Disparities in Patient Portal Engagement Among Patients With Hypertension Treated in Primary Care

Key Points Question Are there disparities in engagement with patient portals among patients treated for hypertension in primary care? Findings This cohort study of 366 871 patients found that patients with hypertension who are Black or Hispanic, who are uninsured, whose preferred language is not English, and who smoke are less likely to engage with patient portals. Meaning These findings suggest that as health care systems use patient portals more frequently for hypertension-related interventions and programs, disadvantaged patient groups are being left behind, potentially exacerbating existing disparities in blood pressure control.

Findings This cohort study of 366 871 patients found that patients with hypertension who are Black or Hispanic, who are uninsured, whose preferred language is not English, and who smoke are less likely to engage with patient portals.

Introduction
[5] Health care systems and care teams are increasingly relying on patient portals to deliver interventions. 6Patients with hypertension can use the patient portal to communicate with the care team about medication titration questions, review their after-visit summary, view their laboratory results, and more.8][9] Patient portals are used as a platform by the care team to provide evidence-based educational material to the patient. 9Patient portals can be used to collect patientreported data, such as home blood pressure readings, to enhance disease management, which has been linked to improved blood pressure control and has been recommended in clinical practice guidelines. 10,11More recently, patient portals have been used to invite patients to participate in clinical trials, giving these patients the potential advantage of better care that is associated with clinical trial participation. 124][15][16] It is unclear how engaged disadvantaged patients are with the patient portal.Given the importance of ongoing asynchronous communications for patients with hypertension, it is crucial to characterize patient portal use to understand its adoption and reach as a platform for hypertension management.It is also important to monitor and evaluate the potential digital divide among disadvantaged patient populations who share the greatest burden of hypertension and its clinical consequences. 14With data from a large and diverse Midwestern health care system, we aim to characterize patient portal access, using multiple definitions, and the use of hypertension-specific functionalities, including messaging the care team and sharing home blood pressure readings.

Methods
This retrospective cohort study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. 17The study was approved by the Advocate Aurora Health Institutional Review Board, which waived informed consent for this retrospective, deidentified data set because of minimal risk to patients.

Study Setting, Data Sources, and Population
We used data from Advocate Health-Midwest, a large and diverse health care system serving Wisconsin's urban and rural populations and Illinois' metropolitan population.The patient portal was introduced to the system in 2012.All patients are automatically offered an activation code at no cost.defined as the first visit in 2021.Each patient was followed up for 1 year to evaluate patient portal activity.Clinical characteristics included depression and diabetes identified using ICD-10 codes.

Main Independent Variables
Uncontrolled baseline blood pressure was defined as a systolic blood pressure of 140 mm Hg or higher or a diastolic blood pressure of 90 mm Hg or higher at the baseline visit.Blood pressurelowering medications were grouped by class and categorized into at least 2 medications currently on the active medication list vs fewer than 2 medications. 2The number of primary care physician (PCP) visits was defined as clinical encounters at internal medicine, family practice, or geriatrics departments and was categorized as 1 to 2 visits vs 3 or more, per clinical practice guideline recommendations. 19

Outcomes
Five indicators for patient portal engagement were created.Access was defined by the patient having logged into their patient portal account at least once during the 1-year follow-up.Frequency of access was also presented as medians (IQRs).The number of patient logins was winsorized at the 95th percentile. 20,21Accessed around PCP visit times was defined as logging in up to 7 days before or after at least 50.0% of the patient's PCP encounters during the 1-year study period.Frequent access was defined as having at least 28 logins during the 1-year study period.This cutoff was selected based on the median number of logins among patients who accessed the portal at least once.Messaging was defined as sending more than 1 free-text secure message to the care team during the 1-year study period via the patient portal messaging functionality.This function has been reported in the literature as a proxy for more active patient engagement given the required involvement in sending messages. 22,23Sharing home blood pressure readings was evaluated among the subgroup of patients who received an self-measured blood pressure order from their care team and was defined as sharing at least 1 reading.This functionality allows the care team to create an order for the patient to manually enter discrete systolic and diastolic readings from their home to be shared with the care team.

Statistical Analysis
Descriptive statistics are reported as numbers (percentages) for categorical or ordinal variables and means (SDs) or medians (IQRs) for continuous variables.Five multivariable logistic regression models were created to evaluate associations between patient characteristics and each of the 5 patient portal engagement indicators.All models included the same independent variables (age, sex, race and ethnicity, insurance, primary language, diabetes, tobacco use, depression, uncontrolled baseline blood pressure, blood pressure-lowering medications, and number of PCP visits) and presented as adjusted odds ratios (ORs) and 95% CIs.The analysis used ORs instead of relative risks even though the outcomes are not expected to be rare; ORs may slightly overestimate relative risks but are commonly reported in the literature, and logistic regression models may be more familiar to readers. 24Given the low proportions of missing data from the EHRs used in this study, no imputation of missing data was performed.Instead, patients missing any of the baseline visit variables included in the regression models were excluded from the cohort (93 patients [<0.1%]).A 2-tailed P < .05 was considered statistically significant.Analyses were completed using SAS software, version 9.4 (SAS Institute Inc).

Description of Study Population
A total of 366 871 patients (mean [SD], 63.5 [12.6] years) with hypertension were included in the analysis (eAppendix in Supplement 1).Of these patients, 52.8% were female and 47.2% were male, 3.4% were Asian, 7.8% were Hispanic, 19.7% were non-Hispanic Black, 66.9% were non-Hispanic White, and 2.3% were of other races and ethnicities.Blood pressure during the baseline visit was controlled among 83.3% of patients, 28.1% were prescribed at least 1 blood pressure-lowering medication, and 59.1% had 3 or more PCP visits during the 1-year study period (Table 1).

Overall Engagement With the Patient Portal
The median (IQR) number of patient portal logins was 28.0 (10.0-65.0)among active users.On average, patients logged in a mean (SD) of 46.7 (49.8) times during the year.Only 553 patients (<1.0%) received an order for home blood pressure monitoring; among them, 48 (8.7%) engaged in sharing at least 1 reading with the care team (Figure 2).

Associations With Sociodemographic Characteristics
Adjusted analyses indicate clear differences in the sociodemographic characteristics of patients with hypertension who engaged with the patient portal (  2).Associations between patient sociodemographic characteristics and receiving an order for and engaging with home blood pressure monitoring showed similar trends with differences in terms of age, race and ethnicity, insurance, preferred language, and tobacco use, although not all these associations were statistically significant (Table 3).

JAMA Network Open | Cardiology
Disparities in Patient Portal Use Among Patients With Hypertension

Associations With Clinical Characteristics
In the adjusted analysis, the number of PCP visits that the patient attended during the 1-year follow-up showed associations across all measures of engagement, with greater odds among patients who had 3 or more visits for any access (OR, 1.59; 95% CI,   Home blood pressure readings (shared ≥1 reading) a Among all patients (n = 366 871) Among patients who accessed at least once (n = 258 539) PCP indicates primary care physician.a Limited to 553 patients who received an order from their care team to share home blood pressure (BP) readings via the patient portal.Abbreviations: BP, blood pressure; NA, not applicable; OR, odds ratio; PCP, primary care physician.

JAMA Network Open | Cardiology
a Adjusted for age, sex, race and ethnicity, insurance, primary language, diabetes, tobacco use, depression, uncontrolled baseline BP, BP-lowering medications, and number of PCP visits.
b Other includes American Indian or Alaska Native, Native Hawaiian, other Pacific Islander, or unknown.
c All patients in the self-pay insurance category accessed the patient portal at least once.

Discussion
In this cohort of more than 300 000 primary care patients with hypertension, 7 of 10 patients accessed the patient portal at least once between 2021 and 2022 but only 28.9% engaged in messages and 8.7% shared home blood pressure readings.The literature on quantifying patient portal engagement shows variable results depending on the patient population, how engagement is Given the lack of a unified definition for engagement, we included several definitions beyond ever accessing the portal or frequency of access.Our results showed consistent findings of disparities in engagement regardless of the engagement definition used and after adjusting for several demographic and clinical characteristics.Although an access code for patient portal use is generated automatically for all patients attending this health care system, patients belonging to racial and ethnic minority groups were less likely to access the patient portal than non-Hispanic White patients.
Patients whose preferred language is not English were also significantly less likely to engage with the patient portal, which is similar to findings in the study by Casillas et al. 27 Although 100% of self-pay patients accessed the patient least once, engagement was still low in this subgroup, which aligns with lack of engagement in other disadvantaged subgroups evaluated in this analysis.An explanation may be that these patients are using the self-pay function in the patient portal.Smoking, which is a social determinant of health-associated behavior, was also associated with less engagement with patient portals.
Our results extend the existing literature on the potential digital divide in patient portal engagement and highlight major disparities among patients treated for hypertension in primary care. 28,29The digital divide is the uneven distribution of access to, use of, and impact of information technologies between demographics and regions. 30Low patient portal use in vulnerable populations may result in intervention-generated inequity, which is a situation whereby an intervention inadvertently worsens existing health disparities instead of improving them. 31,32Research studies, interventions, and quality improvement initiatives in hypertension that are delivered via patient portals should be mindful of the disparities in engaging with patient portals.Efforts should be made and evaluated to reach all patients regardless of their race and ethnicity, insurance status, or preferred language.This is especially true for patients with hypertension because these subpopulations share the burden of clinical consequences. 7r study did not explore the underlying reasons behind the digital divide, which may include education levels, low socioeconomic status, and lack of internet access. 25,33These characteristics likely mediate some of the associations observed in our analysis.In addition, physician and health care system characteristics also likely influence patient portal engagement. 30Self-reported survey data, such as the Health Information National Trends Study, found similar disparities in patient portal engagement.In that survey, patients who were non-Hispanic White were more likely to report being offered access to the patient portal (76.7%) compared with patients who are non-Hispanic Black (66.8%).This disparity highlights the importance of the care team actively discussing and offering the patient portal to patients and offering it equally regardless of race or ethnicity. 34In our analysis, the number of patients who received a self-monitored blood pressure order from the care team was too small to draw conclusions; however, data suggest that patients who are Hispanic or Latino and smokers were less likely to receive an order for self-monitored blood pressure.Future work should incorporate implementation science approaches and directly address the key role of physicians and staff in promoting more equitable patient portal use. 35r study also sheds light on clinical characteristics related to hypertension that may be associated with patient portal engagement.More PCP visits were associated with greater patient portal engagement.This finding suggests that patients engaged with the patient portals are more engaged with their PCPs and have more in-clinic communication.Efforts should be made to engage patients with less than the guideline-recommended number of PCP visits (ie, <3 visits per year) with the patient portal as an alternative (backup) to in-person communications with the PCP. 25,30,33rther research is needed to understand how patient portal engagement is associated with reaching blood pressure control.hypertension showed an association between patient portal access (evaluated subjectively through a survey) and blood pressure control. 36However, the association disappeared after adjusting for potential confounders. 36More recently, a study evaluating associations between patient portal access and glycated hemoglobin control among patients with diabetes showed a clear link between accessing the portal and better control. 4More research exploring these associations among patients with hypertension is warranted.

Limitations
A few limitations should be noted when interpreting our results.First, our data are 1 health care system, and patients may seek care and use patient portals elsewhere.Second, unlike other studies that report on all the common functionalities of the patient portal, we focus our results on 2 functionalities that are relevant to patients with hypertension.Third, many factors may be associated with low patient portal engagement (eg, education level, income, and internet access); however, our analysis was limited to factors found within the EHR.Fourth, we were only able to evaluate patients' engagement in home blood pressure monitoring via the patient portal-specific functionality.Patients may share their home blood pressure readings via the messaging functionality; however, this rate has been reported to be very low at 1.2% of patients only. 10Patients may also engage in home blood pressure monitoring outside patient portals, such as using automated blood pressure cuffs that transmit readings directly to the care team.However, few of these initiatives have been implemented in clinical settings given barriers related to the cost of the device to the patient and barriers to connecting these devices to the EHR.

Conclusions
In this cohort study of patients with hypertension, two-thirds had accessed the patient portal at least once, with a median number of logins of 28 per year.Despite the observed high rates of engagement, clear disparities were observed in terms of race and ethnicity, insurance status, preferred language, and tobacco use.To overcome the digital divide and prevent further disparities in hypertension outcomes, it is important to further evaluate patient portal adoption among these subpopulations and identify ways of increasing their engagement.
These findings suggest that as health care systems use patient portals more frequently for hypertensionrelated interventions and programs, disadvantaged patient groups are being left behind, potentially exacerbating existing disparities in blood pressure control.
the patient portal since it was first introduced, and both patients and clinicians can initiate messages.The home blood pressure monitoring functionality was introduced in 2018.The care team must create and share a home blood pressure monitoring order with the patient before the patient can use the functionality to manually enter readings.The analytic cohort included adult (aged Ն18 years) patients who have a documented hypertension diagnosis (using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] code I10) and at least 1 visit in 2021 to 1 of 312 primary care sites (family medicine, internal medicine, or geriatrics) and at least 1 additional primary care visit JAMA Network Open | Cardiology Disparities in Patient Portal Use Among Patients With Hypertension JAMA Network Open.2024;7(5):e2411649.doi:10.1001/jamanetworkopen.2024.11649(Reprinted) May 15, 2024 2/13 Downloaded from jamanetwork.comby guest on 05/22/2024 before 2021.Sociodemographic and clinical data were extracted from the EHR at the baseline visit,

Figure 1
presents the proportion of patient portal engagement by patient sociodemographic characteristics among all patients.During the 1-year study, 70.5% of patients accessed the patient portal at least once, 60.2% accessed it around PCP visit time, 35.7% accessed frequently, and 28.9% engaged in messaging.

Figure 1 .
Figure 1.Patient Portal Engagement by Characteristics of the 366 871 Study Patients

Figure 2 .
Figure 2. Proportion of Patients Who Engaged With the Patient Portal 100

Table 3 .
26sociations Between Patient Characteristics and Home BP Reading Functionalities in the Patient Portal a None of the patients under other race or ethnicity, self-pay insurance, and non-English speakers shared BP readings with the care team, and therefore an OR could not be calculated.wereactiveusers,defined based on accessing specific functionalities that require more active patient involvement.Sun et al26reported that 32.9% of patients with diabetes attending outpatient clinics used patient portals. e An analysis from 2010 among patients with newly diagnosed JAMA Network Open | Cardiology Disparities in Patient Portal Use Among Patients With Hypertension JAMA Network Open.2024;7(5):e2411649.doi:10.1001/jamanetworkopen.2024.11649(Reprinted) May 15, 2024 9/13 Downloaded from jamanetwork.comby guest on 05/22/2024 Manard W, Scherrer JF, Salas J, Schneider FD.Patient portal use and blood pressure control in newly diagnosed hypertension.J Am Board Fam Med.2016;29(4):452-459.doi:10.3122/jabfm.2016.04.160008 36.