Disparities in Patient Portal Use Among Adults With Chronic Conditions

Key Points Question Did patient access and use of patient health care portals change during the COVID-19 pandemic, and are there differences by sociodemographic characteristics? Findings In this cohort study of 536 participants, significant disparities in portal use by sex, age, multimorbidity, and health literacy were found. While disparities by sex and age decreased and were no longer statistically significant by 2021, disparities by multimorbidity remained consistent throughout the pandemic and disparities by health literacy were exacerbated. Meaning These findings suggest that health systems and practices must understand and address persistent disparities in patient portal utilization among some populations (eg, lower health literacy) as they leverage digital health tools.


Introduction
The COVID-19 pandemic disrupted face-to-face health care delivery and accelerated the adoption and use of digital health modalities, like patient portals. 1,2][9][10] There have also been well-documented disparities: patients with lower socioeconomic status (SES), educational attainment, health literacy, and subsequently those in racial and ethnic minority communities (eg, Black patients) have had lower portal adoption, access, and use. 4,7,11,12Studies have also highlighted patient-specific barriers, including concerns around privacy and security, access to technology and internet, limited digital or technology literacy, limited health literacy, and a general preference for the face-to-face modality of care. 9,13ss is known about how portal adoption and use have shifted during the COVID-19 pandemic, despite greater health system adoption and meaningful use.5][16][17] Although studies have indicated that patients have found portals to be useful tools for managing health during the pandemic, digital literacy was found to be a significant barrier to portal use. 18,19Limited information is currently available on the longitudinal trends of portal use during the pandemic era.[22] In this investigation, we evaluated portal use between 2019 and 2022 among a diverse cohort of middle-aged and older adults with at least 1 chronic condition at a large health system.The study aims were to characterize level of portal use, evaluate temporal changes in use, and to examine any sociodemographic disparities in use.

Methods
This cohort study used data from the COVID-19 & Chronic Conditions (C3) study, approved by Northwestern University institutional review board.As part of their participation in the C3 study, participants provided written informed consent and completed Health Insurance Portability and Accountability Act authorization.
This study was a retrospective cohort study of an ongoing, longitudinal cohort study, the C3 study.The C3 study is a telephone-based survey of participants enrolled in 1 of 5 primary care-based, National Institutes of Health-funded studies (eTable 1 in Supplement 1).The objective was to track the experiences of middle-aged and older adults with underlying health conditions that placed them at higher risk for SARS-CoV-2 infection and adverse outcomes from COVID-19 through the pandemic.
C3 parent studies were chosen due to enrollment of participants who would have greater risk for COVID-19 (eg, largely middle-aged or older adult participants, with Ն1 chronic conditions) and included detailed information on sociodemographic characteristics (eg, education, income), health literacy, and patient-reported outcomes that are not routinely collected in clinical care.
To assess the prevalence of C3 participants' portal use before, during, and after the most restrictive phase of the pandemic and examine sociodemographic disparities in portal use, data from the C3 cohort were matched (using unique patient hospital identification numbers) to data on use and activity of Northwestern Medicine's patient portal (ie, MyChart; Epic Systems), recorded by the enterprise data warehouse (EDW) between January 1, 2019, and December 31, 2022.

JAMA Network Open | Public Health
Disparities in Patient Portal Use Among Adults With Chronic Conditions

Sociodemographic and Psychosocial Characteristics
The C3 study collected self-reported information on patient psychosocial characteristics, COVID-19related beliefs and actions, health and lifestyle behaviors, health services use, and mental and physical health (eTable 2 in Supplement 1).Depression and anxiety were measured using the respective Patient Reported Outcomes Measurement Information Service short-form instruments, which are validated and normed among the general US population. 23For each scale, a raw score was calculated and transformed into corresponding T-scores and categorized using the following severity thresholds: none (T-score, <55), mild (T-score, 55.0-59.9),moderate (T-score, 60.0-69.9),and severe (T-score, Ն70).For this analysis, we collapsed depression and anxiety measures into 3 severity groups: none (T-score, <55), mild (T-score, 55.0-59.9),and moderate or severe (T-score, Ն60.0).
All parent studies had uniform collection of patient information via interview, including demographic characteristics (ie, age, sex, self-reported race and ethnicity), SES (ie, household income, number in household, educational attainment, employment status, and health insurance), self-reported chronic conditions, and a 1-item, self-reported overall health question (assessed as excellent, very good, good, fair, or poor).Race and ethnicity were categorized as Hispanic or Latinx, non-Hispanic Black, non-Hispanic White, and other race (eg, Asian, Native American or Alaskan Native, and self-reported other race).[26][27]

Portal Use and Activity
Number of days of portal login by year was recorded for all study participants by the EDW.The following portal activities are reported in this study: electronic check-ins, requesting appointments, cancelling appointments, confirming appointments, viewing clinical notes, viewing after-visit summaries, downloading after-visit summaries, checking test results, viewing scans, viewing documents, and patient-clinician messaging.All portal activities were reported by frequency by year (2019-2022).

Statistical Analysis
Statistical analysis was conducted using RStudio software version 4.3.0(R Project for Statistical Computing) and Stata/SE software version 18 (StataCorp).Descriptive statistics were conducted on all patient variables.Covariates that were significantly associated in the bivariate analyses were included in the regression model.As the primary outcome, portal login activity, was continuous and nonnormally distributed, we applied a generalized estimating equation with negative binomial regression to model mean change in yearly portal login activity during 2019 through 2022, adjusting for sociodemographic characteristics and year as independent variables.We implemented an autoregressive correlation structure, as we assumed correlations between portal usage are highest between adjacent time points.We used 2019 as our baseline value to compare changes in portal use over time.Incidence rate ratios (IRRs) and estimated probability were reported, with significance set at 2-sided P < .05.To determine whether portal activity differed by year across certain sociodemographic characteristics, interaction terms between years and significant variables identified in our initial multivariate model (ie, race, sex, age, multimorbidity [ie, Ն3 chronic conditions], and health literacy) were tested separately.We adjusted these interaction models by race, sex, age, multimorbidity, health literacy, depression, anxiety, and tangible support.
We sought to better understand outliers and conducted an exploratory analysis of frequent portal users, given extreme usage among select participants between 2019 and 2022.We created a new variable of mean portal use between 2020 and 2022 for each participant and reviewed the distribution of mean portal use.We created a dichotomous variable for high portal user (yes or no)

Results
Among 718 C3 study participants, 536 (74.7%) had data on portal use (ie, were ever users of the portal) and were included in the study (eTable 3 in Supplement 1 All other activities, such as scheduling (ie, electronic check-ins or requesting, cancelling, and confirming appointments) and messaging, were low and had medians at or close to 0 across each year.

Associations Between Portal Login Activity and Sociodemographic Characteristics
Multivariable results for portal use are reported in

Discussion
This cohort study highlights longitudinal changes in disparities in patient portal use by key sociodemographic characteristics.Before the COVID-19 pandemic (2019), patients who were female, were older, had fewer comorbidities, or had lower health literacy had significantly fewer portal logins.
While disparities associated with sex and age were reduced as the pandemic progressed, disparities by multimorbidity remained and disparities by health literacy were exacerbated, highlighting that populations with pre-existing risk factors, including those with low health literacy, may continue to be left behind in the shift toward digital health.
[37] Our findings also suggest that, compared with younger adults and male patients, older adults and female patients were less likely to use patient portals before the pandemic but had marked increases during the pandemic.9][40] Given health disparities in the C3 parent studies, it is possible that C3 participants include a more health care-seeking, proactive male patient population than typical prior studies or a general sample.Another possibility might be caregiver messaging on the patient's behalf by proxy, which is observed among adults managing chronic conditions. 41,42The data for this analysis do not delineate whether portal activity was driven by patients vs caregivers on the patient's behalf.
Although portal logins increased overall during the pandemic, specific portal activity was limited to reviewing test and laboratory results rather than scheduling, messaging, or viewing other documents, similar to results found in a 2019 study on patient portal use and activity. 43Our analyses found that outliers in portal logins (ie, superusers) were more likely to have multimorbidity and mild anxiety.
This study is novel in its analysis of how disparities in the level of portal use evolve over time, providing a unique longitudinal analysis for evaluating sociodemographic differences in portal use

Limitations
This study had several limitations.Our analysis was reliant on the EDW database and evaluated specific portal activities; therefore, it is unclear whether we fully capture a patient's portal activity.
Our analysis may capture accidental logins or (also known as phantom logins) in which a patient logs in and passively views their portal dashboard or notification or message alerts (ie, logging in with no further portal activity completed).
This analysis used a process-based outcome (ie, frequency of annual portal logins) and did not examine whether portal use was associated with better perceived health care quality or improved health outcomes.Logging in to the portal may not equate to meaningful use of the portal or meaningful engagement with health care practitioners in the system. 45We were not able to evaluate the association of portal use with clinical decision-making or clinician factors that may be associated with use.Thus, we cannot infer that increased portal logins are a positive outcome.We also did not investigate caregiver or clinician perspectives on portal use.Additionally, analysis did not include portal nonusers, who might be more likely to have barriers in accessing or using healthcare and thereby have poorer health outcomes and face greater health disparities.
Furthermore, generalizability of findings is limited, as the C3 study surveyed patients with underlying health conditions actively enrolled at a single health care system located in 1 large US city.

JAMA Network Open | Public Health
, after-visit summary.

Figure 2 .
Figure 2. Number of Annual Portal Logins by Sex, Age, Number of Chronic Conditions, and Health Literacy Level

during 4
years before, during, and the most restrictive phase of the COVID-19 pandemic among a diverse sample of patients.As telemedicine and digital health continue to evolve, it is important to consider how future directions for health care organizations might address digital health disparities and meaningful use.It may be important to consider how attitudes and perceptions of patient portals might hinder (eg, concerns about privacy and data security) or facilitate portal adoption and meaningful use.44 Disparities in Patient Portal Use Among Adults With Chronic Conditions JAMA Network Open.2024;7(2):e240680.doi:10.1001/jamanetworkopen.2024.0680(Reprinted) February 29, 2024 3/13 Downloaded from jamanetwork.comby guest on 03/30/2024 using the upper quintile and determined a mean of 69.66 logins to be the minimum threshold to categorize high use (ie, 20% of population that are the highest users).A logistic regression model was fit, adjusting for any covariates associated with high portal use in univariable analysis at P < .05,and unadjusted odds ratios (ORs) and adjusted ORs (AORs) are reported.Data were analyzed between March and June 2023.
Portal Use and Activity Over TimeThe distributions of portal logins by year are shown in Figure1.Frequency of portal activity across 2019 to 2022 are reported in Table2.With respect to the median number of days of portal logins, logins increased from a median (IQR) of 16 (0 to 45.3) days in 2019 to 31 (2 to 52) days in 2020.The median (IQR) number of days of portal logins was 31.5 (6 to 65.3) days in 2021 and 31 (4.8 to 65) days in 2022.The most frequent portal activity was checking laboratory or test results, with a median (IQR) of 4 (0 to 13) logins in 2019 and 2020, 6 (0 to 14) logins in 2021, and 7 (0 to 17) logins in 2022.

Table 2 .
Frequency of Portal Activity From 2019 to 2022
a Includes Asian, Native American or Alaskan Native, and self-reported other race.JAMA Network Open | Public HealthJAMA Network Open.2024;7(2):e240680.doi:10.1001/jamanetworkopen.2024.0680(Reprinted) February 29, 2024 7/13 Downloaded from jamanetwork.comby guest on 03/30/2024 literacy logged in Disparities in Patient Portal Use Among Adults With Chronic Conditions Follow-up investigations are currently underway to examine disparities among C3 participants who sought care in other community health care locations.In this cohort study, we include a novel analysis of sociodemographic disparities in portal use over 4 time points before and during the COVID-19 pandemic.Additional research may be warranted to fully understand effective interventions at sitewide and systemwide levels to bridge the gap in portal use and to minimize vulnerable populations continuing to be left behind.Although this study was focused on sociodemographic disparities of portal activity among patients with an active portal account, it is important to consider existing and shifting disparities among individuals who have never used the portal and addressing possible barriers in portal adoption.Furthermore, as telehealth and digital health tools continue to be an integral part of health care systems, future research would benefit from evaluating and optimizing digital literacy challenges as a potential barrier to portal adoption and use, as well as optimizing access to reliable internet or broadband services, particularly for communities that have historically had poor digital access due to limitations in neighborhood infrastructure.42.Raj M, Iott B. Evaluation of family caregivers' use of their adult care recipient's patient portal from the 2019 Health Information National Trends Survey: secondary analysis.JMIR Aging.2021;4(4):e29074.doi:10.2196/29074 43.Wong JIS, Steitz BD, Rosenbloom ST.Characterizing the impact of health literacy, computer ability, patient demographics, and portal usage on patient satisfaction with a patient portal.JAMIA Open.2019;2(4):456-464.doi:10.1093/jamiaopen/ooz05844.Carini E, Villani L, Pezzullo AM, et al.The impact of digital patient portals on health outcomes, system efficiency, and patient attitudes: updated systematic literature review.J Med Internet Res.2021;23(9):e26189.doi: 10.2196/26189 45.Beal LL, Kolman JM, Jones SL, Khleif A, Menser T. Quantifying patient portal use: systematic review of utilization metrics.J Med Internet Res.2021;23(2):e23493.doi:10.2196/23493Comparison of Patients Without vs Without Patient Portal Accounts eTable 4. Bivariate and Multivariable Analyses of Factors Associated With High Portal Utilization JAMA Network Open.2024;7(2):e240680.doi:10.1001/jamanetworkopen.2024.0680(Reprinted) February 29, 2024 9/13 Downloaded from jamanetwork.comby guest on 03/30/2024