Using AI Text Messaging to Improve AHA's Life's Essential 8 Health Behaviors

Not Applicable

Recruiting

• Conditions: Lifestyle Factors; Cardiovascular Diseases
• Interventions: Behavioral: Generic text messages; Behavioral: Interactive AI chatbot text messaging; Behavioral: Proactive pharmacist support

• Registration Number: NCT06324981
• Lead Sponsor: University of Colorado, Denver

Brief Summary

The goal of our pragmatic clinical trial is to compare how well three different strategies might do to reduce risk factors for cardiovascular disease in patients experiencing health disparities. The three different strategies are: 1) text messages, 2) interactive chatbot messages, and 3) chatbot messages with proactive pharmacist support. To measure cardiovascular risk factors, the investigators are using the American Heart Association's Life's Essential 8 (LE8) factors-blood glucose, cholesterol, blood pressure, physical activity, body mass index, diet, and smoking.

This study focuses on improving cardiovascular risk factors for individuals facing health disparities, such as ethnic minorities, limited English proficiency, and low-income groups. These groups are more likely to be seriously affected by cardiovascular diseases. Self-management, or an individual's roles in managing their own chronic disease, includes lifestyle changes, medication adherence. Improving patients' self-management has been shown to improve health behaviors, better disease control and improved patient outcomes.

The main question this study aims to answer is if one of the strategies (texting, chatbot, or chatbot with pharmacist support) may improve patient self-management and patient outcomes.

The investigators will enroll up to 2,100 patients from three health systems that serve large populations experiencing health disparities: Denver Health, Salud Family Health Centers, and STRIDE Community Health Center.

The results might help researchers and health care systems find the best ways to involve patients with health disparities to managing their chronic cardiovascular disease.

Detailed Description

Our goal is to improve control of cardiovascular (CV) disease risk factors by engaging patients experiencing health disparities in an innovative technology-based self-management intervention with linkages to health system providers. The investigators will focus on the American Heart Association's Life's Essential 8 (LE8) lifestyle factors (blood glucose, cholesterol, blood pressure, physical activity, body mass index, diet, and smoking), that when uncontrolled lead to common co-existing chronic conditions (e.g., hypertension, diabetes), morbidity, health care costs and death. Patients disproportionately affected by these risk factors (e.g., Black, Hispanic/Latino), have worse disease control with greater adverse sequelae (e.g., heart attacks and death).

Self-management is an individual's role in managing chronic disease and has strong evidence of benefit. It includes self-care, a healthy lifestyle (e.g., being physically active), taking medications as prescribed and managing exacerbations of chronic condition(s). Self-management for patients experiencing disparities is enhanced when programs recognize patient context and sociocultural factors that may modify healthy behavior. Self-management can be further enriched when patients are directly supported by their health care provider. Ample evidence shows text messaging can impact self-management behaviors, with the advantage of being universally available through mobile phones. Emerging technologies utilize artificially intelligent (AI) chatbots for the delivery of text messages have the promise of improving the impact of text messaging, particularly if they integrate evidence based communication strategies, including tailoring, behavioral nudges that support intuitive decision-making, and persuasive messaging. These strategies can optimize message content beyond generic, "one size fits all" communication. It is unknown if AI chatbot text messaging with linkages to providers can improve self-management support in large diverse patient populations.

Using a patient level randomized pragmatic trial in 3 health systems caring for large patient populations experiencing health disparities, the investigators will test the comparative effectiveness of theory-based, tailored and socially contextualized communications for self-management support. Patients with CV disease risk factors will be randomized to 1 of 3 automated communication approaches: 1) generic text messages; 2) interactive AI chatbot text messaging leveraging evidenced-based communication strategies with attention to patient context and sociocultural factors influencing self-management; or 3) interactive AI chatbot text messaging plus proactive pharmacist management. Our goal is to increase patient self-management autonomy, competence, and relatedness to health systems, leading to improved and sustained health behaviors, better disease control and improved patient outcomes. The primary effectiveness outcome will be an improved LE8 health score. The investigators will partner with: 1) Salud Family Health Centers, a Federally Qualified Health Center (FQHC) with 13 clinics across Colorado, 2) Denver Health and Hospital Authority, a safety net health system with 9 FQHC clinics, and 3) STRIDE Community Health Center, a FQHC with18 locations surrounding Denver county. The investigators will enroll diverse patients including: Black, Hispanic/Latino, low-income, Spanish speaking-only and rural patients with at least one LE8 factor in the poor/intermediate health category and poor adherence to CV medications. Patients will be identified using demographic, clinical and pharmacy EHR data from each health system. In Year 1 (UG3 phase), applying the Health Equity in Implementation Framework, the investigators will partner with patients, providers, community advocates and health systems stakeholders to develop the AI chatbot infrastructure and message content relevant to the patient population using an intervention mapping approach; assess how best to integrate the intervention within each health system's existing CV prevention programs; and conduct a pilot study of the intervention. In Years 2-5 (UH3 phase), the investigators will conduct a pragmatic patient randomized trial.

Aim 1 (UG3; Year 1): Iteratively update the infrastructure and expand content for the AI text message chatbot with attention to social determinants of health and sociocultural contextual relevant to the target population through stakeholder engaged N-of-1 and focus group interviews and nominal group sessions.

Aim 2 (UG3; Year 1): Conduct a randomized pilot to demonstrate feasibility of intervention delivery and outcomes data collection to assess preliminary effects and to refine the intervention prior to widespread implementation Aim 3 (UH3; Years 2-5): Conduct a pragmatic patient-level randomized intervention of 3 text messaging delivery strategies for self-management support of CV risk factors. Primary outcome will be change in LE8 health score. Secondary effectiveness outcomes will include individual components of the LE8 lifestyle factors, Framingham risk score, self-efficacy, medication adherence, clinical outcomes (e.g., CV related hospitalizations), and healthcare utilization.

Aim 4 (UH3; Years 2-5): Evaluate the intervention using PRISM and a mixed methods approach to evaluate pragmatic clinical and implementation outcomes (reach, effectiveness, adoption, implementation, and maintenance) with an emphasis on equity and representativeness, and systematically assess contextual influences to inform sustainment and future tailoring, adaptations, and dissemination.

Study Timeline

• Study First Submitted: 2023-12-08
• Study Start: 2024-02-26
• Status Verified: 2024-03
• Study First Submitted QC: 2024-03-14
• Last Update Submitted: 2024-03-14
• Study First Posted: 2024-03-22
• Last Update Posted: 2024-03-22
• Primary Completion: 2028-02
• Study Completion: 2028-06

Recruitment & Eligibility

• Status: RECRUITING
• Sex: All
• Target Recruitment: 2100

Inclusion Criteria

• diagnosis of one or more of the following CV risk factors (i.e., hypertension, diabetes or hyperlipidemia); and
• the risk factor is at poor or intermediate health levels as defined by LE8 (e.g., BP>140/90 mm Hg); and
• the patient exhibits poor adherence to prescribed medication to treat the CV risk factor as defined by a delay in refilling the medication within the past 6 months.

Read More

Exclusion Criteria

• patients who do not have cellphone; or
• enrolled in hospice or palliative care; or
• Non-English or Spanish speaking; or
• enrolled in another clinical trial if denoted in the EHR.

Read More

Study & Design

• Study Type: INTERVENTIONAL
• Study Design: PARALLEL

Arm && Interventions

Group	Intervention	Description
Interactive AI chatbot text messaging + proactive pharmacist management	Proactive pharmacist support	The AI chatbot will be the same as arm 2 (Interactive AI chatbot text messaging alone). In this arm, however, pharmacists will review patient's baseline LE8 risk factors and proactively contact patients via telephone and/or the EHR patient portal to address any risk factor that is in poor/intermediate health categories. The investigators are proposing proactive pharmacist involvement as a population-based approach to address patients with uncontrolled CV risk factors.
Interactive AI chatbot text messaging	Proactive pharmacist support	This AI system will utilize NLP and ML to facilitate bi-directional system-patient dialogue with messages that incorporate content utilizing tailoring, behavioral nudges and persuasive messaging as described above. An example message would be: Make a promise to yourself to check your blood pressure today! Your goal is to have the top number at 120 or lower and the bottom number at 80 or lower. Each message will end with a question for the participant that will encourage engagement with the AI conversational chatbot that allows greater opportunity to use theoretical content to engage patient autonomy, competence and relatedness, the mechanisms through which we will impact behaviors.
Generic text messages	Generic text messages	The information content for these messages will be derived from trusted sources of medical information and contain links to websites such as American Heart Association. An example of such a message would be: Remember to take your blood pressure today! You can find more information from the American Heart Association by clicking here. Patients will be able to return texts with questions which will be addressed by the study team, including a clinical pharmacist if needed.
Interactive AI chatbot text messaging + proactive pharmacist management	Interactive AI chatbot text messaging	The AI chatbot will be the same as arm 2 (Interactive AI chatbot text messaging alone). In this arm, however, pharmacists will review patient's baseline LE8 risk factors and proactively contact patients via telephone and/or the EHR patient portal to address any risk factor that is in poor/intermediate health categories. The investigators are proposing proactive pharmacist involvement as a population-based approach to address patients with uncontrolled CV risk factors.

Primary Outcome Measures

Name	Time	Method
Change in Life's Essential 8 risk score	Baseline and 12 months after randomization	The primary outcome is change in Life's Essential 8 risk score from baseline and 12 months following randomization.; The Life's Essential 8 (LE8) is a cardiovascular health score that uses a 0-100 scale. The score is calculated based on a participant's adherence to eight healthy lifestyle components: diet, physical activity, smoking habits, body mass index, total cholesterol, blood sugar, blood pressure, and sleep. Each component has scoring algorithm ranging from 0 to 100 points, allowing generation of a composite cardiovascular health score (the unweighted average of all components) that also varies from 0 to 100 points. 0 will indicate the lowest cardiovascular health scores and 100 will indicate the highest cardiovascular health scores.

Secondary Outcome Measures

Name	Time	Method
Self-Efficacy for Managing Chronic Diseases	Baseline and 12 months after randomization	The Self-Efficacy for Managing Chronic Disease Scale is a valid and reliable instrument available in English and Spanish. The English version is made up of 6-items on a visual analog scale, ranging from 1 (not at all confident) to 10 (totally confident).
Sleep (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8, including change in sleep, between baseline and 12-months following enrollment will be assessed. Since sleep is not observable in the EHR, the investigators will ask patients via text to self-report their status at baseline and 12-months following enrollment. Patients will be asked to report how many hours of sleep do you get per night on average. Patients can enter the number of hours, or can choose to skip the question or select "Prefer not to answer".; For the primary outcome composite score: 100 points: 7-\<9 hours of sleep per night; 0 points: \<4 hours of sleep per night.
Total cholesterol (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8, including change in total cholesterol, between baseline and 12-months following enrollment will be assessed. Data will be derived from the EHR. For patients without a baseline measure derived from the EHR, the investigators will encourage patients to talk to their physician about obtaining a measure consistent with LE8 recommendations. For the 12-month measurement, the investigators will take the value closest to the 12-month post enrollment date with a 3-month window prior to and after the 12-month enrollment date.; For the primary outcome composite score: 100 points: \<130 mg/dL; 0 points: ≥220 mg/dL
Blood pressure (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8 (LE8), including change in blood pressure (both Systolic and Diastolic Blood Pressure), between baseline and 12-months following enrollment will be assessed. Data will be derived from the EHR. For patients without a baseline measure derived from the EHR, the investigators will encourage patients to talk to their physician about obtaining a measure consistent with LE8 recommendations. For the 12-month measurement, the investigators will take the value closest to the 12-month post enrollment date with a 3-month window prior to and after the 12-month enrollment date.; For the primary outcome composite score: 100 points: \<120/\<80 mm Hg; 0 points: ≥160 mm Hg or ≥100 mm Hg.
Blood sugar (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8, including change in blood sugar, between baseline and 12-months following enrollment will be assessed. Data will be derived from the EHR. For patients without a baseline measure derived from the EHR, the investigators will encourage patients to talk to their physician about obtaining a measure consistent with LE8 recommendations. For the 12-month measurement, the investigators will take the value closest to the 12-month post enrollment date with a 3-month window prior to and after the 12-month enrollment date.; For the primary outcome composite score: 100 points: No history of diabetes and hemoglobin A1c \<5.7; 0 points: Diabetes with hemoglobin A1c ≥10.0
Smoking habits (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8, including change in smoking habits, between baseline and 12-months following enrollment will be assessed. Since smoking habits are not observable in the EHR, the investigators will ask patients via text to self-report their status at baseline and 12-months following enrollment. Patients will be asked if they smoke cigarettes, cigars, little cigars, pipes, water pipes, hookah, use e-cigarettes or any other tobacco product? Options may include Yes, No, Prefer not to answer, or Skip.; For the primary outcome composite score: 100 points: Never smoker; 0 points: Current smoker.
Health diet pattern (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8, including change in diet patterns between baseline and 12-months following enrollment. As diet patterns are not observable in the EHR, the investigators will ask patients via text to self-report at baseline and 12-months following enrollment. The study will use the Mini-Eating Assessment Tool (Mini-EAT), a 9-item dietary screener that includes fruits/vegetables, whole grains, refined grains, fish/seafood, legumes/nuts/seeds, low-fat dairy, high-fat dairy, and sweets consumption. The 9-item survey will provide a single score (range 0-100; scores \<61 indicate an unhealthy diet, 61-69 indicate an intermediate diet, \>69 indicate a healthy diet).; For the primary outcome composite score: 100 points: ≥95th percentile (top/ideal diet); 0: 1st-24th percentile (bottom/least ideal quartile).
Number of medication refill gaps (Medication adherence)	Baseline and 12 months after randomization	The investigators will measure medication adherence in one of two ways. First, investigators will identify the number of gaps (frequency) for every patient and medication. Gaps will be determined using pharmacy refill data based on the date of refill, the number of days supplied, and the subsequent refill date during the 12-month intervention period. Worse medication adherence will be identified as an increase in the frequency of gaps. The study is currently using this same methodology in the Nudge study.
Clinic events	Baseline and 12 months after randomization	Clinical events are defined as emergency department (ED) visits or hospitalizations.
Body Mass Index (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8, including change in body mass index, between baseline and 12-months following enrollment will be assessed. Data will be derived from the EHR. For patients without a baseline measure derived from the EHR, the investigators will encourage patients to talk to their physician about obtaining a measure consistent with LE8 recommendations. For the 12-month measurement, the investigators will take the value closest to the 12-month post enrollment date with a 3-month window prior to and after the 12-month enrollment date.; For the primary outcome composite score: 100 points: Body mass index (kg/m2) \<25; 0 points: Body mass index (kg/m2) ≥40.0
Physical activity (Individual Life's Essential 8 components)	Baseline and 12 months after randomization	Change in the individual risk factors of the Life's Essential 8, including change in physical activity, between baseline and 12-months following enrollment will be assessed. Since physical activity is not observable in the EHR, the investigators will ask patients via text to self-report their status at baseline and 12-months following enrollment. Patients will be asked to report the total number of minutes of vigorous physical activity they get in a typical week, and the total number of minutes of moderate physical activity they get in a typical week. Patients can enter the number of minutes for each question, or can choose to skip the question or select "Prefer not to answer".; For the primary outcome composite score: 100 points: ≥150 minutes of moderate or vigorous activities per week; 0 points: 0 minutes
Cost	Baseline and 12 months after randomization	Medical care costs will be estimated using a resource-based method previously developed to assign costs to encounter data. Inpatient utilization will be measured using diagnostic-related groups (DRGs), outpatient utilization using relative value units (RVUs), and pharmacy utilization using average wholesale prices (AWPs). Inpatient costs will be estimated by applying national payment weights to DRGs, outpatient costs by applying a national conversion factor to RVUs, and pharmacy costs at 69% of the AWP during a reference year. Cost data will be analyzed using generalized gamma regression accounting for study arm and health system. The resources to both develop and implement the intervention will also be collected.
Length of refill gaps (Medication adherence)	Baseline and 12 months after randomization	The investigators will measure medication adherence in one of two ways. In the second method, investigators will measure medication adherence by measuring the length of each gap (severity) for every patient and medication. The length of each gap will be determined using pharmacy refill data based on the date of refill, the number of days supplied, and the subsequent refill date during the 12-month intervention period. Worse medication adherence will be identified as the length (severity) of the gaps. We are currently using this same methodology in the Nudge study
Risk Score for Recurrent Coronary Heart Disease, Framingham Risk Score	Baseline and 12 months after randomization	Investigators will determine subsequent Coronary Heart Disease risk based on the following risk factors obtained via the Electronic Health Records and/or patient self-report: age, systolic blood pressure, cigarette smoking status, fasting lipid level (totals and HDL Cholesterol), and diagnosis of diabetes. Points are assigned based on the presence or level of risk factors, and will be converted to risk percentages using a conversion algorithm. The range is 0%-22%, with 0% being the lowest risk and 22% being the highest risk of recurrent coronary heart disease. This score will be calculated for patients who have had a prior CHD event at the time of study enrollment based on diagnosis codes of prior Coronary Heart Disease event or ischemic stroke.
Rate of routine clinical visits and/or other procedures associated with the clinical condition	Baseline and 12 months after randomization	The study will also measure healthcare utilization defined by routine clinical visits and/or other procedures associated with the clinical condition.
Risk Score for Coronary Heart Disease (2-year risk) - First Event, Framingham Risk Score	Baseline and 12 months after randomization	Investigators will determine the 2 year risk of Coronary Heart Disease in patients free of cardiovascular diseases at the time of study enrollment and include the following risk factors to be gathered via the Electronic Health Records or patient self-report: age, systolic blood pressure, cigarette smoking status, fasting lipid level (totals and HDL Cholesterol), diagnosis of diabetes, use of antihypertensive medication. Points are assigned based on the presence or level of risk factors, and can be converted to risk percentages using a conversion table. The range is 0%-43%, with 0% being the lowest risk and 43% being the highest risk of developing coronary heart disease.

Trial Locations

Locations (3)

Salud Family Health Centers; 🇺🇸 Fort Lupton, Colorado, United States

STRIDE Community Health Centers; 🇺🇸 Wheat Ridge, Colorado, United States

Denver Health; 🇺🇸 Denver, Colorado, United States