR for racialised participants was 0.36 (95% CI 0.32, 0.42). The limitations of this study include the reliance on investigator-reported ethnicity and race to classify participants as 'white' or 'racialised', the use of estimates for type 2 diabetes prevalence and demographic data, and the high levels of heterogeneity of pooled estimates. However, despite these limitations, the results were consistent with respect to direction. CONCLUSIONS/INTERPRETATION: Racialised participants are under-represented in government- and industry-funded type 2 diabetes trials. Strategies to improve recruitment and enrolment of racialised participants into RCTs should be developed. REGISTRATION: Open Science Framework registration no. f59mk ( https://osf.io/f59mk ) FUNDING: The authors received no financial support for this research or authorship of the article.
Asunto(s)
Diabetes Mellitus Tipo 2 , Adulto , Humanos , Proyectos de Investigación , Costo de Enfermedad , PrevalenciaRESUMEN
Race, ethnicity, and ancestry are common classification variables used in health research. However, there has been no formal agreement on the definitions of these terms, resulting in misuse, confusion, and a lack of clarity surrounding these concepts for researchers and their readers. This article examines past and current understandings of race, ethnicity, and ancestry in research, identifies the distinctions between these terms, examines the reliability of these terms, and provides researchers with guidance on how to use these terms. Although race, ethnicity, and ancestry are often treated synonymously, they should be considered as distinct terms in the context of health research. Researchers should carefully consider which term is most appropriate for their study, define and use the terms consistently, and consider how their classification may be used in future research by others. The classification should be self-reported rather than assigned by an observer wherever possible.