TL;DR
This study found that using a patient portal significantly improved patients' engagement in preventive health behaviors, such as getting flu shots and undergoing screenings. However, it did not show a significant impact on chronic health outcomes like blood pressure or diabetes management, suggesting portals are currently better at promoting proactive health actions than managing existing conditions.
What Problem Does This Paper Address?
The widespread adoption of patient portals in healthcare systems has led to increased focus on their potential to improve patient engagement and health outcomes. However, there's a critical gap in understanding the actual impact of these portals on patient health behaviors and outcomes. Many studies have explored patient and provider attitudes or portal usability, but robust evidence quantifying the effect on health metrics, especially after accounting for confounding factors, has been scarce. This paper aims to quantify this impact, specifically looking at how portal use influences both proactive preventive health actions and the management of existing chronic conditions.
Methods
The researchers conducted a retrospective observational cohort study using electronic health records (EHR) from the University of Pennsylvania Health System (UPHS). They analyzed data from 10,000 patients aged 50 and older treated between September 2014 and October 2016.
A key methodological contribution is the use of a novel cardinality matching approach based on propensity scores. This technique is designed to address the inherent selection bias in observational studies: patients who use portals might already be more health-conscious or have better access to care, which could confound the results. The cardinality matching procedure aims to create balanced groups of portal users and non-users by matching them on a wide range of patient and provider characteristics. This is followed by a bootstrapping method to estimate the variance of the association estimates, which is more robust than traditional statistical tests for matched data.
Datasets
The study utilized the University of Pennsylvania Health System (UPHS) electronic health records (EHR). This dataset provided information on patient demographics, insurance, health behaviors, and chronic health conditions for a cohort of 10,000 patients. No new datasets were introduced.
Key Insights and Ideas
- Preventive Care Boost: The most striking finding is the significant positive impact of patient portal use on preventive health behaviors. This suggests that portals are effective tools for encouraging patients to take proactive steps for their health.
- Chronic Condition Management Lag: Despite the improvements in preventive care, the study found no significant impact on chronic health outcomes. This implies that while portals can encourage actions related to health, they may not yet be effectively driving improvements in managing ongoing conditions.
- User Demographics Matter: Portal users and non-users differed significantly in demographics, with users being younger, more likely to be white, have commercial insurance, and have higher incomes. This highlights potential disparities in access and adoption that need to be considered.
- Sophisticated Matching is Crucial: The study emphasizes the importance of advanced statistical techniques like cardinality matching to control for confounding variables. Simply comparing users and non-users without such adjustments would likely lead to biased conclusions.
Key Results
After matching for potential confounders, the study found:
- Preventive Health Behaviors: Portal users showed significantly higher rates of annual flu vaccination (OR=1.58), blood pressure checks (OR=1.13), and lipid level screenings (OR=1.50). The composite prevention score was also higher for users (mean difference=0.22).
- Chronic Health Outcomes: No statistically significant differences were found for systolic blood pressure, LDL levels, diabetes status, or hypertension status between portal users and non-users.
- Demographic Differences (Pre-Matching): Portal users were, on average, younger (63.46 vs. 66.08 years), had higher incomes ($74,172 vs. $62,940), and were more likely to be white (72.77% vs. 52.58%) and have commercial insurance (60.99% vs. 40.12%) compared to non-users.
Limitations and Future Work
The authors acknowledge several limitations:
- Focus on a Specific Cohort: The study focused on relatively healthy patients aged 50 and older. The findings may not be generalizable to younger populations or those with severe chronic diseases.
- Limited Time Window: The data was collected over a specific period (2014-2016). A longer time window could provide insights into the temporal relationship between portal usage and the development of chronic conditions.
- Unknown Portal Usage Frequency: The study did not track how often patients logged into the portal. The impact might be attenuated if portal usage frequency varies greatly among users.
Future research directions suggested include:
- Investigating the impact of portal use in different patient cohorts.
- Studying the longitudinal impact of portal use on the trajectory of chronic health outcomes.
- Exploring the effect of portal usage frequency on health behaviors and outcomes.
- Understanding how patient portals can ultimately improve chronic health conditions.
Summary updated