Research

Principal Investigator: Stephen D. Persell, MD, MPH

Automated patient messaging based on home blood pressure (BP) monitoring may help address clinician and patient inertia in hypertension management. This pilot study evaluated the feasibility and impact of an automated messaging system integrated into the electronic health record (EHR) that delivered tailored feedback to patients based on remotely reported home BP readings. Seven clinicians participated, and patients with two recent office BP measurements of 140/90 mmHg or greater were eligible.

The system used a deterministic algorithm to:

• Trigger messages that reminded patients to monitor their BP.
• Provide guidance when readings were above or below goal.
• Reinforce progress when BP targets were met.

The study followed participants over nine months to evaluate:

• Changes in home systolic BP (SBP),
• Adjustments in antihypertensive medications
• Message engagement, with subgroup analyses by baseline home BP control.

Results

Of 285 eligible patients, 70 enrolled and 64 completed the nine-month follow-up. Participants received an average of 13 automated messages, with high message view rates. Overall, home SBP decreased by 9.4 mmHg over nine months. Patients with uncontrolled baseline home BP experienced substantial reductions in SBP (-12.5 mmHg), and just over half had intensification of antihypertensive therapy, whereas those with controlled baseline BP showed no meaningful change.

These findings demonstrate that EHR-integrated, automated feedback based on remote BP monitoring is feasible and may substantially improve BP control among patients with sustained hypertension.

Principal Investigator: Stephen D. Persell, MD, MPH

This AUTO2 Trial will evaluate whether computer-generated medical advice sent to patients' electronic medical records can improve blood pressure. The study will enroll patients with uncontrolled hypertension across a large, diverse health system.

Using a 2x2 factorial, patient-randomized controlled trial (the AUTO2 Trial), investigators will evaluate the effects of:

1. AUTOmated transmission of remotely obtained blood pressure values versus manual transmission,
2. AUTOmatically-generated medical advice to patients related to their home blood pressure monitoring results sent via patient portal messages versus no automated feedback.

Study Aims

1. Measure the systolic blood pressure after one year.
2. To assess whether these two interventions interact. Automated messaging that includes prompting to perform self-monitoring and other feedback could lead patients to sufficiently self-report their blood pressures and obviate the need for direct transmission of blood pressures. Alternatively, the two interventions could have synergistic effects leading to more than additive benefits.
3. To evaluate intervention effects on resource utilization and clinician attitudes toward the use of these remote monitoring systems and examine how the remote blood pressure monitoring information is used by physicians, advanced practice providers and their teams to deliver patient care.

Potential Impact

This trial has the potential to demonstrate the effectiveness of practical strategies to accelerate the achievement of evidence-based hypertension treatment goals that could be incorporated into existing health systems. Public health benefit: A modest reduction in population systolic blood pressure of five mm Hg or even less is expected to produce important reductions in cardiovascular events, stroke and heart failure and to confer other important health benefits. Identifying and testing light-touch strategies to accelerate the achievement of evidence-based hypertension goals that are feasible and affordable for health systems to implement and for patients to use is needed to realize the proven benefits of controlling hypertension.

Principal Investigator: Stephen D. Persell, MD, MPH

Mobile health applications may support hypertension self-management, but their effectiveness compared with simpler tools remains uncertain.

This randomized clinical trial evaluated whether an artificial intelligence (AI)-based smartphone coaching app could improve systolic blood pressure (SBP) and hypertension-related behaviors compared with a standard blood pressure (BP) tracking app. A total of 333 adults with uncontrolled hypertension were randomized to receive either a coaching app designed to promote home BP monitoring and behavior change or a BP tracking app; all participants received a home BP monitor. Participants were followed for six months.

Study Outcomes

• SBP after six months (primary outcome)
• Medication adherence
• Home monitoring practices
• Self-efficacy
• Health behaviors

Results

Among the 297 participants who completed follow-up, SBP decreased in both groups over six months. The intervention group had a slightly lower mean SBP than the control group, but the between-group difference was not statistically significant.

Participants using the coaching app reported significantly greater confidence in their ability to control blood pressure, while no significant differences were observed in other secondary outcomes. Subgroup analyses suggested that effects may vary by age.

Overall, the AI-based coaching app did not significantly reduce SBP compared with a BP tracking app, though improvements in self-efficacy and trends toward greater BP reduction suggest that further research is needed to identify populations most likely to benefit.

Related Publication

• Development and Pilot Testing of Hypertension Management Support With EHR-Integrated Telemonitoring and Proactive Patient Messaging

Principal Investigator: Stephen D. Persell, MD, MPH

Diagnostic and treatment strategies often carry different risks and benefits for older adults than for middle-aged patients, and misapplying standard approaches can increase morbidity and mortality in geriatric care.

Common examples of such misuse, highlighted by the American Geriatrics Society's Choosing Wisely initiative, include:

• Testing and treating asymptomatic bacteriuria.
• Prostate-specific antigen screening in older men without prostate cancer.
• Overly intensive glucose-lowering therapy in Type 2 diabetes.

Clinicians may deviate from best evidence due to underestimating downstream harms, overestimating the risks of inaction, responding to social norms, relying on habitual practices or avoiding perceptions of ageism. Clinical decision-support nudges delivered through electronic health records (EHRs) offer a promising, low-cost approach to addressing these issues by targeting both conscious and unconscious drivers of decision-making.

The BEAGLE study will develop, pilot and refine EHR-based nudges to reduce overuse in older adults, culminating in a large cluster-randomized trial to evaluate their impact on care quality, patient safety and clinician attitudes.

Related Publications

• Reducing Care Overuse in Older Patients Using Professional Norms and Accountability: A Cluster Randomized Controlled Trial
• Persistence of Effects of Behavioral Interventions on Reducing Overuse of Care in Older Patients After Discontinuation

Principal Investigator: Stephen D. Persell, MD, MPH

High-risk polypharmacy is common among older adults in the United States, particularly among those with dementia or cognitive impairment and is associated with adverse drug reactions, falls and increased healthcare costs.

Study Aims

This project will test whether low-cost, scalable behavioral economic nudges embedded in electronic health records (EHRs) can help clinicians reduce high-risk polypharmacy, addressing limitations of prior interventions that relied on rational decision-making models and had limited effectiveness and uptake.

Using a pragmatic, clinic-randomized controlled trial with a factorial design across two large health systems, the study will evaluate the impact of commitment and justification nudges — alone and in combination — on high-risk polypharmacy over an 18-month intervention period with additional follow-up to assess durability of effects.

The project will also examine clinician experiences, workflow impacts and acceptability through surveys and interviews, with the goal of generating effective, generalizable EHR-based strategies to reduce medication-related harm in older adults nationwide.

Related Publication

• Pilot Project Manuscript

Principal Investigators: Stephen D. Persell, MD, MPH, and Kimberly Kenton, MD

This project proposes an innovative, automated approach to improve the identification, education and non-surgical treatment of urinary incontinence (UI) in women within a large, geographically diverse health system. Using a single validated screening question during annual primary care visits, women with bothersome UI will be identified and directed to a web-based shared decision-making module (Wisercare©) that educates patients about UI subtypes, causes, and treatment options and generates a personalized treatment preference map.

Results are automatically integrated into the electronic health record and shared with both patients and primary care clinicians, enabling streamlined, patient-centered treatment initiation without adding burden to already overextended providers. Close collaboration among primary care leaders, UI subspecialists, and pelvic floor physical therapists ensures seamless community-based, non-surgical care, while ongoing evidence updates and automated workflows enhance scalability, sustainability and generalizability to other primary care settings.

Principal Investigators: Jeffrey Linder, MD, MPH, and Jason Doctor, PhD

Chronic insomnia guidelines recommend cognitive behavioral therapy for insomnia (CBT-I) as first-line treatment, yet pharmacologic sedatives known as Z-drugs remain the most commonly prescribed therapy despite not being recommended for long-term use.

The Study in Outpatient Medicine Using Nudges to Improve Sleep (SOMNUS) is an 18-month, factorial, cluster-randomized trial designed to promote guideline-concordant insomnia care in primary care by encouraging CBT-I use and reducing Z-drug prescribing through behavioral nudges.

Conducted at Northwestern Medicine, the trial will enroll 444 primary care clinicians across 64 outpatient clinics.

Study Design

Using a 2x2 factorial design, clinics will be randomized to receive one or both of two electronic health record (EHR)-embedded nudges or neither:

• A default limit of 10 pills with no refills for new Z-drug prescriptions.
• A just-in-time prompt offering a one-click guideline-aligned order and requiring justification for non-adherence.

Outcomes

The primary outcome is the change in encounter-level Z-drug pill count over 18 months before and after intervention initiation, with secondary outcomes including CBT-I referrals, monthly Z-drug use, treatment duration and benzodiazepine prescribing. Overall, SOMNUS will evaluate whether scalable, EHR-based behavioral nudges can improve adherence to insomnia treatment guidelines and optimize care in primary care settings.

Related Publication

• The Study in Outpatient Medicine using Nudges to Improve Sleep (SOMNUS): study protocol for an 18-month factorial, cluster-randomized trial to increase guideline-concordant treatment of insomnia in primary care

Principal Investigators: Jeffrey Linder, MD, MPH, and Jason Doctor, PhD

Long-term opioid therapy for noncancer pain lacks evidence of benefit and is associated with substantial harm, yet opioid prescribing remains higher than pre-opioid crisis levels despite clinical guidelines and education efforts. Trial 1 of the Application of Economics & Social Psychology to Improve Opioid Prescribing Safety (AESOPS-1) aims to reduce unnecessary opioid prescribing in primary care by leveraging behavioral insights — evidence-based social and psychological interventions designed to influence clinician decision-making.

Study Design

AESOPS-1 is a randomized trial of primary care clinics in Illinois and California comparing a behavioral intervention with usual care.

All clinics receive opioid guideline education, while intervention clinics also receive electronic health record-based nudges, including accountable justification for prescriptions, clinician precommitment to address high-risk prescribing and a PainTracker tool to broaden discussions of pain.

Outcomes

The primary outcome is the change in weekly morphine milligram equivalents (MME) prescribed, with a secondary outcome examining changes in the proportion of patients receiving at least 50 daily MME. Outcomes will be analyzed using an intent-to-treat, difference-in-differences mixed-effects regression model over an 18-month intervention period and a six-month follow-up.

Overall, AESOPS-1 will assess whether scalable, EHR-integrated behavioral interventions can effectively reduce noncancer opioid prescribing and lower population-level exposure to opioids in primary care.

Related Publication

• The Protocol of the Application of Economics and Social Psychology to Improve Opioid Prescribing Safety Trial 1 (AESOPS-1): Electronic Health Record Nudges

Principal Investigators: Jeffrey Linder, MD, MPH, and Julie Szymczak, PhD

Antibiotic overuse threatens patient safety through adverse drug effects and accelerates the development of antibiotic resistance. While prior interventions using behavioral economics and communication strategies have modestly improved antibiotic prescribing, they have largely focused on clinician behavior during office visits. Patient-directed approaches have typically been generic, educational and visit-based, leaving a critical gap in addressing patient care-seeking behaviors — particularly among patients with high levels of inappropriate antibiotic use.

This project advances a patient-centered stewardship approach that directly engages patients outside of clinical encounters using theory-informed behavioral strategies.

The proposed mixed-methods research includes three aims:

1. Retrospective controlled cohort studies of more than 1.5 million adult and pediatric patients across two health systems will categorize high antibiotic use as appropriate, partially appropriate, potentially appropriate or inappropriate.
2. Qualitative interviews with patients who exhibit high inappropriate antibiotic use will identify social and behavioral drivers of care-seeking and antibiotic use, guided by the Theoretical Domains Framework.
3. Insights from the first two aims will inform the development and pilot testing of a patient-centered behavioral nudge in a controlled randomized trial.

Outcomes

The primary outcome will be the number of inappropriate antibiotic prescriptions over eight months, with secondary outcomes assessing non-antibiotic prescribing, feasibility, acceptability, spillover effects and unintended consequences.

Overall, this project seeks to better understand and influence patient behaviors related to antibiotic use, laying the foundation for a larger randomized trial. By improving patient engagement, diagnostic accuracy and antibiotic prescribing quality, the work aims to enhance patient safety and reduce the spread of antibiotic-resistant bacteria.

Principal Investigator: Jeffrey Linder, MD, MPH

This project combines the rehabilitation medicine expertise at the University of Pittsburgh/UPMC with the primary care, patient-reported outcomes and implementation expertise at Northwestern to evaluate a system of screening older adults for physical function impairments using the PROMIS PF patient-reported outcome; referring screen-positive patients (PROMIS PF at 40 or greater) to physical therapy; and having physical therapists deliver PRIMA, an evidence-based intervention to preserve or improve physical function.

Study Aims

The project includes three aims:

• Aim 1: Engage with a wide range of clinical and community advisors, PCORI Health System Implementation Initiative leaders and health information staff to identify implementation facilitators and barriers. We will map these facilitators and barriers to evidence-based implementation strategies to optimize real-world PRIMA delivery. By the conclusion of Aim 1, we will have developed a formal implementation plan for screening and referral to a PRIMA-based PT intervention.
• Aim 2: After the design of the implementation plan, we will survey at least 10 multidisciplinary clinicians (physicians, physical therapists) and 10 older adults at each site who were not involved with Aim 1 with the Acceptability of Intervention Measure (AIM), Intervention Appropriateness Measure (IAM) and Feasibility of Intervention Measure (FIM). Iterative refinements to the implementation plan will be made in response to feedback.
• Aim 3: Implement and evaluate proactive, primary care-based functional status assessment, appropriate PT referrals for older adults with reduced physical function and delivery of the PRIMA PT-based evidence-based intervention.

Outcomes

The main clinical outcome will be the 12-week PROMIS PF, and the main implementation outcomes will include reach, fidelity and patient acceptability.