Insights Into Patient Portal Engagement Leveraging Observational Electronic Health Data

OBJECTIVE: To understand patient portal engagement stratified by patient characteristics among adults 50 years and older with at least 1 common chronic medical condition using electronic health records data. STUDY DESIGN: This exploratory study retrospectively analyzed categorical and numerical data for a patient cohort receiving any kind of care in the M Health Fairview system. METHODS: Data were retrieved from the Epic Clarity database for patients 50 years and older with 1 or more chronic illnesses during the study period and mapped to International Classification of Diseases codes. Portal activation and usage and health care encounters were stratified by patient characteristics. We performed descriptive analysis, Spearman correlation, and multivariable regression. RESULTS: Of 250,345 adults 50 years and older with at least 1 chronic condition, 61% of them activated the portal and 54% logged at least 1 session between 2011 and 2024. Enrollment disparities were observed by age, race, language, education, and certain chronic conditions. Lower usage was noted among adults 65 years and older (42%), Black patients (40%), non-English speakers (Hmong [38%], Somali [21%], Spanish [28%]), those with less than a college degree (no diploma [53%], General Educational Development/high school diploma [76%]), and patients with certain conditions. Patients with diabetes, neoplasms, ischemic heart disease, and hypertension showed greater engagement, and those with heart failure or chronic obstructive pulmonary disease had lower engagement. Higher portal use was correlated with a higher number of completed encounters but less so with no-shows. Odds of portal use were lower for patients who were 65 years and older, men, non-White, and non-English speakers. Those with neoplasms, heart disease, and hypertension had highest odds of portal usage. Proxy usage was minimal. CONCLUSIONS: Disparities in patient portal use among adults with chronic conditions varied by patient characteristics. Precisely targeted strategies toward suboptimal users of patient portals could enhance their adoption and sustained use.