Speaking the Language of Inclusion: Examining English Languages Requirements in Cardiovascular Digital Health Trials

BackgroundCardiovascular medicine is rapidly evolving, as it integrates digital technologies intended to decentralize care from the clinic and/or hospital setting into patients homes and communities. However, clinical trials of digital technologies are frequently written and communicated solely in English, potentially exacerbating existing racial and ethnic differences in cardiovascular-related health outcomes. In this work, we evaluated the language-related inclusion and exclusion criteria of digital health trials related to cardiovascular medicine registered on ClinicalTrials.gov. MethodsPlanned, ongoing, and completed digital health trials were identified using the ClinicalTrials.gov website interface using terms related to digital health (see https://cardio-lang-inclusion.streamlit.app/ for detailed methods and data visualization). Only digital trials testing an intervention, conducted within the United States, and studying adults were included in analyses. The identified digital health trials were further subset to select for those related to cardiovascular medicine (i.e., prevention, weight loss and cardiometabolic disease, coronary disease, vascular disease, and arrhythmias). Natural language processing was leveraged to process the 1192 digital health trials inclusion and exclusion criteria. ResultsThere were a total of 1192 digital health trials related to cardiovascular medicine were identified. Over a third (470 (39.4%)) listed English language fluency as an inclusion criteria, while 196 (16.4%) excluded non-English language fluent participants. In total, 611 (51.3%) of the analyzed digital trials in cardiovascular medicine excluded participation of non-English language fluent patients (see Figure 1). Manual chart review of 50 random digital health trials in cardiovascular medicine demonstrated that 23 (46%) required English language fluency as an inclusion criteria and 2 (4%) excluded non-English language fluent patients from participation. These findings align with the broader automated natural language processing analysis. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=149 SRC="FIGDIR/small/25329222v1_fig1.gif" ALT="Figure 1"> View larger version (25K): org.highwire.dtl.DTLVardef@1b65425org.highwire.dtl.DTLVardef@14e5966org.highwire.dtl.DTLVardef@cf76c8org.highwire.dtl.DTLVardef@7f6ff8_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOFigure 1.C_FLOATNO Use of English language fluency as an inclusion or exclusion criteria in clinical trials of digital health interventions (N=1196). A total of 611 (51.3%) of digital health trials used English language fluency as either an inclusion or exclusion criteria (i.e., excluded non-English language fluent patients from participation). 196 (16.4%) excluded non-English language fluent patients. 470 (39.4%) required English language fluency as an inclusion criteria for participation. Created in BioRender. Kim, D. (2025) https://BioRender.com/o46f078 C_FIG ConclusionsWe found that there was frequent exclusion of non-English language fluent patients from digital health trials in cardiovascular medicine. This may limit the generalizability of digital health trials, which is critical as we transition to decentralized clinical practice using computers and smartphones and require an accompanying rigorous evidence base. What is the clinical question being addressed?Trials in digital health use text curated by investigators. However, with the large numbers of non-English-fluent individuals in the US, these trials may exclude minority populations. Hence, we sought to investigate whether digital health trials required English fluency for participation. What is the main finding?611/1192 (51.3%) digital health trials required English language fluency for participation. This limits the generalizability of current digital health trials. As we move toward more decentralized care, we will need a broad evidence base that includes non-English-fluent individuals.