Abstract
BACKGROUND: Type 2 diabetes is a serious, pervasive metabolic condition that disproportionately affects ethnic minority patients. Telehealth interventions can facilitate type 2 diabetes monitoring and prevent secondary complications. However, trials designed to test the effectiveness of telehealth interventions may underrecruit or exclude ethnic minority patients, with language a potential barrier to recruitment. The underrepresentation of minorities in trials limits the external validity of the findings for this key patient demographic. OBJECTIVE: This systematic review examines (1) the research reporting practices and prevalence of ethnic minority patients included in telehealth randomized controlled trials (RCTs) targeting type 2 diabetes and the trial characteristics associated with recruiting a high proportion of minority patients, and (2) the proportion of included RCTs that report using English language proficiency as a patient screening criterion and how and why they do so. METHODS: Telehealth RCTs published in refereed journals targeting type 2 diabetes as a primary condition for adults in Western majority English-speaking countries were included. Ethnically targeted RCTs were excluded from the main review, but were included in a post hoc subgroup analysis. Abstract and full-text screening, risk of bias assessment, and data extraction were independently conducted by two reviewers. RESULTS: Of 3358 records identified in the search, 79 articles comprising 58 RCTs were included. Nearly two-thirds of the RCTs (38/58) reported on the ethnic composition of participants, with a median proportion of 23.5% patients (range 0%-97.7%). Fourteen studies (24%) that included at least 30% minority patients were all US-based, predominantly recruited from urban areas, and described the target population as underserved, financially deprived, or uninsured. Eight of these 14 studies (57%) offered intervention materials in a language other than English or employed bilingual staff. Half of all identified RCTs (29/58) included language proficiency as a participant-screening criterion. Language proficiency was operationalized using nonstandardized measures (eg, having sufficient "verbal fluency"), with only three studies providing reasons for excluding patients on language grounds. CONCLUSIONS: There was considerable variability across studies in the inclusion of ethnic minority patients in RCTs, with higher participation rates in countries with legislation to mandate their inclusion (eg, United States) than in those without such legislation (eg, United Kingdom). Less than 25% of the RCTs recruited a sizeable proportion of ethnic minorities, which raises concerns about external validity. The lack of objective measures or common procedures for assessing language proficiency across trials implies that language-related eligibility decisions are often based on trial recruiters' impressionistic judgments, which could be subject to bias. The variability and inconsistent reporting on ethnicity and other socioeconomic factors in descriptions of research participants could be more specifically emphasized in trial reporting guidelines to promote best practice. TRIAL REGISTRATION: PROSPERO International Prospective Register of Systematic Reviews: CRD42015024899; http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015024899 (Archived by WebCite at http://www.webcitation.org/6kQmI2bdF).