Abstract Background De-implementation—reducing low-value or harmful care—is critical but difficult in clinical practice. Clinical decision support (CDS) “nudges” in electronic health records (EHRs) aim to promote guideline-concordant deprescribing, but effects are inconsistent. In a pragmatic randomized controlled trial across a large health system, we tested a suite of EHR-based CDS nudges to support Choosing Wisely-aligned deprescribing of glycemic medications in older adults with type 2 diabetes. Although a prior pilot showed modest improvement in guideline concordance (5.1%), the full trial found no significant changes in prescribing; this process evaluation examines clinicians’ comments on alerts to explain why. Methods We conducted a mixed-methods process evaluation of comments within EHR-based alerts from a null-result RCT that promoted Choosing Wisely deprescribing for older adults with type 2 diabetes. Among 66,634 alerts firing across EHR encounters (December 2016-July 2023), providers commented on 764 (1.2%). Two researchers independently coded comments using reflexive thematic analysis, identifying four themes (three negative). Exploratory logistic and multinomial regressions examined predictors of commenting, valence, and themes among acknowledged firings, adjusting for patient, provider, and encounter factors. Results Thematic analysis of comments revealed three barriers to deprescribing: (1) disagreement with Choosing Wisely guidelines (308 comments, e.g., perceived low overtreatment risk); (2) workflow misalignment (203 comments, e.g., wrong provider responsibility); and (3) patient preferences (69 comments). Logistic regression showed multiple concurrent OPAs reduced action odds by 31.6% (OR 0.684, 95% CI 0.560–0.835); comments were 2.57 times more likely to be negative than positive (OR 2.565, 95% CI 1.637–4.018). Disparities in engagement were found, with female providers, patients, and socially vulnerable individuals less likely to comment. Conclusion This process evaluation demonstrates scalable real-time feedback for clinical decision support refinement in de-implementation, with regressions identifying context-specific predictors. Provider disagreement, alert firings misaligned to workflows, and patient resistance hinder effectiveness. Future work should refine clinical decision support design to address complexity, enhance guideline explainability to build provider concordance, align with provider roles and workflows, and include patient-centered approaches. Trial registration The NYU School of Medicine Institutional Review Board (i17-01308) approved the trial, which has the clinicaltrials.gov ID NCT04181307 (https://clinicaltrials.gov/study/NCT04181307), with a first record date of November 26, 2019.

ABSTRACT:Manufacturing and use of pharmaceuticals is responsible for nearly 20% of US health care's greenhouse gas footprint. As highly effective weight-loss medications such as incretin mimetics become widely prescribed, it is important to understand environmental implications. More than 33.5 million Americans have tried incretin mimetic medications (IMMs), and nearly 30 million are projected to be consistent users by 2030. This article examines the broader potential implications of glucagon-like peptide-1 and other IMMs on climate change. We conducted a preliminary and speculative carbon footprint of IMMs using a life cycle assessment approach. Our findings suggest widespread IMM use could reduce greenhouse gas emissions by decreasing caloric consumption, food production, and health care activities, leading to a maximum estimated reduction of 760 kg CO2e/person/year. This reduction would be greater than the environmental benefits of switching to electric vehicles or adopting a vegetarian diet. This study highlights the need for more research into potential environmental benefits of IMMs.

INTRODUCTION:Linking electronic health record (EHR) use to care quality may offer insights into potential interventions improving guideline adherence and closing care gaps. We examine how EHR metadata can measure cognitive load in primary care providers during statin prescribing and identify cognitive load points in EHR workflows associated with guideline-concordant statin initiation. METHODS:We retrospectively extracted 2024 data from EHR primary care encounters from a large academic health system. We identified adult patients who met the criteria for statin initiation and calculated their atherosclerotic cardiovascular disease (ASCVD) risk scores. Cognitive load metrics were derived from EHR metadata. Logistic regressions evaluate associations between cognitive load and statin initiation, adjusting for patient covariates and provider fixed effects. Gradient-boosted forests and Shapley Additive explanations (SHAP) values were used to identify key EHR events and cognitive load patterns associated with statin initiation. RESULTS:Longer encounter duration was associated with increased likelihood of statin initiation, whereas more time spent per EHR event was associated with a decreased likelihood. Nonlinear associations were observed for loop count and distinct event count: predicted initiation probability decreased with increasing loop count to 93.9 loops, then increased beyond this threshold. For distinct events, initiation probability increased up to approximately 18 events and declined at higher counts. In a gradient-boosted decision tree model, average event time was the strongest predictor (72.2% relative contribution). Additional positive predictors included time spent reviewing lab results and on suggested medication order sets. Order list modification and looping back to it were negatively associated with statin initiation. DISCUSSION:EHR metadata can associate cognitive load with appropriate clinical behavior, revealing nonlinear associations between cognitive load and statin initiation rates. This work suggests opportunities to optimize EHR systems to reduce cognitive burden and support clinical decision-making. Connecting cognitive load to prescribing behavior generates hypotheses about how workflow adjustments and enhanced decision support might improve guideline adherence and patient care through prospective evaluation.

Generative AI is increasingly used in speculative design to support creative exploration, yet most systems function as tools for generating content rather than as social facilitators that coordinate collaborative imagination. We present Muse, a conversational AI designed to act as a social facilitator in participatory speculative design workshops exploring healthcare futures. Muse guides groups through world-building, persona creation, and artifact development while maintaining narrative continuity and reflective pacing alongside human facilitators. The system was iteratively developed through internal testing, piloted with ten workshop participants, and deployed at scale at a medical informatics conference with 43 participants across five groups. All groups completed the full activity sequence, producing speculative worlds, personas, and artifacts. Post-session surveys indicated that participants found Muse valuable, easy to use, and fun, with most agreeing it added to their design process. Observational and participant-reported challenges included a communicator bottleneck that filtered group deliberation, structured optimism that constrained the speculative space, and inherent limitations of chat-based facilitation. This work contributes the design, iterative refinement, and deployment of Muse, positioning social facilitation as a novel and distinct role for conversational AI in collaborative design.

Lung cancer screening (LCS) with low-dose computed tomography (CT) remains underused in the US, partly because of incomplete smoking history documentation in electronic health records (EHRs) and limited time for shared decision-making in primary care. To determine whether a patient-facing, EHR-integrated tool combined with clinician-facing clinical decision support improves the identification of LCS-eligible patients and the ordering of low-dose CT compared with clinician-facing tools alone. This pragmatic, unstratified, randomized clinical trial with parallel groups was conducted from March 29, 2024, to March 28, 2025, at primary care clinics at University of Utah Health and New York University Langone Health. Adults aged 50 to 79 years with a documented smoking history, an active patient portal account, and a primary care visit in the preceding year were included. Study 1 enrolled patients with uncertain LCS eligibility (10 to 19 pack-years, unknown pack-years, or missing quit date); study 2 enrolled patients with documented eligibility (20 or more pack-years and currently smoking or quit smoking within 15 years). The control included the clinician-facing Decision Precision+ tool (preventive care reminders and a shared decision-making tool). The intervention included the Decision Precision+ tool as well as the MyLungHealth tool, which collected detailed smoking history (study 1) and delivered personalized education and risk/benefit information (studies 1 and 2) via the patient portal in English and Spanish. The primary outcomes were the proportion of patients newly identified as eligible for LCS (study 1) and low-dose CT ordering rates (study 2) over 12 months. Analyses used intention-to-treat mixed-effects logistic regression. There were 31 303 randomized participants, including 26 729 in study 1 (13 144 [49.2%] female; 13 580 [50.8%] male; median [IQR] age, 62 [55-69] years) and 4574 in study 2 (2230 [48.8%] female; 2344 [51.2%] male; median [IQR] age, 63 [56-69] years). In study 1, the MyLungHealth tool increased new LCS eligibility identification (635 of 13 412 [4.7%] vs 308 of 13 317 [2.3%]; adjusted odds ratio, 2.19; 95% CI, 1.99-2.42; P < .001). In study 2, low-dose CT ordering was higher in the intervention arm (474 of 2312 [20.5%] vs 434 of 2262 [19.2%]; adjusted odds ratio, 1.16; 95% CI, 1.04-1.30; P = .008). In this randomized clinical trial, integrating a patient-centered tool into primary care EHR workflows increased the identification of patients eligible for LCS and the ordering of low-dose CTs. The relative increases in these primary outcomes were substantial, but absolute increases were more modest. Research on more intensive interventions is warranted to evaluate their ability to further improve LCS screening. ClinicalTrials.gov Identifier: NCT06338592

Background: Clinical decision support (CDS) tools embedded in electronic health records in the form of integrated clinical prediction rules provide a potentially effective intervention to reduce inappropriate antibiotic prescribing for acute respiratory infections. However, their effectiveness has been limited by workflow barriers and low adoption by health care providers. Nurses are well positioned to implement evidence-based protocols using CDS tools. In a multicenter randomized controlled trial, a nurse-led implementation strategy for acute respiratory infection integrated clinical prediction rules was evaluated for use in primary care and urgent care settings. Objective: This study aimed to examine nurse and nurse leader perspectives on the sustainability of an electronic health record-integrated CDS tool for antibiotic stewardship and explored factors influencing its potential long-term integration into ambulatory nursing practice beyond the clinical trial. Methods: We interviewed 22 nurses and nurse leaders from 37 clinics across 3 academic medical centers that participated in the clinical trial. Two semistructured interview guides, one for nurses and one for nursing leadership, were developed to understand the barriers and facilitators to implementing a decision aid tool for nurses and to elicit challenges specific to nursing interactions with the CDS tool. Interviews were recorded and transcribed. Using thematic content analysis and iterative coding, our team collaboratively identified emerging themes related to sustainability and refined the results with consensus. Results: Five themes emerged: (1) importance of staffing stability and capacity, (2) impact of dedicated clinic resource availability, (3) variable nurse readiness with CDS-guided clinical care, (4) influence of openness to change and a nurse-supportive clinic culture, and (5) ongoing need for training and support. Specific recommendations for future actions were also noted. Conclusions: Our findings revealed specific barriers and facilitators to the sustainability of a CDS tool from the nursing perspective that can inform further implementation of nurse-led delegation protocols in the ambulatory setting. Future solutions should consider mapping physical workflows, scheduling specific to nurse visits, continuing education, and treating cough and sore throat as 2 distinct processes.

Importance:Lung cancer screening (LCS) with low-dose computed tomography (CT) remains underused in the US, partly because of incomplete smoking history documentation in electronic health records (EHRs) and limited time for shared decision-making in primary care. Objective:To determine whether a patient-facing, EHR-integrated tool combined with clinician-facing clinical decision support improves the identification of LCS-eligible patients and the ordering of low-dose CT compared with clinician-facing tools alone. Design, Setting, and Participants:This pragmatic, unstratified, randomized clinical trial with parallel groups was conducted from March 29, 2024, to March 28, 2025, at primary care clinics at University of Utah Health and New York University Langone Health. Adults aged 50 to 79 years with a documented smoking history, an active patient portal account, and a primary care visit in the preceding year were included. Study 1 enrolled patients with uncertain LCS eligibility (10 to 19 pack-years, unknown pack-years, or missing quit date); study 2 enrolled patients with documented eligibility (20 or more pack-years and currently smoking or quit smoking within 15 years). Interventions:The control included the clinician-facing Decision Precision+ tool (preventive care reminders and a shared decision-making tool). The intervention included the Decision Precision+ tool as well as the MyLungHealth tool, which collected detailed smoking history (study 1) and delivered personalized education and risk/benefit information (studies 1 and 2) via the patient portal in English and Spanish. Main Outcomes and Measures:The primary outcomes were the proportion of patients newly identified as eligible for LCS (study 1) and low-dose CT ordering rates (study 2) over 12 months. Analyses used intention-to-treat mixed-effects logistic regression. Results:There were 31 303 randomized participants, including 26 729 in study 1 (13 144 [49.2%] female; 13 580 [50.8%] male; median [IQR] age, 62 [55-69] years) and 4574 in study 2 (2230 [48.8%] female; 2344 [51.2%] male; median [IQR] age, 63 [56-69] years). In study 1, the MyLungHealth tool increased new LCS eligibility identification (635 of 13 412 [4.7%] vs 308 of 13 317 [2.3%]; adjusted odds ratio, 2.19; 95% CI, 1.99-2.42; P < .001). In study 2, low-dose CT ordering was higher in the intervention arm (474 of 2312 [20.5%] vs 434 of 2262 [19.2%]; adjusted odds ratio, 1.16; 95% CI, 1.04-1.30; P = .008). Conclusions and Relevance:In this randomized clinical trial, integrating a patient-centered tool into primary care EHR workflows increased the identification of patients eligible for LCS and the ordering of low-dose CTs. The relative increases in these primary outcomes were substantial, but absolute increases were more modest. Research on more intensive interventions is warranted to evaluate their ability to further improve LCS screening. Trial Registration:ClinicalTrials.gov Identifier: NCT06338592.

Importance Medication nonadherence is present in nearly half of patients with hypertension but is underrecognized in clinical care. Data linkages between electronic health records and pharmacies have created opportunities for scalable assessment of medication adherence at the point of care. Objective To test the effectiveness of a multicomponent intervention that identified patients with uncontrolled hypertension and medication nonadherence using linked electronic health record-pharmacy data combined with team-based care to address adherence barriers. Design, Setting, and Participants TEAMLET (Leveraging Electronic Health Record Technology and Team Care to Address Medication Adherence) was a pragmatic, 2-arm, cluster randomized clinical trial conducted between October 2022 and November 2024 in 10 primary care sites in New York. The study included adults with uncontrolled hypertension and low medication adherence, defined as proportion of days covered (PDC) less than 80%. Data analysis was performed from November 2024 to January 2025. Intervention The intervention consisted of the following: (1) automated identification of patients with medication nonadherence at the time of the visit; (2) prompting of medical assistants to screen for barriers to adherence; (3) clinical decision support alerting the primary care physicians and nurse practitioners to barriers to adherence; and (4) adherence discussion between the primary care physician or nurse practitioner and the patient. The comparator was usual care. Main Outcomes and Measures The primary outcome was change in PDC from baseline to 12 months. Results Among 1726 patients (mean [SD] age, 67.2 [13.9] years; 887 [51.4%] female), the mean (SD) baseline PDC was 33.2% (30.5%) overall (32.4% [30.4%] in the intervention group and 34.0% [30.6%] in the control group). The mean (SD) PDC at 12 months was 51.1% (39.5%) for the intervention group and 53.1% (39.6%) for the control group. No difference was found in the change in PDC from baseline to 12 months between the intervention and control groups (mean [SD] absolute change in PDC, 18.5 [41.1] vs 18.2 [40.9] percentage points, respectively; adjusted difference, -0.15 percentage point; 95% CI, -4.06 to 3.76 percentage points). Change in systolic blood pressure and patients who became adherent (PDC >= 80%) at 12 months were also similar between groups. Conclusions and Relevance In this pragmatic trial, an intervention that combined team-based primary care with automated identification of patients with antihypertensive medication nonadherence did not lead to improvements in adherence or blood pressure.

Uncontrolled hypertension is common and frequently related to inadequate adherence to prescribed medications, resulting in suboptimal blood pressure control and increased healthcare utilization. Although healthcare providers have the opportunity to improve medication adherence, they may lack the tools to address adherence at the point of care. This study aims to assess the usability of a digital tool designed to improve medication adherence and blood pressure control among patients with hypertension who are not adherent to therapy. By evaluating usability, the study seeks to refine the tool's design, underscore the role of technology in managing hypertension, and provide insights to inform clinical decisions. We performed qualitative usability testing of an electronic health record (EHR)-integrated intervention with medical assistants (MAs) and primary care providers (PCPs) from a large integrated health system. Usability was assessed with these end-users using the “think aloud” and “near live” approaches. This evaluation was guided by two frameworks: the End-User Computing Satisfaction Index (EUCSI) and the Technology Acceptance Model (TAM). Interviews were analyzed using a thematic analysis approach. Thematic saturation was reached after usability testing was performed with 10 participants, comprising 5 PCPs and 5 MAs. The study identified several strengths within the content, format, ease of use, timeliness, accuracy, and usefulness of the tool, including the user-friendly content presentation, the usefulness of adherence information, and timely alerts that fit into the workflow. Challenges centered around alert visibility and specificity of information. Leveraging the two conceptual frameworks (TAM and EUCSI) to test the usability of the medication adherence tool was helpful. The tool's several strengths and opportunities for improvement were found. The resulting suggestions will be used to support the enhancement of the design for optimal implementation in a clinical trial.

Patients with neurodiverse conditions, including autism spectrum disorder and intellectual disabilities, face persistent barriers in healthcare settings due to provider bias, communication breakdowns, and environmental sensitivity. These challenges can contribute to negative care experiences and lower utilization of essential services by this population. This project proposes a Virtual Reality (VR) training program to educate clinicians on delivering inclusive, high-quality care for patients with neurodiverse conditions. Using generative AI tools, we developed patient and provider personas and categorized underlying common needs into five evidence-based modules: communication, environmental sensitivity, time and pacing, bias and assumptions, and caregiver inclusion. Based on these needs, we propose training modules which incorporate dynamic avatars, real-time feedback, motion tracking, and haptic feedback to simulate realistic exam room encounters and train providers in scenario-based learning. The training emphasizes user-centered design, integrating the needs of individuals with neurodiverse conditions and their clinicians. This work contributes a scalable, immersive training model that aims to reduce bias, improve clinical communication, and enhance accessibility in healthcare delivery.

BackgroundLaypeople have easy access to health information through large language models (LLMs), such as ChatGPT, and search engines, such as Google. Search engines transformed health information access, and LLMs offer a new avenue for answering laypeople’s questions. ObjectiveWe aimed to compare the frequency of use and attitudes toward LLMs and search engines as well as their comparative relevance, usefulness, ease of use, and trustworthiness in responding to health queries. MethodsWe conducted a screening survey to compare the demographics of LLM users and nonusers seeking health information, analyzing results with logistic regression. LLM users from the screening survey were invited to a follow-up survey to report the types of health information they sought. We compared the frequency of use of LLMs and search engines using ANOVA and Tukey post hoc tests. Lastly, paired-sample Wilcoxon tests compared LLMs and search engines on perceived usefulness, ease of use, trustworthiness, feelings, bias, and anthropomorphism. ResultsIn total, 2002 US participants recruited on Prolific participated in the screening survey about the use of LLMs and search engines. Of them, 52% (n=1045) of the participants were female, with a mean age of 39 (SD 13) years. Participants were 9.7% (n=194) Asian, 12.1% (n=242) Black, 73.3% (n=1467) White, 1.1% (n=22) Hispanic, and 3.8% (n=77) were of other races and ethnicities. Further, 1913 (95.6%) used search engines to look up health queries versus 642 (32.6%) for LLMs. Men had higher odds (odds ratio [OR] 1.63, 95% CI 1.34-1.99; P<.001) of using LLMs for health questions than women. Black (OR 1.90, 95% CI 1.42-2.54; P<.001) and Asian (OR 1.66, 95% CI 1.19-2.30; P<.01) individuals had higher odds than White individuals. Those with excellent perceived health (OR 1.46, 95% CI 1.1-1.93; P=.01) were more likely to use LLMs than those with good health. Higher technical proficiency increased the likelihood of LLM use (OR 1.26, 95% CI 1.14-1.39; P<.001). In a follow-up survey of 281 LLM users for health, most participants used search engines first (n=174, 62%) to answer health questions, but the second most common first source consulted was LLMs (n=39, 14%). LLMs were perceived as less useful (P<.01) and less relevant (P=.07), but elicited fewer negative feelings (P<.001), appeared more human (LLM: n=160, vs search: n=32), and were seen as less biased (P<.001). Trust (P=.56) and ease of use (P=.27) showed no differences. ConclusionsSearch engines are the primary source of health information; yet, positive perceptions of LLMs suggest growing use. Future work could explore whether LLM trust and usefulness are enhanced by supplementing answers with external references and limiting persuasive language to curb overreliance. Collaboration with health organizations can help improve the quality of LLMs’ health output.

Objective: This study aimed to explore the effects of demographics and social determinants of health (SDOH) on remote patient monitoring (RPM) utilization and blood pressure (BP) improvement. Methods: A secondary data analysis of an RPM program for hypertension in federally qualified health centers (FQHCs). This observational study, guided by the Digital Health Equity-focused Implementation Research framework (DH-EquIR), used linear mixed effect models to investigate the effects of demographics and mean area deprivation index (ADI) on utilization and BP change among patients with three months of home BP monitoring data. Utilization was measured as a count of missed days per week, indicating days without transmitted BP readings. Results: There were 105 participants, averaging 55.4 years old, with 64.8% Black or African American race, and 33.4% of Hispanic/Latino ethnicity. On a scale of 1-10, with 1 indicating the lowest level of deprivation, the mean ADI NY state rank was 3.3. As weeks on RPM progressed, participants experienced significant increases in missed days per week overall. For every point increase in ADI NY state rank, missed days per week increased by 0.24 (p < 0.05). Regardless of ADI, for every increasing week on RPM, the systolic BP value decreased by 0.55 mmHg (p < 0.0001). Conclusion: This DH-EquIR-guided RPM study, among the first in FQHCs, found minimal RPM usage differences by demographics and SDOH. Overall, participants in this sample effectively utilized RPM and showed improvement in BP, including in participants living in areas of high ADI NY state rank and inconsistent RPM utilization.

BackgroundAmong the alternative solutions being tested to improve access to genetic services, chatbots (or conversational agents) are being increasingly used for service delivery. Despite the growing number of studies on the accessibility and feasibility of chatbot genetic service delivery, limited attention has been paid to user interactions with chatbots in a real-world health care context. ObjectiveWe examined users’ interaction patterns with a pretest cancer genetics education chatbot as well as the associations between users’ clinical and sociodemographic characteristics, chatbot interaction patterns, and genetic testing decisions. MethodsWe analyzed data from the experimental arm of Broadening the Reach, Impact, and Delivery of Genetic Services, a multisite genetic services pragmatic trial in which participants eligible for hereditary cancer genetic testing based on family history were randomized to receive a chatbot intervention or standard care. In the experimental chatbot arm, participants were offered access to core educational content delivered by the chatbot with the option to select up to 9 supplementary informational prompts and ask open-ended questions. We computed descriptive statistics for the following interaction patterns: prompt selections, open-ended questions, completion status, dropout points, and postchat decisions regarding genetic testing. Logistic regression models were used to examine the relationships between clinical and sociodemographic factors and chatbot interaction variables, examining how these factors affected genetic testing decisions. ResultsOf the 468 participants who initiated a chat, 391 (83.5%) completed it, with 315 (80.6%) of the completers expressing a willingness to pursue genetic testing. Of the 391 completers, 336 (85.9%) selected at least one informational prompt, 41 (10.5%) asked open-ended questions, and 3 (0.8%) opted for extra examples of risk information. Of the 77 noncompleters, 57 (74%) dropped out before accessing any informational content. Interaction patterns were not associated with clinical and sociodemographic factors except for prompt selection (varied by study site) and completion status (varied by family cancer history type). Participants who selected ≥3 prompts (odds ratio 0.33, 95% CI 0.12-0.91; P=.03) or asked open-ended questions (odds ratio 0.46, 95% CI 0.22-0.96; P=.04) were less likely to opt for genetic testing. ConclusionsFindings highlight the chatbot’s effectiveness in engaging users and its high acceptability, with most participants completing the chat, opting for additional information, and showing a high willingness to pursue genetic testing. Sociodemographic factors were not associated with interaction patterns, potentially indicating the chatbot’s scalability across diverse populations provided they have internet access. Future efforts should address the concerns of users with high information needs and integrate them into chatbot design to better support informed genetic decision-making.

Introduction:Linking EHR use to care quality offers insights for interventions to improve guideline adherence and close care gaps. We examine how EHR metadata can measure cognitive load in primary care providers during statin prescribing and identify points of cognitive load in the EHR workflow. Methods:EHR primary care encounter data from a large academic health system in 2024 were retrospectively extracted. We identified adult patients who met the criteria for statin initiation and calculated their ASCVD risk scores. Cognitive load metrics were derived from EHR metadata. Logistic regressions evaluate associations between cognitive load and statin initiation, adjusting for patient covariates and provider fixed effects. Gradient-boosted forests and SHAP values identified key EHR events and cognitive load associated with the initiation of statin therapy. Results:Longer encounter duration increased the likelihood of statin initiation, whereas more time spent per EHR event decreased it. Non-linear effects were observed for loop count and distinct event count: the probability of initiation decreased with increasing loop counts up to 93.9 loops, then increased beyond this threshold. For distinct events, the initiation probability increased up to approximately 18 events and declined at higher counts. In a gradient-boosted decision tree model, average time per event was the strongest predictor (72.2% relative contribution). Additional positive predictors included the time spent reviewing lab results and on suggested medication order sets. Modifying the order list and looping back to it were negatively associated with statin initiation. Discussion:EHR metadata can associate cognitive load with appropriate clinical behavior, finding nonlinear relations between cognitive load and statin initiation rates. This work highlights the need to optimize EHR systems to reduce cognitive burden and support clinical decision-making. Connecting cognitive load to prescribing behavior gives insight into how workflow adjustments and enhanced decision support can improve adherence to guidelines and patient care.

BACKGROUND:Medication nonadherence is a common issue among patients with hypertension. Healthcare professionals often overlook medication nonadherence due to limited tools, time constraints, and competing demands. Integrating pharmacy medication fill data into electronic health records (EHRs) presents an opportunity to enhance medication adherence measurement and monitoring in real-time. This study identified facilitators and barriers to addressing adherence to antihypertensive medications by Medical Assistants (MAs), Registered Nurses (RNs), and Primary Care Providers (PCPs) in primary care settings. METHODS:We conducted a qualitative study with, 15 healthcare professionals (5 MAs, 5 RNs, and 5 PCPs) caring for patients with hypertension. Semi-structured interviews, guided by the Consolidated Framework for Implementation Research (CFIR), explored barriers and facilitators related to screening and addressing medication non-adherence during primary care clinical encounters. Thematic analysis and deductive coding were used to analyze the data. RESULTS:Four major themes emerged: motivation, work infrastructure, capability, and opportunity. MAs and PCPs were motivated to discuss medication adherence and build relationships. Capability varied; RNs were confident in their counseling skills based on their training and patient trust, and PCPs described adherence counseling as part of their role, particularly through motivational interviewing. Work infrastructure presented structural hurdles due to RN workflow limitations and MA role constraints. Opportunity to address non-adherence were constrained by tight schedules and competing clinical demands during brief visits. CONCLUSIONS:RNs and PCPs felt capable in their ability to address medication adherence but cited time and competing demands as significant barriers; conversely, MAs reported motivation but were limited by their role. These findings suggest opportunities for effective management of medication adherence in practice settings through a more coordinated strategy across multiple healthcare professionals. TRIAL REGISTRATION:ClinicalTrials.gov; NCT05349422; https://clinicaltrials.gov/ct2/show/NCT05349422.

Overuse and misuse of antibiotics is an urgent health care problem and one of the key factors in antibiotic resistance. Validated clinical prediction rules have shown effectiveness in guiding providers to an appropriate diagnosis and identifying when antibiotics are the recommended choice for treatment.We aimed to study the relative ability of registered nurses using clinical prediction rules to guide the management of acute respiratory infections in a simulated environment compared with practicing primary care physicians.We evaluated a case-based simulation of the diagnosis and treatment for acute respiratory infections using clinical prediction rules. As a secondary outcome, we examined nursing self-efficacy by administering a survey before and after case evaluations. Participants included 40 registered nurses from three academic medical centers and five primary care physicians as comparators. Participants evaluated six simulated case studies, three for patients presenting with cough symptoms, and three for sore throat.Compared with physicians, nurses determined risk and treatment for simulated sore throat cases using clinical prediction rules with 100% accuracy in low-risk sore throat cases versus 80% for physicians. We found great variability in the accuracy of the risk level and appropriate treatment for cough cases. Nurses reported slight increases in self-efficacy from baseline to postcase evaluation suggesting further information is needed to understand correlation.Clinical prediction rules used by nurses in sore throat management workflows can guide accurate diagnosis and treatment in simulated cases, while cough management requires further exploration. Our results support the future implementation of automated prediction rules in a clinical decision support tool and a thorough examination of their effect on clinical practice and patient outcomes.

OBJECTIVE:To examine whether patient sociodemographic and clinical characteristics and prior interactions with the healthcare system were associated with opening patient portal messages related to cancer genetic services and beginning services. STUDY SETTING AND DESIGN:The trial was conducted in the University of Utah Health (UHealth) and NYU Langone Health (NYULH) systems. Between 2020 and 2023, 3073 eligible primary care patients aged 25-60 years meeting family history-based criteria for cancer genetic evaluation were randomized 1:1 to receive a patient portal message with a hyperlink to a pretest genetics education chatbot or information about scheduling a pretest standard of care (SOC) appointment. DATA SOURCES AND ANALYTIC SAMPLE:Primary data were collected. Eligible patients had a primary care visit in the previous 3 years, a patient portal account, no prior cancer diagnosis except nonmelanoma skin cancer, no prior cancer genetic services, and English or Spanish as their preferred language. Multivariable models identified predictors of opening patient portal messages by site and beginning pretest genetic services by site and experimental condition. PRINCIPAL FINDINGS:Number of previous patient portal logins (UHealth average marginal effect [AME]: 0.32; 95% CI: 0.27, 0.38; NYULH AME: 0.33; 95% CI: 0.27, 0.39), having a recorded primary care provider (NYULH AME: 0.15; 95% CI: 0.08, 0.22), and more primary care visits in the previous 3 years (NYULH AME: 0.09; 95% CI: 0.02, 0.16) were associated with opening patient portal messages about genetic services. Number of previous patient portal logins (UHealth AME: 0.14; 95% CI: 0.08, 0.21; NYULH AME: 0.18; 95% CI: 0.12, 0.23), having a recorded primary care provider (NYULH AME: 0.08; 95% CI: 0.01, 0.14), and more primary care visits in the previous 3 years (NYULH AME: 0.07; 95% CI: 0.01, 0.13) were associated with beginning pretest genetic services. Patient sociodemographic and clinical characteristics were not significantly associated with either outcome. CONCLUSIONS:As system-level initiatives aim to reach patients eligible for cancer genetic services, patients already interacting with the healthcare system may be most likely to respond. Addressing barriers to accessing healthcare and technology may increase engagement with genetic services.

Background:De-implementation-reducing low-value or harmful care-is critical but often difficult in practice. Nudges via clinical decision support (CDS) tools in electronic health records aim to promote guideline-concordant care, but their effectiveness is mixed. In a randomized trial, we tested CDS nudges to support deprescribing glycemic medications in older adults, aligned with Choosing Wisely guidelines. Despite prior success elsewhere, the intervention had limited impact. The current study evaluated potential reasons why the EHR-based nudges to encourage guideline-based, relaxed glycemic control for older adults with Type 2 Diabetes were not effective in influencing clinician behavior. Methods:We conducted a retrospective cohort analysis of EHR data from 67,412 alerts issued to clinicians, promoting different types of glycemic control, including reducing metformin, switching from non-metformin medications to metformin, and discontinuing medication. Comments left by providers on 779 of those firings were coded and thematically analyzed by two authors. Logistic and multinomial logistic regressions were performed to understand the contexts behind the lack of nudge effectiveness at the alert, encounter, patient, and physician levels. Results:Out of 67,412 alerts, providers commented in only 1.15% of cases. When they did, they were about 10.7% more likely to act on the alert, but comments were mostly negative (3.28 times more likely). Feedback highlighted three themes: disagreement with guidelines (most common), poor alert fit in workflow, and patient reluctance to change medications. Logistic regressions showed providers were less likely to act on alerts with multiple triggers and more likely to leave negative comments. Multinomial models linked rejection themes to patient and medication traits, noting less rejection related to workflow in patients with limited life expectancy. Disparities in engagement were found, with female providers, patients, and socially vulnerable individuals less likely to comment. Conclusion:These findings highlight barriers to de-implementation via CDS. Provider disagreement, misaligned alerts, and patient resistance hinder effectiveness. Low engagement and negative feedback suggest nudges alone may not change behavior without integration into routines. Engagement variation stresses the need for tailored strategies. Future work should refine nudge design to address complexity, align with provider roles, and include patient-centered approaches. Trial registration:The NYU School of Medicine Institutional Review Board (i17-01308) approved the trial, which has the clinicaltrials.gov ID NCT04181307 (https://clinicaltrials.gov/study/NCT04181307), with date of first record on November 26, 2019.

OBJECTIVE:To evaluate trends in telemedicine utilization overall and across clinical specialties, providing insights into its evolving role in health care delivery. STUDY DESIGN:This retrospective cross-sectional study analyzed 1.9 million telemedicine video visits from a large academic health care system in New York City between 2020 and 2023. The data, collected from the health care system's electronic health records, included telemedicine encounters across more than 500 ambulatory locations. METHODS:We used descriptive statistics to outline telemedicine usage trends and compared telemedicine utilization rates and evaluation and management characteristics across clinical specialties. RESULTS:Telemedicine utilization peaked during the COVID-19 pandemic, then declined and stabilized. Despite an overall decline, 2 non-primary care specialties (behavioral health and psychiatry) experienced continued growth in telemedicine visits. Primary care and urgent care visits were mainly characterized by low-complexity visits, whereas non-primary care specialties witnessed a rise in moderate- and high-complexity visits, with the number of moderate-level visits surpassing those of low complexity. CONCLUSIONS:The findings highlight a dynamic shift in telemedicine utilization, with non-primary care settings witnessing an increase in the complexity of cases. To address future demands from increasingly complex medical cases managed through telemedicine in non-primary care, appropriate resource allocation is essential.