Reducing Ethnic and racial Disparities by improving Undertreatment, Control, and Engagement in Blood Pressure management with health information technology (REDUCE-BP) hybrid effectiveness-implementation pragmatic trial: Rationale and design

Abstract

Background:

While racial/ethnic disparities in blood pressure control are documented, few interventions have successfully reduced these gaps. Under-prescribing, lack of treatment intensification, and suboptimal follow-up care are thought to be central contributors. Electronic health record (EHR) tools may help address these barriers and may be enhanced with behavioral science techniques.

Objective:

To evaluate the impact of a multicomponent behaviorally-informed EHR-based intervention on blood pressure control.

Trial Design:

Reducing Ethnic and racial Disparities by improving Undertreatment, Control, and Engagement in Blood Pressure management with health information technology (REDUCE-BP) (NCT05030467) is a two-arm cluster-randomized hybrid type 1 pragmatic trial in a large multi-ethnic healthcare system. Twenty-four clinics (>350 primary care providers [PCPs] and >10,000 eligible patients) are assigned to either multi-component EHR-based intervention or usual care. Intervention clinic PCPs will receive several EHR tools designed to reduce disparities delivered at different points, including a: 1) dashboard of all patients visible upon logging on to the EHR displaying blood pressure control by race/ethnicity compared to their peers and 2) set of tools in an individual patient’s chart containing decision support to encourage treatment intensification, ordering home blood pressure measurement, interventions to address health-related social needs, default text for note documentation, and enhanced patient education materials. The primary outcome is patient-level change in systolic blood pressure over 12 months between arms; secondary outcomes include changes in disparities and other clinical outcomes.

Conclusion:

REDUCE-BP will provide important insights into whether an EHR-based intervention designed using behavioral science can improve hypertension control and reduce disparities.

INTRODUCTION

Racial disparities in hypertension control have been recognized for decades.^1–5 Despite modest improvements in treatment initiation^6,7, blood pressure control among Black and Hispanic/Latino adults remains substantially lower than non-Hispanic whites.^7–11 Many factors contribute to these persistent care gaps. Racial differences in treatment intensification alone contribute to more than 20% of observed racial/ethnic variation in blood pressure control.² Follow-up care is challenging for providers and health systems, in part because of inability to afford in-home blood pressure monitoring cuffs as well as differential recommendations by providers about self-monitoring.¹² Further, social determinants of health, such as financial resource strain, are often more prevalent among Black and Hispanic/Latino patients and socially disadvantaged individuals and can compound issues of health system access.^13,14

A strategy to address some of these challenges is with the use of electronic health record (EHR)-embedded tools.^2,15,16 Their widespread use by providers already in their clinical workflow enhances the potential for scalability.^15,17 Many EHRs systems contain a range of possible clinical decision support tools such as alerts, dashboards, reminders, and defaults.^18,19 Despite evidence supporting the ability of these interventions to improve health care quality, in many cases they have only been modestly effective, attributed to issues with their timing within the clinical workflow, the salience of the information, and alert fatigue.^18,20–22 To date, very few trials have evaluated whether EHR-based interventions can reduce racial/ethnic disparities.^23,24

Incorporating behavioral science principles into EHR tools could improve their performance and ability to improve health equity.^{18,21,25–28} For example, EHR alerts are often delivered when clinicians are actively using the medical record to provide care; while this outreach occurs in usual workflow, clinicians may be less likely to respond to new information at this point.¹⁸ Priming physicians by providing information outside of visits and using behavioral science techniques to increase salience, such as showing differences in blood pressure control by race/ethnicity compared to one’s peers (i.e., a behavioral science principle called social norming)^18,25, may encourage clinicians to proactively engage with their patients. Adjusting the timing and presentation of tools also leverages a form of choice architecture.^18,29 Using priming and enhancing salience may also increase receptivity to tools that are subsequently delivered in a clinical encounter.^18,29,30 However, despite the promise of these approaches, relatively few interventions have specifically incorporated behavioral science principles into EHR tools and evaluated them in randomized trials, especially in hypertension.²²

Further, in the specific context of racial/ethnic disparities in blood pressure control, existing EHR interventions have often addressed only one contributor to poor control (e.g., treatment intensification);³¹ their scope could be expanded to encourage screening for social determinants of health, simplifying access to appropriate community resources, and facilitating routine clinical follow-up in tandem.³²

Accordingly, we launched Reducing Ethnic and racial Disparities by improving Undertreatment, Control, and Engagement in Blood Pressure management with health information technology (REDUCE-BP). The overall objective was to evaluate the impact of a multi-component behaviorally-informed EHR intervention on blood pressure control and racial/ethnic disparities in patients with hypertension and measure implementation outcomes to facilitate potential future dissemination.

METHODS/DESIGN

Overall study design

REDUCE-BP is a pragmatic, intention-to-treat, cluster-randomized trial that uses a hybrid type 1 effectiveness-implementation research design. The primary purpose of the trial is to test effectiveness of a multi-component EHR intervention while collecting information on implementation of the intervention to facilitate future, wider-scale application.³³ The trial was designed based on Pragmatic Explanatory Continuum Indicator Summary (PRECIS-2) trial guidance and reported using SPIRIT reporting guidelines (protocol in Appendix 1). The study was approved by the institutional review board of Mass General Brigham, overseen by a Data and Safety Monitoring Board, and registered on clinicaltrials.gov (NCT05030467). Informed consent was waived for all subjects on the basis that the intervention involved testing decision support directly for providers using information that would be available to them within the EHR system and because obtaining informed consent could reduce participation from underrepresented populations.

This research was supported by the NIH National Institute on Minority Health and Health Disparities (NIMHD) (R01MD014874) to BWH (Choudhry/Lauffenburger). The authors are solely responsible for the design and conduct of this study and drafting and editing of the paper and its final contents.

Study setting and randomization

The trial is being conducted at 24 outpatient primary care practices of Advocate Aurora Health (AAH), a large integrated delivery network in Wisconsin and Illinois. AAH serves a highly racially/ethnically diverse population of more than 1 million patients. AAH has a fully functional EHR system, Epic^®, that supports computerized ordering of medication and other decision support tools.

Participating clinics were randomly assigned to a multicomponent EHR intervention or usual care (i.e., no intervention). We used clinic-level covariate-constrained (CCR) randomization to limit the potential for contamination while reducing clinic-level imbalances between treatment arms.^34,35 In CCR, baseline factors that may be potential confounders are selected, and simulation is used to generate possible randomization schemes, with covariate balance assessed for each possible scheme. From those that fall below maximum allowable differences between the arms, a single randomization scheme was selected. As baseline covariates in the CCR, we incorporated geographic region (i.e., patient service area), number of potentially-eligible patients, distribution of race/ethnicity of patients in the clinic, number of primary care providers (PCPs) and departments, zip-code level social deprivation index³⁶, and whether it was a clinic designated as a “priority clinic” (i.e., hypertension is a key quality improvement focus) by AAH. Using the CCR, we randomized clinics in a 1:1 ratio to intervention and usual care arms.

Participants

Provider-subjects are eligible for inclusion in the study based on adult primary care specialty (internal medicine, family medicine, or geriatrics) and employment by AAH in one of the randomized 24 clinics. Patients are eligible for inclusion if they meet the following criteria: 1) ≥1 visit with an included AAH PCP in the past 24 months, 2) 18–85 years of age, 3) diagnosis of hypertension on their EHR problem list or having ≥2 office, telehealth, or telephone encounters coded with a diagnosis of hypertension occurring on different dates with at least one encounter occurring in the prior 24 months, 4) latest outpatient/ambulatory (excluding urgent care) systolic blood pressure (SBP) ≥140 or diastolic blood pressure (DBP) ≥90 in the last 12 months. We selected the patients’ lower blood pressure for inclusion if more than one was taken at the same time. These criteria were chosen as they match Healthcare Effectiveness Data and Information Set (HEDIS) criteria used for provider performance at AAH and are also highly pragmatic to identify using routinely-collected data.³⁷

Interventions

Our intervention design was informed by 30 qualitative interviews conducted with patients and PCPs at AAH to understand perspectives on hypertension management and opportunities to improve EHR tools, other stakeholder meetings at AAH, and prior peer-reviewed literature.^16,18,38 Using this information, the EHR tools were co-designed using behavioral principles with a multidisciplinary investigative team, with an emphasis on reducing the use of interruptive alerts and in-basket messages to limit alert fatigue.^39,40 We also incorporated several behavioral science principles for our intervention, including salience, social norming, choice architecture, and framing, based on their preliminary effectiveness in other settings and ability to be adapted to EHRs.^18,29,41,42

The multicomponent intervention consists of several EHR tools delivered at different time points during a patient’s care episode (Figure 1). The overall design leverages the concept that individual decision-making by physicians and their ability to take action is influenced by pressures like time or stress.⁴³ Specifically, before or after an office visit, providers generally have more time to consider and address patient care issues, but patients are not physically present so making therapeutic changes could be challenging. In contrast, patient interaction occurs during potentially hurried visits where making rational decisions is more difficult.

FIGURE 1.

Key components of the intervention

The tools are primarily for PCPs but include some aspects for other parts of the care team. The core PCP-facing tools included two main components designed to address these different decision-making points between and during visits: 1) a dashboard of all patients immediately visible upon logging in to the EHR displaying blood pressure control by race/ethnicity in comparison to that among their peers and 2) a set of tools in an individual patient’s chart (in the “Plan” tab, which is the default view on chart opening) containing decision support to encourage treatment intensification, the prescribing of home blood pressure measurement, addressing social determinants of health, default text to include in encounter note documentation, and enhanced patient education materials. Each is described in further detail below.

Between visits: Dashboard

Within the EHR system, we assembled a patient registry consisting of a PCP’s patients who had ≥1 visit in last 24 months, were 18–85 years of age, and had a diagnosis of hypertension (i.e., the same inclusion criteria used for trial eligibility, except for blood pressure control). Using this registry, we created a dashboard that can be automatically displayed when a PCP logs into the EHR system. To be visible at EHR login, intervention PCPs were instructed to add the dashboard to their login screen, as they do with other dashboards at AAH.

The dashboard presents the proportion of patients whose blood pressures are poorly controlled (as defined above) overall and by race/ethnicity for the largest 3 groups at AAH (i.e., Hispanic/Latino, Black and White patients) (Figure 2). Graphics within the dashboard are presented for the PCP’s patient panel and also compared with their department (e.g., other providers at their clinic). This type of “social norming” has successfully changed provider behavior for antibiotic and antihypertensive prescribing and mammography screenings.^42,44,45 Providers can select dashboard elements to obtain detailed individual-level information about patients, including longitudinal blood pressure values, medication use, upcoming appointments, and key social determinants of health related to hypertension (i.e., food insecurity, financial resource strain, and transportation needs) (Appendix 2- Figure 1). To highlight patients in greatest need of blood pressure management, patients with poor blood pressure control are listed first in the individual-level drilldown. Within categories of control, patients are then sorted by upcoming appointment (with those with no scheduled appointment listed first) and then finally by their date of last visit (i.e., oldest visit to newest). The dashboard also provides information about dates of patient visits and has functionality to send patients and clinic staff messages for follow-up and monitoring.

FIGURE 2.

Electronic health record embedded hypertension dashboard

This same dashboard, at the department level, is also available for nurse managers at the intervention clinics when they log into the EHR system to support hypertension care delivery.

During visit: Clinical decision support

Medical assistants: Social determinants of health alert during patient rooming

To facilitate the capture of social determinants of health data required for PCP decision support, an Epic alert (called a Best Practice Advisory) fires for medical assistants if social determinants of health have not been reviewed for the patient in the prior 90 days. The alert appears when they first room eligible patients and requests that the medical assistant complete a screening for 11 social determinants of health, including food insecurity, financial resource strain, and transportation needs; these social determinants of health were determined at the AAH system level. For any identified unmet social needs, the alert links to community resources that can be provided to the patient during or after their PCP visit.

Primary care providers: Default landing page containing decision support, order set, note documentation, and patient materials

When PCPs open a patient encounter for patients whose latest outpatient/ambulatory blood pressures are elevated (i.e., SBP≥140 or DBP≥90), they will be brought, by default,⁴⁶ to the “Plan” tab in the patient’s chart (Figure 3). This contains an alert indicating the need for treatment intensification and, if applicable, to address any identified social determinants of health needs. In addition, the decision support displays hypertension-specific data about patient medication use and blood pressure values, the status of 11 social determinants of health, an order set prompting the prescribing of guideline recommended racially-concordant treatments^47,48, relevant lab tests, self-monitoring tools, a default to add hypertension-specific documentation to the progress note, and enhanced patient education materials. Recent home blood pressure readings are also presented to PCPs if available, which could be used to support decision-making.

FIGURE 3. Clinical decision support elements within electronic health record plan tab for primary care providers.

A. Alert Box; B. Longitudinal blood-pressure specific patient information; C. Social determinants of health wheel (wheel not shown); D. SmartSet order set

The alert indicating the need for treatment intensification or a social need to be addressed was designed using behavioral principles of salience.⁴⁹ It is displayed at the time of opening an encounter, as our prior qualitative interviews suggested that most providers begin this encounter the day of or day before a patient visit and therefore the information could assist both before and during a visit. Social determinant of health needs are displayed on a “wheel” that is already in use at AAH and highlights patients with identified unmet needs and links to community resources to address these social needs. The wheel is updated in real-time based on the social needs screening conducted by the medical assistant during patient rooming.

The Epic order set (known as a “SmartSet”) provides recommended medication additions or substitutions, recommends labs (e.g., basic metabolic panel or aldosterone and renal levels), streamlined ways to order a home blood pressure cuff (e.g., directly ordering cuffs from a list of validated devices available at AAH), blood pressure measurement and recording handouts, and follow-up visit scheduling for a nurse blood pressure check. These handouts are available in both English and Spanish and were drawn from enhanced materials from the American Heart Association⁵⁰. PCPs or support staff (i.e., clinic managers or nurses) can also send patients a flowsheet via the patient portal that allows for remote entry of blood pressure values. Based on clinical practice guidelines^37,51, the order set is tailored slightly based on race as recorded in the EHR. If the patient is a Black individual, the order set first recommends combination therapy, thiazides, or calcium channel blockers (Appendix 2- Figure 2). The order set also provides a 1-page handout (under “clinical information”) with updated evidence about special populations and emerging suggestions to test renin and aldosterone activity, particularly in Black patients.^3,37,48 Finally, the order set also defaults to appending a tailored hypertension note to their end of their progress note. This note template prompts PCPs to document pre-populated actions such as modifying medication, addressing social needs, or social work referrals.

Patients whose blood pressure was elevated in the prior encounter but is normal on the current visit will also have the Plan tab displayed to provide the PCP with access to the order set and other supporting resources. However, for these patients, no note documentation will be prompted unless the patient has an unaddressed social determinant of health.

Study procedures

Prior to study launch, PCPs and allied staff (e.g., clinic managers, nurses, and medical assistants) who assist with managing EHR encounters in the intervention arms received brief training on the dashboard and decision support tools, conducted by the site principal investigators and clinical leadership. In addition, PCPs were given access to an optional training video and EHR tip sheets. These activities occurred as part of AAH’s robust existing infrastructure to train providers about EHR tools and system updates; in other words, training for this trial did not introduce any significant co-intervention.

Providers in usual care clinics do not have access to the intervention tools.

Outcomes

Primary outcome

The clinical outcomes will be drawn from routinely-collected EHR data (Table 1).^52,53 The primary outcome is change in SBP among patients who met eligibility criteria for the decision support; we selected SBP as our primary outcome because SBP has a greater effect on cardiovascular clinical outcomes than DBP.^54,55 Similar to prior studies^52,53, we will only use readings from ambulatory clinics where blood pressure is routinely measured and managed (i.e., primary care, geriatrics, cardiology, nephrology, and endocrinology). The change will be calculated as the difference between the last reading prior to randomization to the last value prior to the end of the 12-month follow-up. If more than one blood pressure value is recorded on these days, we will use the lowest, as our prior qualitative interviews and design meetings for the trial suggested that clinicians often make decisions based on the lowest value measured that day; the decision support is also appearing based on the lowest value in a prior visit, rather than the mean. However, in sensitivity analyses, we will take a mean of the values recorded in the EHR for a given clinic day, rather than just the latest value.

TABLE 1.

Study Outcomes

Outcome	Measurement	Assessment
Primary	Change in systolic blood pressure	Change in systolic blood pressure from the last ambulatory care encounter before randomization through 12 months after, via values recorded in the EHR from a calibrated sphygmomanometer
Secondary	Change in racial/ethnic disparities	Change in the gap in systolic blood pressure between each of Black and Hispanic/Latino patients and non-Hispanic White patients from the last ambulatory care encounter before randomization through 12 months after, via values recorded in the EHR from a calibrated sphygmomanometer
Secondary	Change in diastolic blood pressure	Change in diastolic blood pressure from the last ambulatory care encounter before randomization through 12 months after, via values recorded in the EHR from a calibrated sphygmomanometer
Secondary	Percentage of patients with well-controlled blood pressure	Percentage of patients with SBP <140/80 mmHg or <130/80 mmHg for those with comorbid conditions defined by clinical guidelines, using values in the EHR from a calibrated sphygmomanometer
Secondary	Intensification of antihypertensive medications	Percentage of patients with intensification of medications in the follow-up period (defined by addition of therapy or increasing dose)
Secondary	Guideline-concordant medication prescribing	Percentage of patients with guideline-concordant medications ordered in the follow-up period

Abbreviations: EHR, electronic health record

To maximize generalizability and minimize impact on clinic workflow, we will use measurements as they are conducted during routine care, rather than having study-specific procedures. However, AAH has standard procedures for measuring blood pressure across clinics, and values are regularly captured using automated devices by medical assistants during rooming.

In secondary analyses of this primary outcome, we will measure 1) change in SBP across all patients who met eligibility criteria regardless of whether they had an in-person visit during follow-up, 2) change in SBP across all available follow-up time (i.e., until the end of the study, up to 24 months after the trial start date), and 3) use longitudinal modeling to include all, not just the last, blood pressure value during follow-up. In additional sensitivity analyses, we will include home blood pressure values recorded in the EHR.

Secondary outcomes

Secondary blood pressure outcomes include change in the gap in SBP between Black and Hispanic/Latino patients and non-Hispanic White patients, change in DBP from baseline to at least 12 months after randomization, and the percentage of patients with well-controlled blood pressure, defined separately using guideline metrics before trial launch and any changes to guidelines during the conduct of the trial. We plan to explore gaps in other race/ethnicities, but Black, non-Hispanic White, and Hispanic/Latino patients are the largest subgroups in these clinics at AAH. We will also measure the number of treatment intensifications and percentage of patients with guideline-concordant medications ordered during follow-up as secondary outcomes.

Finally, as part of the hybrid type 1 study design, we will also measure implementation outcomes informed by the Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM) framework.⁵⁶ These will include frequency of using the dashboard, clinical decision support tools, note documentation, completeness of social determinants of health wheels, the provision of resources to meet social determinants of health needs, number of referrals, number of lab tests ordered, sharing of patient education materials, and ordering of blood pressure cuffs.⁵⁶ We will also measure the number of in-person visits, telehealth visits, and follow-up blood pressure checks during the follow-up period. Other implementation outcomes, including usability, acceptability, and sustainability will be measured at the end of the trial, through clinic exit surveys and qualitative interviews with patients and PCPs and quantitative data from the EHR. We will also explore changes social determinants of health and their association with blood pressure control and disparities, including the relationship between improving social determinants of health and changes in blood pressure using multivariable linear regression and marginal structural model approaches that handle time-dependent exposures.

The trial fully began in May 2022 with follow-up of all trial subjects expected by Spring 2024.

Analytic plan

We will report the means and frequencies of pre-randomization variables separately for the intervention and the control arm; we will compare these using absolute standardized differences to evaluate balance between arms. We will then also evaluate the distribution of outcomes, notably the primary outcome of change in SBP.

All analyses will use intention-to-treat principles. We hypothesize that the intervention will lead to a greater change in SBP from baseline to follow-up between the intervention arm and usual care, with the null hypothesis being no difference. For this primary outcome, we will use generalized estimating equations (GEE) to adjust for the cluster design with an identity link function and normally-distributed errors. If >10% of data is missing (i.e., no ambulatory clinic measured SBP available during the 12-month follow-up), we will use multiple imputation for our study outcomes, as we have done in prior research; if ≤10%, we will use complete case analysis.⁵² In sensitivity analyses for patients with missing data, we will use the last blood pressure value carried forward. In secondary analyses, we will adjust for age, sex, race/ethnicity, and baseline levels of poor hypertension control. In secondary analyses, we will also use longitudinal modeling to account for multiple SBP values over follow-up.

For secondary outcomes, we hypothesize that the intervention will reduce the gaps in SBP between Black and Hispanic/Latino patients and White patients, lead to more patients with well-controlled blood pressures, greater change in DBP and higher rates of treatment intensification and guideline concordance. For these outcomes, we will use GEE with an identity link and normally-distributed errors for continuous variables and with a log-link function and Poisson-distributed errors for categorical variables. For outcomes that are non-normal, we will use GEE with a log-link function and Poisson-distributed errors. Investigators will be blinded to treatment allocation until after the analyses are complete.

We will conduct analyses of change in blood pressure from baseline to follow-up by racial/ethnic subgroup using GEE and test whether the intervention effectiveness differed by subgroup using interaction terms. Then, we will conduct a differences-in-differences analysis using GEE with an identity link and normally-distributed errors to verify that the magnitude of disparities by Black and Hispanic/Latino patients was reduced from baseline to follow-up compared with non-Hispanic White patients. We will also conduct other subgroup analyses by age, gender, and degree of baseline blood pressure control.

Sample size considerations

From baseline data from the 24 clinics, we expect that >350 PCPs and >10,000 patients will be eligible for inclusion in the analysis. Of these, we estimate that 40% of patients will be Black and 12% will be Hispanic/Latino individuals. The remainder will be non-Hispanic White (~40%), Asian or multi-racial/other race, which will provide sufficient diversity to evaluate changes in hypertension disparities.

We powered the study to detect a 5.0 mmHg mean change in SBP between the intervention and control arms from randomization through 12 months. Our assumptions included 24 clinics, a standard deviation (SD) of 13.5 based on baseline data, α=0.05, an intra-cluster correlation (ICC) of 0.1, and ≥6,000 patients would be included in the analysis (i.e., 3,000 patients per arm). With these assumptions, we would have >80% power to detect differences in our primary outcome. The 5.0 mmHg mean change was chosen as this is considered to be a clinically meaningful change and observed in prior interventions in hypertension.^51,57,58 This estimate also accounts for regression to the mean and assumptions derived from prior work and feasibility data.⁵²

Limitations

There are several limitations that should be acknowledged. First, patients may not have a blood pressure reading in the 12-month follow-up, which may lead to incomplete assessment of the primary outcome; however, based on prior work, we expect less than 10% missingness.^10,52 Second, while multi-component interventions are increasingly being recognized as necessary to overcome barriers to hypertension control, should the intervention be shown to be effective, we will not be able to explicitly identify which aspect was most effective in the primary analysis. Third, this intervention is not designed to increase the rates of diagnosing hypertension; however, data within AAH and nationwide indicate that greater racial/ethnic disparities exist in treatment and control of hypertension rather than underdiagnosis.⁵⁹ Fourth, while our trial was designed to be pragmatic and incorporate EHR tools typically available at most healthcare systems, the results may not fully generalize. The tools are also largely PCP-facing; their effectiveness could be limited with poor uptake, but we will evaluate this in the implementation evaluation. Fifth, blood pressures are captured from routine care, which may lead to some variation across clinics; however, our primary outcome is change in SBP by patient so we expect any variation to have a minimal impact. Sixth, while the intervention primarily focuses on enhancing treatment intensification, we are also unable to measure adherence to medication using the EHR data alone, which could reduce our ability to understand this contribution on blood pressure change. Finally, while social determinants of health are being collected and likely impact blood pressure management and control, we are not able to carefully assess the impact of addressing them.

Conclusion

The rising prevalence of poorly-controlled hypertension and racial and ethnic disparities in blood pressure control represent a major public health problem. The REDUCE-BP cluster randomized pragmatic trial is designed to leverage EHR tools with high potential for scalability to improve blood pressure and ultimately reduce racial/ethnic disparities in hypertension control. Most EHR decision support tools for providers have been designed with the goal of reducing provider liability or meeting performance targets for specific process measures.⁶⁰ Despite the promise of using rapid recent advances in behavioral economics and related behavioral sciences, relatively few interventions have incorporated these principles into EHR tools and evaluated them in randomized trials, particularly in hypertension.²² By contrast, we will not only develop and evaluate the impact of behaviorally-based EHR tools in the context of hypertension, we will also explicitly test dissemination and implementation by evaluating system-wide effects, including any unintended negative effects of EHR use on health disparity populations.^26,61

The trial will provide important insight into whether an EHR-based intervention can improve blood pressure control and reduce disparities. If successful, this model can be widely and efficiently implemented within AAH and other health care systems.