MT25 Assessing the Impact of a Patient Portal Application for Cancer Patients on Missed Appointments


Objectives Caregiving dyads in palliative care are confronted with complex care needs. Respite care services can be highly beneficial in alleviating the caregiving burden, supporting survivorship and dying at home. Yet, respite care services are difficult to locate and access in the province of Quebec, Canada, particularly when navigating ubiquitous sources of online health information of varying quality. Objective A digital patient-centred portal is currently available to cancer patients at the McGill University Health Centre (MUHC), providing real-time access to personal health records and information on upcoming appointments. We aimed to quantify the initial impact of patient portal use on missed appointments. Methods We conducted a retrospective analysis to compare differences in missed appointment (i.e., no-show) rate between portal users and non-users who had appointments at the Cedars Cancer Centre between May 31st, 2018, and July 29th, 2022. We linked the portal data to patient-level data to match users and non-users 1-1 using a nearest neighbour matching approach without replacement, controlling for patient demographic (sex and age) and clinical characteristics (cancer diagnosis, staging, treatment, and progression). We fit a logistic regression model to investigate the association between these covariates and portal use. Differences in no-show rate were compared using two-proportions Z-test. Results A total of 654 cancer patients were matched. Adequate balance between users and non-users was achieved after matching. Radiation treatment could predict a significantly higher odd (3.287, p<0.01) of patients being users. Female patients and an increase in age were associated with lower odds (0.674, p=0.017; 0.969, p<0.01). The overall no-show rate for portal users seemed lower compared to non-users (0.052 vs. 0.058, 95% confidence interval [-0.0125 to 0.0006], p=0.076). While the finding was not statistically significant with an alpha of 0.05, the confidence interval upper bound was close to 0. Conclusions Patient portal usage may be associated with a modest decrease in patient no-show rate, but results were not statistically significant. Additional features like appointment reminders might lead to greater reductions in patient no-shows. We cannot disambiguate whether covariates predicting portal usage are due to self-selection or uneven roll-out. Fully assessing the impact of the patient portal will require considering more outcomes in larger patient populations.