Publications

Achieving digital health equity and proper use of identity credentials is crucial as reliance on electronic modalities increases. Proxy access-now increasingly referred to as shared access-is a widely available functionality that offers identity credentials to care partners who assist loved ones in navigating the electronic care delivery demands of patients with complex care needs. However, adoption of these tools has been hindered by complicated user interfaces and low awareness.Drawing on frameworks and principles rooted in human-centered design (HCD), we conducted an evaluation of a multisite quality improvement study designed to increase the awareness and adoption of shared access to patient portals for older adults and their care partners. Through feedback gathered from key informants, we identified barriers to the adoption of materials created for the parent quality improvement project, and synthesize additional implementation strategies from informant feedback to improve shared access.We employed the Double Diamond Model (DDM) of HCD to guide our research. The DDM includes engaging a diverse group of community partners-older adults, care partners, health care system leaders, communications professionals-through focus groups and individual interviews. Our process involved identifying pain points related to registration for shared access, then synthesizing these insights through inductive coding and affinity mapping to generate solutions.An analysis of our community partner feedback revealed several themes, including the necessity for simplified patient portal registration, standardized terminology about shared access, and clear messaging strategies. A step-by-step video tutorial was developed as a prototype. The prototype was then implemented at a partner health system and received positive feedback, suggesting its potential for broader use.These findings emphasize the importance of involving "end users" (patients, care partners, health care system leaders, communications professionals) in the evaluation and implementation of digital health tools. Approaching challenges with an HCD mindset helped our team identify barriers to shared access adoption and led to the development of a tangible resource (prototype and video). This project highlights the potential for HCD to drive improvements in digital health equity.This research demonstrates a practical application of HCD methods in developing effective solutions for enhancing shared access for older adults, and all people using patient portals.

BACKGROUND: Hypertension-related deaths in the United States have sharply increased over the past decade. Understanding factors behind this concerning trend is crucial to guide public health strategies.

OBJECTIVES: This study sought to examine changes in prevalence, treatment, and control of cardiometabolic risk factors among U.S. adults with hypertension.

METHODS: Using National Health and Nutrition Examination Survey, we identified 21,822 adults with hypertension from 1999 to 2023. Trends in age-adjusted prevalence of diabetes, hyperlipidemia, obesity, smoking, and high-risk drinking, as well as in age-adjusted treatment and control rates for hypertension, diabetes, and hyperlipidemia, were assessed.

RESULTS: In adults with hypertension, prevalence of diabetes increased from 17.2% (95% CI: 14.6%-20.2%) in 1999-2000 to 27.8% (95% CI: 25.2%-30.7%) in 2021-2023. Among patients with hypertension and diabetes, the proportion receiving treatment for both conditions rose until 2009-2010 but subsequently plateaued (71.6% in 2021-2023; 95% CI: 67.5%-75.4%). A similar trend was observed for the proportion achieving control of both conditions, which remained at 31.2% in 2021-2023 (95% CI: 26.5%-36.3%). Prevalence of hyperlipidemia decreased in adults with hypertension but was still high at 73.1% (95% CI: 70.6%-75.4%) in 2021-2023. Among patients with hypertension and hyperlipidemia, treatment and control rates for both conditions increased only until around 2007; the respective rates were 50.4% (95% CI: 47.2%-53.6%) and 32.3% (95% CI: 29.5%-35.2%) in 2021-2023. Prevalence of concurrent diabetes and hyperlipidemia in adults with hypertension nearly doubled from 12.5% (95% CI: 10.6%-14.7%) in 1999-2000 to 21.3% (95% CI: 19.0%-23.8%) in 2021-2023. Among patients with hypertension, diabetes, and hyperlipidemia, the treatment rate for all 3 conditions increased until 2007-2008 and plateaued afterward (62.1% in 2021-2023; 95% CI: 56.6%-67.2%). The control rate for all 3 conditions followed a similar trend but remained low at 26.3% (95% CI: 20.8%-32.6%) in 2021-2023. Prevalence of obesity and high-risk drinking increased over time, whereas prevalence of smoking held steady.

CONCLUSIONS: The burden of cardiometabolic risk factors among U.S. adults with hypertension has increased overall, with concurrent diabetes and hyperlipidemia now affecting more than 1 in 5. Only one-quarter of adults with hypertension, diabetes, and hyperlipidemia have all 3 conditions controlled-a proportion that has not improved in more than a decade.

The current guideline-recommended clinic blood pressure (BP) measurement procedure takes nearly 10 minutes to complete and may not be feasible to implement in busy clinical practice settings. Additionally, evidence supporting the steps in the current guideline-recommended procedure is of uneven quality. A streamlined, evidence-based approach to clinic BP measurement that still produces accurate and precise BP measurements may facilitate improved hypertension diagnosis and management. We summarized the latest evidence from studies that have quantified the impact of streamlining certain steps in the BP measurement procedure on BP measurement accuracy and precision. We translated this evidence into a practical, streamlined protocol for office BP measurement in usual primary care and potentially other settings. Studies have demonstrated it is possible to reduce the rest period before measurements from 5 to 0 minutes, and the interval between measurements from 60 to 30 seconds, without compromising accuracy. Additionally, analyses of studies with replicate BP measurements performed according to clinical practice guideline recommendations showed that repeating the initial screening measurement only when the first one is ≥130/80 mm Hg optimized accuracy and efficiency. Meanwhile, using the proper cuff size, arm support, and patient positioning are critical for BP measurement accuracy and recommendations for these steps remain unchanged from current guidelines. Broad implementation of a streamlined approach would result in more efficient BP measurement without compromising accuracy or precision, thereby increasing capacity to screen, diagnose, and manage hypertension.

No study has evaluated whether associations of plasma phosphorylated tau 217 (p-tau217) with mild cognitive impairment (MCI) or dementia vary by race or hormone therapy (HT) use. We examined 2,766 cognitively unimpaired women ≥65 years randomized to HT vs placebo with 25-year follow-up. P-tau217 was associated with incident MCI/dementia (hazard ratio [HR], 2.43; 95% CI, 2.18-2.71) and each individual outcome (MCI: HR, 1.94; 95% CI, 1.72-2.20; dementia: HR, 3.17; 95% CI, 2.79-3.61). Associations between p-tau217 and dementia were stronger for women randomized to estrogen plus progestin vs placebo (HR, 4.18; 95% CI, 3.41-5.13 vs HR, 3.07; 95% CI, 2.41-3.91, respectively; P interaction=0.044) but did not vary for estrogen alone vs placebo. The combination of p-tau217 and age performed similarly in White and Black women (AUC=72.0% and 70.4%, respectively). Findings show the value of plasma p-tau217 for prediction of MCI and dementia up to 25 years in advance in older women.

BACKGROUND: Self-measured blood pressure monitoring (SMBP) is a proposed strategy to improve hypertension control, but few studies compare SMBP with automated office blood pressure (AOBP) measurements over time. Moreover, little is known about reimbursement for these services.

METHODS: We describe a quality improvement initiative in our academic internal medicine practice. Patients received validated home blood pressure (BP) devices and were instructed to check their BP twice daily for seven days each month. Devices transmitted readings seamlessly via a smart phone application that averaged weekly readings. Clinicians reviewed the data and adjusted therapy if indicated. We tracked changes in SMBP, AOBP, antihypertensive medications, and reimbursement.

RESULTS: Among 140 patients referred (mean age 57.6 years, 57.1% women, 25.7% Black), 59 completed the program. Over 6 months, the mean self-measured systolic BP decreased by 5.8 mm Hg (95% CI: -8.2 to -3.5) and the mean diastolic BP decreased by 2.8 mm Hg (95% CI: -4.2 to -1.4) and the number with BP < 130/<80 mm Hg increased by 22 percentage points (P = 0.007). However, unattended AOBP showed no change in either systolic (change: 0.1 mm Hg; P = 0.98) or diastolic (change: -0.6 mm Hg; P = 0.64) readings. Reimbursement was variable and ranged from no payment to a maximum payment of $51; when paid, the average payment was $13.81.

CONCLUSIONS: In this program, SMBP decreased over time while AOBP remained similar. Reimbursement, when received, was modest. Future work should evaluate whether using SMBP as a therapeutic target reduces cardiovascular events.

BACKGROUND: Hospitalized older adults are commonly discharged with changes to antihypertensive and glucose-lowering (cardiometabolic) medications. Though adverse drug events remain a leading cause of readmissions, there is little contemporary data on how medication discharge planning is communicated and how often medication errors occur post-discharge.

OBJECTIVE: To assess older adults' post-hospital medication use and ambulatory follow-up after receiving cardiometabolic medication changes during hospitalization.

DESIGN: Prospective cohort study from 11/2022 to 01/2024.

PARTICIPANTS: Adults aged 65 years or older from discharged home from an academic medical center with changes to pre-admission cardiometabolic medications.

MAIN MEASURES: Participants completed 7- and 90-day telephonic surveys on health status, medication use, and discharge planning. Self-report of medication use was compared to discharge summaries to identify medication errors (not initiating, not stopping, or taking incorrect dose). Multivariable regression models were used to identify characteristics associated with errors.

KEY RESULTS: The cohort included 151 participants (median [IQR] age 74 [70-78] years; 54% male; 17% Black, 82% White, 41% frail). Participants were admitted with a median (IQR) of 3 (2-4) cardiometabolic medications and discharged with a median (IQR) of 2 (1-4) medication changes. Of the 319 individual medications changed at discharge, 33% were further modified by 90 days. Participants reported comprehensive medication discharge planning for only 13% of medication changes. Though 93% of participants reported they understood the purpose of each of their medications at discharge, 39% had ≥ 1 medication errors at 7 days and 50% at 90 days. Use of ≥ 5 cardiometabolic medications was associated with higher rates of medication errors at 7 days (IRR 1.63; 95% CI 1.07-2.48) and 90 days (IRR 1.66; 95% CI 1.13-2.45).

CONCLUSIONS: Most hospitalized older adults discharged with cardiometabolic medication changes experienced medication errors or gaps in discharge planning. Steps to ensure all patients receive high-quality medication discharge planning are needed.

BACKGROUND: For most physicians trained in internal medicine (IM), residency represents the last opportunity to compare and contrast generalist versus subspecialty careers. Decisions about subspecialty practice impact physician supply and distribution in the United States. This study tested the association between randomly assigned intern year rotation duration and career choice and assessed the importance of resident characteristics versus residency experiences in predictive modeling.

METHODS: This retrospective observational study included all categorical residents graduating from a university-affiliated IM program between 2017 and 2023 (N = 338). The association between randomly assigned intern year rotation duration and career choice was evaluated with univariable and multivariable log-link Poisson regression models. "Career choice" was defined as the field in which residents were practicing one year after residency. Predictive models were built using random forest modeling to compare the relative importance of resident characteristics and residency experiences.

RESULTS: Pre-residency career interests and ultimate career choice were highly concordant (P < 0.001, Cramer's V 0.50 (95% CI: 0.41-0.51)), particularly among residents pursuing cardiology, gastroenterology, and hematology/oncology. Rotation duration and career choice were not consistently associated; in multivariable analyses, weeks on service in the intensive care unit were associated with a career in pulmonary/critical care (RR 1.40, 95% CI 1.23-1.56), but there was no association between rotation duration and career in hematology/oncology, cardiology, gastroenterology, or general medicine. Internally cross-validated predictive models revealed that pre-residency demographics and interests were highly discriminative of career choice (AUC 0.824). The addition of residency experiences did not improve model performance (AUC 0.829).

CONCLUSIONS: Pre-residency career interests are a key determinant of career choice and hold greater weight in predictive models than intern year clinical exposure.

BACKGROUND: Traditional, complementary, and integrative (TCI) medicine is an essential component of health systems worldwide, especially in low- and middle-income countries. Despite its widespread use, existing research on the safety, efficacy, and integration of TCI medicine within conventional healthcare systems is fragmented. This fragmentation highlights the urgent need for a clearly defined global research agenda to guide future studies, secure funding, and inform governance in this field.

METHODS: The Traditional, Complementary, and Integrative Medicine Unit at the World Health Organization Headquarters in Geneva, Switzerland coordinated an international research priority-setting exercise using the Child Health and Nutrition Research Initiative (CHNRI) method between June and December 2023. We invited a purposive sample of 120 experts from established academic networks to participate; 53 experts (44.16% response rate) contributed, and 34 of them scored 157 unique research ideas according to five CHNRI criteria: feasibility, effectiveness, deliverability, equity, and potential for disease burden reduction. Additionally, we performed a comparative analysis by generating research priorities using large language models (LLMs), including ChatGPT-4o, Claude 3.5, and Grok 3, and these outputs were compared with the expert-derived priorities.

RESULTS: Top-ranked research priorities focused on chronic disease management (e.g. diabetes, dyslipidemia), geriatric safety (e.g. herb-drug interactions), mental health (e.g. resilience and mood disorders), and integration of TCI into health systems. Priorities varied by income setting. Comparison with LLM-generated lists showed thematic overlap in efficacy and safety but divergence in focus, with LLMs emphasising research capacity, policy, and systems-level priorities.

CONCLUSIONS: We established a global, expert-informed research agenda to guide the future direction of TCI medicine and ensure alignment with public health needs. The comparison with LLMs highlights the complementary potential of artificial intelligence in research governance and agenda-setting.

BACKGROUND AND OBJECTIVES: Electronic health record (EHR) tools are widely used to influence prescribing behaviors. However, the application of EHR tools to deprescribing in older adults, particularly persons living with dementia, is understudied, despite the potential to ameliorate cognitive decline by targeting medications like benzodiazepines and anticholinergics. We explored the design and implementation of EHR-based tools for deprescribing using a multi-phase mixed-methods approach.

RESEARCH DESIGN AND METHODS: Within a large healthcare system, we first conducted semi-structured qualitative interviews to understand perspectives with primary care providers (PCPs) and care partners. Transcripts were analyzed using immersion/crystallization approaches to generate themes. Second, we designed potential EHR deprescribing tools, including pre-visit provider and patient messages and provider decision support, using an interdisciplinary team based on the interviews. Third, we conducted additional qualitative interviews of drafted tools. Lastly, we conducted pilot testing of the EHR tools with PCPs.

RESULTS: We conducted interviews with 16 care partners of persons living with dementia and 7 PCPs and pilot testing of tools with 12 PCPs. From qualitative interviews, we observed several key themes around the complexity of managing deprescribing, effective framing for deprescribing interventions, health-system limitations influencing deprescribing, and leveraging technology and EHR deprescribing tools. From pilot testing, we observed the acceptability and feasibility of the tools, with salient concerns including the potential for information overload and ensuring adequate personalization.

DISCUSSION AND IMPLICATIONS: This multi-phase implementation study uncovered ways to use EHR tools to overcome deprescribing barriers, with pilot-tested tools that show promise of acceptability and feasibility, warranting further testing.