Abstract
BACKGROUND: Patient portals are a common electronic medical record tool that allow for the asynchronous exchange of health information between patients and their health care teams. Patients can leverage patient portals to perform tasks such as viewing test results, reviewing clinical notes, and messaging their health care team. The impact of patient portal use on clinical outcomes in cirrhosis is unknown. OBJECTIVE: In this study, we evaluated the relationship between patient portal use patterns and readmissions in cirrhosis. METHODS: We identified 131 patients with decompensated cirrhosis with an index cirrhosis-related admission between May 1, 2018, and May 1, 2019. We then examined patient portal enrollment and use data during the 6-month period preceding the study period. Portal functions evaluated included sending a message, reading a message, and reading a test result. Use was categorized as active (sending a message) and passive (reading a message or test result) and was further stratified as no, moderate, or frequent use based on the frequency of portal function use compared to the mean. The primary outcomes were 90-day and overall readmissions, adjusted for age, model for end-stage liver disease-sodium, alcohol-related cirrhosis etiology, ascites, and hepatic encephalopathy. Portal functions assessed included sending a message, reading a message, and reading a result; the total number of times a portal function was performed was divided by the number of months the patient was enrolled in the patient portal during the 6-month period. RESULTS: The study population was 50.4% (66/131) female, with a mean age of 58 years. Enrollment in the patient portal was 63.4% (83/131), and there was no significant difference in enrollment based on clinical or demographic characteristics. For the entire cohort, 14.5% (19/131) and 22.1% (29/131) of patients were moderate and frequent active users, respectively. Of those enrolled in the patient portal, 97.6% (81/83) of patients were moderate or frequent passive users for both reading a message and reading a test result. Moderate active users had less 90-day readmissions (odds ratio 0.77, 95% CI 0.60-1.00) and overall readmissions (subdistribution hazard ratio 0.42, 95% CI 0.21-0.84), compared to nonactive users. There was no relationship between readmissions and passive use. CONCLUSIONS: Passive use of the patient portal is very high but is not associated with the risk of readmissions in people with decompensated cirrhosis. However, moderately active use of the patient portal is associated with a reduced risk of readmissions. Further work is needed to identify possible confounders and refine key use behaviors that may be protective with regard to the risk of readmission in this population.