Language-related eligibility criteria in UK randomised trials: a systematic review of extended research reports

Background

Trial teams frequently make language-related, non-clinical eligibility decisions during recruitment. They need to ensure that patients understand the conditions and implications of trial participation and either have the necessary language skills to participate, or receive appropriate accommodations (e.g. translation or interpreting). Fair and consistent assessments are necessary to avoid unduly excluding patients, which could limit external validity and exacerbate inequalities. This study examines how trial teams make language-related eligibility decisions.

Methods

We conducted a systematic review of National Institute for Health and Care Research (NIHR) research reports (2010–2022) for UK-based randomised controlled trials (RCTs) recruiting adults for two conditions that disproportionately affect ethnic minority populations: clinical depression and type 2 diabetes mellitus (T2DM). Two researchers independently screened titles and abstracts and extracted data. We analysed the communication demands of the interventions and primary outcome measures in relation to how language screening was reported, including procedures or instruments used as proxies for language-related gatekeeping.

Results

We assessed 185 titles and abstracts from NIHR monographs. Thirty-two RCTs (23 depression, 9 T2DM) ultimately met our inclusion criteria. Ethnic diversity was minimal, particularly in the depression RCTs, where the median proportion of White participants was 97%. Language screening practices were inconsistent across studies and were often poorly aligned with the actual linguistic demands of the trial. Half of the included RCTs explicitly reported a language-based eligibility criterion, including 63% of trials evaluating talking therapies for depression compared to 27% of trials assessing pharmacological, device-based, or surgical interventions. Explicit and implicit language-related gatekeeping measures included the ability to complete research assessments involving language (sometimes to a prespecified score cut-point), provide informed consent, and engage in the intervention as judged by recruiters. Translation and interpreting support were mentioned in one depression study.

Conclusions

This review exposes methodological practices that may impede diverse patients’ participation. Linguistic demands of the interventions and outcomes need to be considered in justifying language-related screening and accommodations. Participants’ language variables need to be disentangled from ethnicity through routine data collection. A purpose-built screening tool that is universally applied to all participants could lead to fairer, more consistent assessments.

Trial registration

PROSPERO International Prospective Register of Systematic Reviews CRD42021267905. https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=267905. Registered on October 21 2021.