Technology-enabled Management Versus Usual Care for Blood Pressure

For this study, we have 3 cohorts that employ varying recruitment strategies before they are enrolled in the interventions.

At Stanford, we have 2 cohorts, where Cohort 1 is recruiting participants through clinic screening and contacting them through text or email.

For Cohort 2, recruitment is divided into two phases: Control Phase I and Intervention Phase II. During Control Phase I, standard recruitment methods were employed, including general outreach through participating clinics and pre-existing clinical networks. These efforts did not incorporate specific strategies aimed at enhancing diversity or prioritizing underrepresented populations. In contrast, Intervention Phase II introduced four targeted strategies designed to improve the inclusion of African-American, Hispanic/Latinx individuals, and women.

1. Digital Outreach via BuildClinical: The study leverages a digital recruitment campaign managed by BuildClinical, utilizing targeted online advertisements across various platforms. Prospective participants who engage with the advertisements are directed to a study landing page, where they receive detailed study information and complete a preliminary eligibility screening before being contacted by the study team.
2. Patient Registry-Based Recruitment: Eligible participants will be identified through structured data queries from the STAnford Research Repository (STARR) or EPIC registry. This approach ensures targeted outreach to individuals who meet the study's inclusion criteria while reducing screening burden.
3. Honest Broker Outreach via Stanford RPEP: To maintain privacy compliance, the Stanford Research Participant Engagement Program (RPEP) will conduct Honest Broker outreach via multiple channels, including secure Epic MyHealth messages, postal mail, and provider co-invitations. This method allows potential participants to opt in or out before direct contact with the research team, ensuring ethical and efficient recruitment.
4. Spanish-Speaking Ambassador Program: A dedicated Spanish-speaking ambassador will support recruitment efforts by engaging with Spanish-speaking participants, providing culturally tailored study information, and assisting with the informed consent process. This initiative ensures equitable access for Spanish-speaking individuals and facilitates participation among historically underrepresented populations. Our TIDE site partners at the Morehouse School of Medicine utilize their own recruitment methods for cohort 3, which uses the following: Medical residents will be randomized into three screening groups using a block randomization approach to ensure demographic balance across the groups (e.g., age, years in training, prior experience). Each group will utilize a different screening method for patient recruitment: Arm 1: Resident-Led Manual Screening Residents apply eligibility criteria manually by reviewing patient EMRs and making judgment-based decisions. Arm 2: AI-Assisted Screening AI software, Deep 6, will scan EMRs for inclusion criteria and flag potentially eligible patients. Residents will then validate flagged cases to confirm or deny eligibility. Arm 3: Coordinator-Led Screening (Control) Coordinators screen patients independently of AI, applying eligibility criteria consistently across all patient records. These comprehensive recruitment strategies are designed to enhance enrollment diversity, improve participant engagement, and ensure a balanced distribution of participants across study arms. Primary and Secondary Endpoints for Cohort 1 Primary Endpoint

• Percentage of participants who consent to participate following text message compared to the percentage of participants who consent to participate following emailoutreach. Secondary Endpoints ● Percentage of participants who consent following either text message or telephone outreach. ● Percentage of participants who consent after opting out after receiving a call or text message.
• Median time from call/text initiation to participant response among all participants who are contacted and respond.
• Percentage of African American, Hispanic, or Asian participants who consent to participate following text message compared to the percentage of African American, Hispanic, and Asian participants who consent to participate following email outreach.
• Percentage of African American or Hispanic participants who consent to participate following text message compared to the percentage of African American or Hispanic, participants who consent to participate following email outreach.
• Percentage of Asian participants who consent to participate following text message compared to the percentage of Asian participants who consent to participate following email outreach. Primary and Secondary Endpoints for Cohort 2 Primary Endpoint ● Percentage of participants enrolled who are either African-American or Hispanic/LatinX during Intervention Phase II compared to Control Phase I. Secondary Endpoints ● Percentage of African-American or Hispanic/LatinX participants who screen positive during Intervention Phase II compared to Control Phase I. ● Percentage of Hispanic/LatinX participants enrolled during Intervention Phase II compared to Control Phase I. ● Percentage of African-American participants enrolled during Intervention Phase II compared to Control Phase I. ● Percentage of women participants enrolled during Intervention Phase II compared to Control Phase I. ● Percentage of African-American, Hispanic/LatinX, or women participants enrolled during Intervention Phase II compared to Control Phase I. ● Percentage of African-American, Hispanic/LatinX, or women participants who screen positive during Intervention Phase II compared to Control Phase I. ● Percentage of Hispanic/LatinX participants who screen positive during Intervention Phase II compared to Control Phase I. ● Percentage of African-American participants who screen positive during Intervention Phase II compared to Control Phase I. ● Percentage of women participants who screen positive during Intervention Phase II compared to Control Phase I. ● Percentage of African-American or Hispanic/LatinX participants contacted through Honest Broker and enrolled during Intervention Phase II compared to the Percentage of African-American or Hispanic/LatinX participants in Control Phase I. ● Percentage of women participants contacted through Honest Broker and enrolled during Intervention Phase II compared to the percentage of women participants in Control Phase I. ● Percentage of African-American or Hispanic/LatinX participants contacted through Patient Registry and enrolled during Intervention Phase II compared to the Percentage of African-American or Hispanic/LatinX participants in Control Phase I. ● Percentage of women participants contacted through Patient Registry and enrolled during Intervention Phase II compared to the percentage of women in Control Phase I.
• Percentage of African-American or Hispanic/LatinX participants contacted through BuildClinical and enrolled during Intervention Phase II compared to the Percentage of African-American or Hispanic/LatinX participants contacted in Control Phase I.
• Percentage of women participants contacted through BuildClinical and enrolled during Intervention Phase II compared to Percentage of women participants enrolled in Control Phase I.
• Percentage of Hispanic/LatinX participants contacted through Lay Coordinator and enrolled during Intervention Phase II compared to the Percentage of African-American or Hispanic/LatinX participants contacted in Control Phase I. Primary and Secondary Endpoints for Cohort 3 Primary Endpoint ● Screening Accuracy: Accuracy metrics include true positive, true negative, false positive, and false negative rates across the three arms, providing insight into each method's reliability and potential for error. Secondary Endpoints ● Representativeness: This measures the number and characteristics (e.g., age, race/ethnicity) of patients recruited in each arm, allowing for assessment of demographic representativeness. ● Enrollment Screening Efficiency: This is measured as the average time per screened patient across the three arms. Efficiency will be assessed based on total time to screen each patient from record access to eligibility determination.
• Workload and Satisfaction: Surveys will capture resident and coordinator perceptions of workload, satisfaction, and feasibility of each method, with Likert scales and qualitative feedback (Ellis \& Kasperbauer, 2020). Statistical Analysis Plan (SAP) Cohort 1 Analysis Methods The proportion of participants who consent following text message or email-telephoneoutreach will be calculated. Median and interquartile range will be estimated for the time from email /text initiation to participant response. Cohort 2 Analysis Methods This study aims to increase recruitment through a two-phase trial that will compare a combination of recruitment strategies through 3 distinct cohort-based recruitment strategies to assess their effectiveness to improve participant diversity and engagement. Participants are control phase I or intervention phase II. The primary statistical test will be a two-sample Z-test for proportions, comparing enrollment percentages in African American/Black and Hispanic/Latino between the control phase I and intervention phase II. The same method will be applied to the secondary endpoints, comparing recruitment rates across demographic subgroups and different recruitment strategies. All tests will be one-sided with α=0.05. Based on the preliminary data from February 2025, the control phase enrolled 135 participants, with the following distributions:
• African American/Black: 8 (5.92%)
• Hispanic/Latino: 17 (12.6%) The intervention phase II aims for a 50% increase in African American/Black and Hispanic/Latino enrollment. With a total of 270 participants (135 per phase), the power calculation indicates that the statistical power to detect this significant increase is 100%. Cohort 3 Analysis Methods This cohort will evaluate the efficiency, accuracy, satisfaction, and representativeness across the three screening arms. The primary metric for assessing efficiency will be the mean screening time per patient. A one-way ANOVA will be conducted to compare mean screening times among the arms, followed by post-hoc tests if significant overall differences are detected. For accuracy, the analysis will be based on true and false positive/negative rates across screening methods. Chi-square tests will be used for categorical comparisons, and logistic regression will be applied to control for potential confounding variables, such as resident training level and experience. Representativeness and satisfaction will be evaluated through the analysis of demographic characteristics, including age, race, and gender along with responses to surveys assessing workload and satisfaction. Chi-square tests will be used to analyze categorical demographic data, while the Kurskal-Wallis test will be employed to compare non-parametric Likert-scale responses across all study arms. Additionally, open-ended feedback will be thematically analyzed to enrich the interpretation of quantitative findings.