Hypertension is a common heart health condition in the United States (US) affecting more than 119 million adults.1 It has been linked to adverse health outcomes such as stroke, heart disease, and all-cause mortality.2–4 In 2021, hypertension contributed to more than 600 000 deaths in the US.5 However, research indicates greater burden of hypertension in males, young adults, and racial and ethnic minority populations in the US.6 Non-Hispanic Black/African American adults, in particular, experience the highest prevalence of hypertension.7,8 Increasing rates of hypertension are also documented in the Hispanic/Latino and Asian/Asian American communities.9 There is an urgent need to address the persistent disparities in hypertension observed across social and clinical demographics.
In this context, primary care based, data-driven approaches are emerging.10,11 Primary care, often the first point of contact for many health care services,12 utilizes a model of care which prioritizes patient-centered care, affordability, and prevention. This model of care has been linked to lower mortality rates, better self-reported health status, and reduced impact of income inequality on health outcomes.13
Regarding data-driven approaches, a current review highlights the role of electronic health records (EHRs) in identifying patients, driving tailored intervention, and monitoring results in primary care-based strategies to address disparities in hypertension management.14 Despite the potential and the mounting evidence regarding data/EHR-driven approaches, additional work is needed. For example, best practices for integrating EHR-driven approaches to strengthen hypertension management in racial and ethnic minority populations in primary care have yet to be established.15 However, there is a growing awareness that disparities dashboards can play an important role in this context. In particular, dashboards provide a platform for data visualization, to highlight disparities, and support initiatives to improve health disparities.16
Between 2018 and 2023, the Utah Department of Health and Human Services (UDHHS) and the University of Utah Health (U of U) collaborated to develop an EHR-driven hypertension disparity dashboard (“dashboard”). The overall purpose of this dashboard is to provide clinicians with insights related to health disparities in hypertension management across 12 Community Health Centers and to serve as a foundation for quality improvement (QI) projects. The present study evaluates this dashboard in terms of the approach, sustainability, and usability.
METHODS Dashboard designThe dashboard's main screen (Figure 1) includes a “Data display (1)” section which shows a visual distribution of hypertension across the 13 U of U primary care clinics. The blue chart represents controlled hypertension, the red for uncontrolled hypertension, and the gray for undiagnosed hypertension. The “Data views (2)” allows a user to switch between different dashboard pages. The “summary information (3)” shows the overall number and percentages of patients with hypertension across the clinics. There is also the “filter controls (4)” which allows a user to filter the distribution by specific provider, clinic, or patients’ demographics (such as age, gender, and race and ethnicity).
Study setting and design
This qualitative explorative study was conducted at the University of Utah Health (U of U Health) system in Salt Lake City, Utah, United States. U of U Health provides primary care through 12 health centers in the greater Salt Lake City area, which all operate a shared electronic health records (EHR) system. Overall, the health centers serve a combined total of about 120 000 patients annually. All participants have provided informed consent. The University of Utah Institutional Review Board (IRB #00163534) exempted the study.
Participants and recruitmentStudy participants included quality improvement (QI) managers, data analysts, program managers, evaluators, primary care clinical and research faculty, and medical students. For recruitment, purposive sampling was used. The recruitment started in January 2023 and concluded in March 2023. Individuals were invited to participate and informed about the research project using emails and in-person office visits.
Semi-structured interviewsSemi-structured interviews were conducted to evaluate the approach and the sustainability of the dashboard. The interview guide was developed in an iterative process, including question development, pre-testing and refining (eTable 1). The interviews were conducted in person or virtually via Zoom, audiotaped and lasted about 45 min. Audio files from interviews were transcribed verbatim.
For data analysis, an iterative applied thematic synthesis procedure was used.17 A codebook was developed based on structural codes from the interview guides. For example, structural codes from the development interviews included approach and sustainability. Transcripts of participant's narratives were reviewed and grouped by the structural codes. Two tables were then created, one table for each structural code. Each table had three-columns, category, findings or results, and supporting quotes (eTables 3 and 4).
Think aloud protocolThe Think Aloud Protocol (TAP) was used to evaluate the usability of the dashboard. The Think Aloud Protocol is an interview technique where participants are asked to verbalize their thoughts during a problem-solving task.18 For this analysis, six main tasks for using the dashboard were developed for participants to complete (eTable 2). The interviews were conducted in person, audiotaped and lasted about 45 min.
Audio files from the interviews were transcribed verbatim and analyzed also using an iterative applied synthesis. A codebook was developed for each task, using participant's responses as they completed the tasks. Transcripts were reviewed and grouped by content codes identified from each participant's narrative. Three tables were created, one table for each task. The tables also had three-columns: category, results, and supporting quotes (eTables 5–7).
RESULTS Presentation of resultsFor each of the evaluation categories—approach, sustainability, and usability—the main results are summarized in a table (Tables 2– 4). In addition, to strengthen the readability, only selected quotations are mentioned in the body text. An overview of all relevant quotations and how they informed results can be found in the online appendix.
Participant characteristicsA total of 18 interviews were conducted, 10 to evaluate the approach and the sustainability and 8 to evaluate the usability. Of all participants, 66.7% were females, 27.8% were evaluation managers, 22.2% were each QI manager and research managers (Table 1).
TABLE 1 Participants characteristics.
| Characteristics |
Approach/sustainability N = 10 (%) |
Usability N = 8 (%) |
Overall N = 18 (%) |
| Gender | |||
| Female | 6 (60.0) | 6 (75.0) | 12 (66.7) |
| Male | 4 (40.0) | 2 (25.0) | 6 (33.3) |
| Profession | |||
| Evaluation manager | 4 (40.0) | 1 (12.5) | 5 (27.8) |
| Primary care physician | 1 (10.0) | 1 (12.5) | 2 (11.1) |
| QI manager | 2 (20.0) | 2 (25.0) | 4 (22.2) |
| Data analyst | 1 (10.0) | – | 1 (5.5) |
| Medical students | – | 2 (25.0) | 2 (11.1) |
| Research manager | 2 (20.0) | 2 (25.0) | 4 (22.2) |
Concerning the dashboard approach, five themes were identified. The themes are “purpose and objectives,” “target audience and users,” “development steps,” the “inter-professional approach” as well as “barriers and challenges.” (Table 2; eTable 3).
TABLE 2 Evaluation of the dashboard approach.
| Themes | Result |
| Purpose and objectives |
|
| Dashboard user |
|
| Development processes |
|
| Barriers and challenges |
|
Overall, the participants described the dashboard as a platform to assist clinicians in identifying disparities in hypertension management in general and regarding uncontrolled and undiagnosed hypertension in particular. The dashboard serves to improve the clinical workflow, to increase the quality of care, and to ensure equality in hypertension management.
“The goals of the hypertension dashboard have been to improve clinical processes for clinicians and clinic team members who are utilizing EPIC within the health systems that have the dashboard set up at and specifically focusing on hypertension.” (P2, 345–347)
Primary care physicians are the main target group for the disparity dashboard. In addition, QI teams and researchers, as well as nurses, and medical directors, or other clinicians who want to better understand the current situation of hypertension management overall within the system, for a specific clinic, or a specific patient population.
“Providers [primary care physicians] are I think our primary audience for the toolkit for the dashboard.” (P5, 1319–1321)
The development of the hypertension dashboard incorporated several processes and steps. In a first step, clinical and academic teams needed to specify the purpose of the dashboard, the potential integration in clinical processes, as well as the involvement of relevant stakeholders. Participants noted also that the team went through the process of identifying similar dashboards that exist.
“Step one is identifying like what is already out there and identifying who already kind of got their skin in the game.” (P4, 1038−1039)
Other critical steps included specifying the definition and presentation of aggregated data and the development of the user interface. Participants mentioned that the team needed to think about how to organize the data to improve visual clarity and interpretation for all users. This process included identifying the kinds of hypertension information that users will be expecting to see and outlining the various levels of detail to show.
“You also need to know what kind of filtering mechanisms you want available, and the lowest level of detail you want to be able to drill into and some high-level aggregate views that you want to display.” (P6, 1759–1761)
One major development challenge was compiling data for the dashboard. Some participants mentioned that the data team experienced difficulties with identifying the data source to be used and the logic of accurately defining hypertension. Another challenge was working with various partners and teams. During the initial stages, for some participants, it was not always clear what their roles were or how they could effectively contribute. One participant stated difficulties “bringing everyone together to the table.” Importantly, the COVID-19 pandemic had a substantial impact on collaborations and partnerships. According to one participant, the pandemic caused burnout because individual teams had to handle multiple priority projects simultaneously.
Evaluation of the dashboard sustainabilityRegarding the dashboard sustainability, two main themes identified were: facilitators and barriers to sustainability (Table 3; eTable 4).
TABLE 3 Evaluation of the dashboard sustainability.
| Themes | Result |
| Facilitators |
|
| Barriers |
|
Participants mentioned that the dashboard should be extended for use to deepen understanding of the Social Determinants of Health (SDOH) and hypertension. One participant stated that the dashboard could offer a way to see trends in hypertension by different SDOH measures. Another participant mentioned that the dashboard can also support grant applications and existing QI projects such as those involving medical residents who may be interested in conducting QI-related research.
“The dashboard is like a huge asset for applying for a grant and then for grant activities. We're trying to get a contract. And there's a lot of grants out there in terms of hypertension.” (P1, 199–201)
Additionally, some participants emphasized the need to involve clinical champions in future dashboard projects to support in the clinical implementation and utilization.
“A lesson learned could be to perhaps have gotten like a clinical champion from a clinic that would be involved throughout to help push this.” (P5, 1612–1613)
Limited resources, funding challenges, and lack of dashboard visibility, were recognized as major barriers to sustaining the dashboard. Some participants emphasized that additional resources will be needed to fully implement the dashboard into clinical practice. Participants also discussed learning challenges occurring at the provider-level could be detrimental to future dashboard sustainability efforts. One participant mentioned that physicians may find it burdensome to learn and integrate the dashboard into their clinical workflow.
“I don't think we intend or expect physicians to be in there that much…. Physicians are, we know they're very busy clinicians don't have a lot of extra time, and this is a new system for them that they may be willing to learn about it, but they don't necessarily have the time.” (P1, 47–48; 53–55)
Concerning the lack of dashboard visibility, many departmental and clinic teams are currently not aware of the hypertension dashboard. Participants discussed that the dashboard should be exposed to more institutional divisions, as this will help create potential future use opportunities.
Evaluation of the dashboard usabilityThe analysis of the dashboard usability identified 11 dashboard themes, categorized into three domains (strengths, challenges, and recommendations). These themes describe key components to ensure the usability of an EHR-driven dashboard in primary care (Table 4, eTables 5–7).
TABLE 4 Evaluation of the dashboard usability.
| Domains | Themes | Result |
| Strength | Data presentation |
|
| Page customization |
|
|
| Challenges | Page navigation |
|
| Data customization |
|
|
| Recommendations | Hypertension definitions |
|
| Multiple group comparisons |
|
|
| Extended gender identity |
|
Participants highlighted the dashboard's use of visualizations such as graphs and charts are its most important strength. There was agreement among participants that colored visual representations was an important usability feature. One participant explained that the use of different colors made it easy to see the trends in hypertension.
Participants also highlighted the hover-over function as another positive usability aspect. This feature enabled participants to display the rates and proportions of patients with hypertension when they hovered on a specific chart or patient demographic.
Page customizationThe interviews revealed that providing page filters and tabs were a critical component of usability. According to one participant, the page tabs were helpful for finding specific information. Another positively received feature was the inclusion of a dedicated tab containing patient-specific information. The patients’ detail tab made it easy to complete the task of identifying patients with undiagnosed and uncontrolled hypertension at a specific clinic.
“Especially for how specific you can get with the queries; I think it is also nice to have a lot of the different tabs and sub-tabs to narrow that down.” (P3, 15–17)
In particular, the page tabs presented challenges, as the tab names did not always indicate the information on their respective pages. One participant mentioned that people who were not familiar with the clinics would have difficulty understanding what the tabs are and what they do.
“It will take a minute to figure out what the different tabs were and what they do. People who are not familiar with the clinic or locations and how they relate would have a much harder time completing the tasks.” (P6, 349–352)
Several participants had difficulties making sociodemographic comparisons. Participants expected to find filter options on the summary page, which would have helped show specific demographic information for hypertension. One participant mentioned this feature should have been included on the summary page since it focuses on the overall hypertension distribution.
“There is no option on the summary page to filter by any of the social demographics. You can filter in the patient detail tab, but it does not show the overall distribution that you see on the summary page.” (P1, 97–100)
Also, display issues occurred with some of the pages. Some participants mentioned that the pages become significantly crowded when making demographic comparisons between the clinics. For example, this happens when comparing the age or race demographic distribution of hypertension between multiple clinics.
Another usability challenge was the sorting functions. Some think aloud tasks required participants to identify the provider (s) with the highest rates of hypertension. To complete these tasks, some participants sorted hypertension by the numbers and percentages. However, the dashboard did not always fully sort the information. According to one participant, the page is only sorted by location and not by provider.
“I am trying to use this function to sort the rates from descending to ascending but it only sorts by location and doesn't fully sort the information.” (P2, 299–300)
A key dashboard recommendation was the inclusion of hypertension definitions. One participant specifically recommended a feature that displayed the definitions when a user selected the charts for hypertension. Currently, the dashboard screen shows the rates of undiagnosed, uncontrolled, and controlled hypertension, but makes no mention of how they are determined.
“I would like to see a description maybe like when you hover on the controlled and undiagnosed, how these were defined.” (P8, 382–383)
Participants recommended extending the demographic comparison feature to show a combination of multiple patient demographics. For example, a dashboard user is currently not able to see hypertension rates of different age groups by gender or by race.
“There is no option to combine more than one social demographic group. For example, showing the distribution of hypertension by race stratified by gender or by racial and ethnic minorities by white patients.” (P2 136–139)
Lastly, one participant highlighted the need to include other gender identities. For one participant, the dashboard is potentially missing on patients who do not identify as male or female.
“I would say, this is not inclusive of other gender demographics. For equity, diversity, and inclusion, we need to include other and not just have female and male.” (P7 224–225)
This study evaluated a primary care hypertension disparities dashboard, in terms of the approach, sustainability, and usability. The overall purpose of the dashboard is to help physicians and other clinical teams, such as nurses or medical assistants, leverage clinical data to identify and address hypertension disparities in their clinic setting and patient population.
To support this, the dashboard aggregates and summarizes this data across locations and patient characteristics, providing this information visually to improve hypertension care and management. The graphical representation of hypertension distribution over time enables physicians or clinical leadership teams to identify sites (e.g., clinics or departments) and/or patient groups (e.g., by age, gender, or race and ethnicity) with high hypertension rates. Physicians can review this information to identify patients for outreach or treatment initiation and intensification. Overall, the study has shown that an EHR-driven dashboard can be a useful tool for clinical practice.
Prior studies have reported similar results.19–23 Connolly and colleagues for example, developed and integrated an equity dashboard to analyze clinical performance across their patient populations as defined by patients sociodemographic characteristics. The dashboard provided a platform with which the health system was able to evaluate data documentation quality and consistency, examine readmission rates, and identify gaps in overall patient encounters.22
Behling and colleagues, implemented a hypertension control dashboard to examine and address racial differences in clinical workflows and processes for hypertension management, including medication prescriptions and therapeutic intensification. Although the dashboard was unable to achieve overall health equity between Black/African American and White adults, the study did report a significant improvement in hypertension control rates and follow-up encounters in individual groups. To strengthen hypertension equity, the study plans to incorporate self-blood pressure monitoring and telemedicine into the dashboard program.23
Furthermore, research in information technology has long identified the involvement of health professionals and collaborative teams as a key determining factor in a dashboard development process. Particularly for dashboards intended for clinical use, the involvement of a clinical team is imperative. A recent study suggested that clinical QI and safety dashboards are often underutilized and lack usefulness due to limited engagement of health professionals during the dashboard development process.24 The involvement of health professionals also helps to ensure that QI dashboards are presenting meaningful data and utilizes clinically accepted data definitions and guidelines.25 It is equally important that collaborative teams are aware of their roles and responsibilities and can establish regular communication.
In terms of usability evaluation, the use of colored charts and graphs, the ability to hover-over the charts, and having a dedicated summary page, received the most positive feedback. Participants noted that colored representations of hypertension were self-explanatory and supported their interpretation of the data. The dashboard used blue charts to represent controlled hypertension, red charts for uncontrolled hypertension, and gray charts for undiagnosed hypertension. Prior work supports this finding26,27 and describes color combinations as a key principle of effective messaging and data communication.
In addition, the hover-over feature was perceived as an important dashboard feature. For example, participants were able to hover-over the charts for uncontrolled and undiagnosed hypertension to show the rates and percentages of the distribution. Some studies have emphasized the need for this feature to improve the usability of a data visualization system.27–29 Sometimes, the hover-over feature is even regarded as one of the ten best practices for health dashboards.29
LESSONS LEANREDThe results of this analysis suggest that it is feasible to design an EHR-integrated dashboard to help identify and address hypertension disparities in a healthcare setting. However, there are some areas that could be improved through changes in the dashboard design, user training, and implementation strategies.
First, our evaluation revealed a wide variation in the dashboard user. Participants believe that all clinical teams (e.g., physicians, medical assistants, nurses), including non-patient-facing roles (e.g., QI/data manager, director-level professionals) can utilize the dashboard. However, the dashboard was originally designed for physicians to identify gaps in hypertension management in their patient population. On the other hand, some physicians saw the dashboard as most useful for QI specialists and upper management teams.
While physicians understood the practical need for a dashboard, they preferred that the QI teams use the dashboard to provide updates on how they or their clinic are doing regarding hypertension management rather than having physicians query the dashboard themselves. This may be due to current practices at our clinics, where QI teams regularly provide clinical performance and quality results for a wide range of health conditions, including hypertension.
Second, no “clinical champion” was assigned when the dashboard was developed or implemented. A clinical champion could have been a clinician or clinical staff member who links the dashboard research and clinical leadership or care teams. They also could have helped mobilize and maintain communication with relevant stakeholders at the clinic. However, during the pandemic, clinical processes were focused on managing the pandemic and not on QI activities. This has been a serious challenge for the project in general and the close collaboration with clinical teams in particular.
Finally, the preference for other clinical systems (e.g., EPIC) led to disincentives and limited buy-in to implement the dashboard. For some clinicians, the dashboard was another system to find time to learn and use in addition to other established systems. Some clinicians saw the dashboard as introducing the potential burden of working with two completely different systems for patient care, leading to increased workload and poor healthcare delivery. Additionally, existing systems such as EPIC offer similar functionality and, in some cases, more features (e.g., provider-patient communication) unavailable on the dashboard.
Despite these challenges, feedback on the hypertension dashboard was overwhelmingly positive. A unique attribute of the dashboard is its ability to display hypertension distribution at individual clinic, provider, and patient levels, including providing comparisons across individual patient's demographics (e.g., age, gender, race, ethnicity, statin use, and others) for all levels. Another unique attribute is the inclusion of overall hypertension as well as the distribution of uncontrolled and undiagnosed hypertension, which existing EHR-driven dashboards do not provide.
LimitationsThis study has a few limitations. First, the study is limited by the sample size. There were only 10 participants in the development interviews and 8 participants in the think aloud protocol. This may have impacted the richness and depth of collected data. Second, the hypertension dashboard was developed and evaluated in a primary care setting. This limits the generalizability of the overall findings to other health care systems and patient population. Third, there is the possibility of bias in the TAP interviews. Participant's ability to complete the TAP tasks may have been driven by how well they remembered or completed previous tasks.
CONCLUSIONSAn EHR-driven dashboard can be a novel tool for addressing hypertension disparities in primary care. It offers a platform where clinicians can identify patients for culturally tailored interventions. Factors such as physician time constraints, data definitions, comprehensive patient demographic information, end-users, and future sustenance, should be considered before implementing a dashboard. Additional research is needed to identify best practices for integrating a dashboard into clinical workflow for hypertension.
AUTHOR CONTRIBUTIONSConception and Design: Dominik Ose, Emmanuel Adediran, Emily Carlson. Analysis: Emmanuel Adediran and Alex Lockrey. Interpretation of data: Emmanuel Adediran, Emily Carlson, Alex Lockrey, Dominik Ose, Robert Owens, Elena Gardner. Drafting of the manuscript: Emmanuel Adediran, Dominik Ose. Critical revision of the manuscript for important intellectual content: Emmanuel Adediran, Dominik Ose, Elena Gardner, Alex Lockrey, John Stuligross, Emily Carlson, Danielle Forbes, Robert Owens. Supervision: Dominik Ose, Robert Owens, John Stuligross, Danielle Forbes, Emily Carlson. Tables and Figures: Emmanuel Adediran, Dominik Ose, Alex Lockrey, Elena Gardner. All authors reviewed and approved final versions of the manuscript.
ACKNOWLEDGMENTSThe study was part of research funded by the Utah Department of Health and Human Services (5NU58DP00609-05-00). The funders had no role in the study design, data collection, data analysis, or the decision to submit the paper for publication.
CONFLICT OF INTEREST STATEMENTThe authors declare that they have no competing interests.
DATA AVAILABILITY STATEMENTData requests should be made to Emmanuel Adediran, the corresponding author. The corresponding author will then forward the request to the U of U software licensing office who will process the request, including issuing a data use agreement with the requesting party and providing relevant access information.
© 2024. This work is published under http://creativecommons.org/licenses/by-nc/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Owens, Robert 1 ; Gardner, Elena 1 ; Lockrey, Alex 1 ; Carlson, Emily 2 ; Forbes, Danielle 3 ; Stuligross, John 4 ; Ose, Dominik 5 1 Department of Family and Preventive Medicine, University of Utah, Salt Lake City, Utah, USA
2 Community Physicians Group, University of Utah, Salt Lake City, Utah, USA
3 Utah Department of Health and Human Services, Salt Lake City, Utah, USA
4 Intermountain Healthcare, Salt Lake City, Utah, USA
5 Department of Family and Preventive Medicine, University of Utah, Salt Lake City, Utah, USA; Department of Health and Healthcare Sciences, Westsächsische Hochschule Zwickau, Zwickau, Saxony, Germany