(151) Case Study With a Cardiovascular Outcomes Trial for a Glucagon-Like Peptide-1 Receptor Agonist: Evaluating Trial Generalizability With an Enriched Electronic Health Record-Based Database

Background: Because cardiovascular (CV) disease is a leading cause of morbidity and mortality in patients with diabetes, a number of clinical trials have been conducted to assess the impact of novel diabetes therapies, such as glucagon-like peptide-1 receptor agonists (GLP-1 RAs), on CV risk. Subsequent evaluations have reported low generalizability of these trials’ criteria to real-world populations but were limited by gaps in relevant data.

Objectives: To revisit a CV outcomes trial with previously reported low CV generalizability—for purposes of this study, the SUSTAIN-6 trial for the GLP-1 RA semaglutide—using a large, novel multicenter electronic health records (EHR)-based database (egnite Database, egnite, Inc., Aliso Viejo, CA, USA) enriched with key CV-relevant parameters.

Methods: This case study used data from the egnite Database (Jan 1, 2018 – Feb 6, 2024), derived from 23 institutions with appropriate permissions, to retrospectively assess patients with a diagnosis of type 2 diabetes (but not type 1 diabetes) who fulfilled minimum data availability requirements (≥1 hemoglobin A1c [HbA1c] and estimated glomerular filtration rate laboratory result; ≥1 echocardiogram) and describe a population potentially eligible for the SUSTAIN-6 trial per key CV enrollment criteria.

Results: Of 187,307 patients with type 2 diabetes who met all minimum data availability requirements, up to 128,249 (68.5%) were potentially eligible for SUSTAIN-6 at some point within this study’s window per key CV criteria. In a sensitivity analysis where minimum data availability did not require a documented echocardiogram, up to 55.2% were still found to have been potentially eligible per key CV criteria. Characteristics of the potentially eligible main cohort were comparable to that of the final SUSTAIN-6 trial population (eg, mean ± standard deviation HbA1c 7.6% ± 1.9% versus 8.7% ± 1.5%, and body mass index of 34.0 ± 7.6 versus 32.8 ± 6.2 kg/m2, respectively). Of note, up to 71.9% of these potentially eligible patients were considered still active (not palliative; not deceased; ≥1 encounter with system in the past year) within their respective healthcare institutions as of dataset lock, but only 26.6% of them had any documented GLP-1 RA prescription history.

Conclusions: This case study suggests that the generalizability of a CV outcomes trial (here, for semaglutide) may be greater than previously reported and that an EHR-based dataset can meaningfully influence the findings of such assessments. Interestingly, it also observed that of contemporary patients who may stand to derive a meaningful benefit per major CV outcomes trials for GLP-1 RAs, only a minority have ever had a documented prescription of a GLP-1 RA, signaling a potential opportunity for quality improvement in diabetes care.