197-OR: Algorithmic Identification May Improve Racial and Ethnic Diversity of Clinical Study Recruitment

Introduction: Early analyses of Rare and Atypical Diabetes Network (RADIANT) study recruitment suggested that goals to recruit underrepresented groups were not being met. We tested whether a validated electronic health record (EHR) algorithm to identify people with an atypical form of diabetes improved identification of racially and ethnically minoritized individuals who may be candidates for the RADIANT study. Methods: Individuals identified by the algorithm were reviewed by research assistants, then classified by endocrinologists as atypical diabetes, a known type of diabetes, or unable to classify (more information needed). Chi-squared tests were used to compare the proportion of self-reported non-Hispanic Black (NHB) and Hispanic or Spanish-speaking (H/SS) participants enrolled through the Mass General site prior to use of the algorithm (mainly through referral by expert clinicians) and proportion of potential individuals identified by the algorithm. Results: Prior to beginning recruitment through the EHR algorithm, 53% of participants enrolled in RADIANT from the Mass General Brigham site identified as NHW, 5% as NHB, 5% as H/SS, 20% as non-Hispanic Asian (NHA). The algorithm initially identified 539 individuals with potentially atypical forms of diabetes. Of these, 452 under the age of 85 were reviewed, and 93 (20.6%) were classified as atypical and possible RADIANT candidates (v. 65.7% with a known type of diabetes and 13.7% unable to be classified). Of those with likely atypical diabetes, 39.8% identified as NHW, 22.6% as NHB, 11.8% as H/SS, and 20.4% as NHA. The algorithm identified a higher percentage of NHB individuals (p