BACKGROUND:The COVID-19 pandemic accelerated the adoption of digital health technologies, including telehealth, electronic medical records, and patient-facing applications. These tools maintained access to care during periods of restriction but exposed gaps in workforce readiness, patient trust, digital literacy, and equity. Despite widespread implementation, limited evidence exists on how patients and healthcare providers experienced this transition and how these experiences influenced digital health adoption. OBJECTIVE:To compare the experiences of patients and healthcare providers in XXXXXXXX during the pandemic, identify key barriers and enablers of digital health adoption, and inform sustainable models of care for future disruptions. METHODS:A qualitative study was conducted between March 2023 and September 2024. Semi-structured Zoom interviews were undertaken with 20 patients diagnosed with COVID-19 and 21 healthcare providers involved in pandemic care in New South Wales and Victoria. Data was analyzed using reflexive thematic analysis and interpreted through the Technology Acceptance Model and Diffusion of Innovations framework to examine technology adoption, communication, and system preparedness. RESULTS:Four themes captured patient and provider experiences. First, rapid digitization shifted healthcare-seeking behavior, with many patients delaying in-person care due to safety concerns. Second, providers reported limited readiness, increased digital workload, and workflow disruption. Third, differences in digital literacy and misaligned expectations undermined trust, communication, and diagnostic confidence in virtual settings. Finally, long-term sustainability requires hybrid models of care, interoperable infrastructure, and supportive governance. Overall, the findings highlight the dual nature of crisis-driven digital transformation: while innovation accelerated rapidly, systemic and relational vulnerabilities in care delivery were also exposed. CONCLUSION:Digital care was essential for maintaining access during the pandemic; however, rapid deployment alone cannot ensure safety, equity, or quality of care. Sustainable integration requires coordinated investment in interoperable infrastructure, workforce capability, and patient-centered design to support hybrid delivery models and strengthen preparedness for future healthcare disruptions.
Abstract:BACKGROUND: Pharmacogenetics is increasingly recognized as essential for optimizing drug therapy and reducing preventable adverse drug events. However, applying pharmacogenetic data in real-world clinical decisions remains challenging. Traditional rule-based clinical decision support systems often fall short. Machine learning-based clinical decision support systems offer a more dynamic solution, combining pharmacogenetic variants, patient-specific clinical data, medication history, and guidelines to generate personalized treatment recommendations. Abstract:OBJECTIVE: This scoping review examined the extent and nature of evidence on integrating machine learning into clinical decision support systems for pharmacogenetics, emphasizing clinical implementation. By focusing on articles where tools have been integrated into clinical workflows, this review highlights progress and persistent gaps in translating pharmacogenetics-informed machine learning models into everyday clinical practice. Abstract:METHODS: A comprehensive search across multiple databases identified studies published from January 2015 to September 2025. Eligible studies included any design reporting on machine learning-based clinical decision support systems incorporating pharmacogenetic data to support therapeutic decisions in clinical settings. Abstract:RESULTS: Of 1262 records screened, 7 studies met inclusion criteria. These studies implemented machine learning-based clinical decision support systems integrating pharmacogenetic data to guide drug therapy in clinical environments. While varying in design, setting, and implementation maturity, most systems demonstrated potential benefits, such as reducing preventable adverse drug events, improving prescribing accuracy, or enhancing workflow integration. Two of the included studies evaluated PGx ML-CDSS tools in live clinical or trial workflows. Abstract:CONCLUSION: Despite growing interest and model development, clinical implementation of machine learning-based pharmacogenetic clinical decision support tools remains limited. Most systems remain rule-based, while a few integrate machine learning with pharmacogenetic data for more personalized and adaptive decision-making. This review underscores the gap between theoretical model development and real-world application. To advance the field, implementation science research is needed to evaluate usability, clinical workflow integration, and patient outcomes. Real-world evidence is essential to unlock the full potential of machine learning-based clinical decision support in pharmacogenetics.
Abstract:BACKGROUND: Excessive alerts, difficult-to-use interfaces, and inefficient workflows increase cognitive load and time in the electronic health record (EHR), contributing to clinician burnout. Ordering workflows are a high-friction component of the EHR experience. Order friction (OF) is a vendor-derived metric quantifying ordering burden as a count of required manipulations and interruptions. Abstract:OBJECTIVES: The objective of this study is to (1) evaluate and characterize OF across inpatient medication orders within a large multihospital pediatric health system and (2) develop, implement, and assess a targeted intervention to reduce OF. Abstract:METHODS: This quasi-experimental pre-post analysis took place in a three-hospital pediatric health system using Epic Systems. We analyzed vendor-supplied OF data from January to December 2024 for all inpatient medication orders. A multidisciplinary workgroup used total order changes (TOC = changes per order [CPO] × order volume) to identify high-volume, high-friction medications, and then implemented a pediatric hospital medicine preference list, available to all inpatient providers. We prepopulated dosing, frequency, and "as needed" indications for 15 medications, built as 45 order variants mapping to 25 orderables. Using an uncontrolled pre-post design, we compared pre- and postintervention CPO at the orderable level (22 paired orderables; 25 in a sensitivity analysis), using a paired Wilcoxon signed-rank test. Abstract:RESULTS: Among 22 paired orderables, median CPO decreased from 3.20 (interquartile range [IQR]: 2.41-3.68) to 0.39 (IQR: 0.18-1.04; mean: 3.15-0.79; W = 253, n = 22, p < 0.001). Sensitivity analysis including 25 orderables yielded similar results (median: 3.21-0.40). At the system level, combining institutional and preference-list ordering for these medications, volume-weighted CPO decreased from 3.06 to 2.72 as adoption of the optimized list rose from 9.6% to 15.8%. Abstract:CONCLUSION: Vendor-derived OF metrics can identify and optimize high-friction pediatric medication orders. Prepopulating clinically appropriate defaults substantially reduced CPO, suggesting a feasible approach to reducing EHR-related burden in pediatric inpatient care.
Abstract:BACKGROUND: Large language models (LLMs) are rapidly transforming medical education, yet their performance in Allergy/Immunology remains insufficiently characterized. Furthermore, concerns regarding accuracy, consistency, and sensitivity to input format persist. Abstract:OBJECTIVES: This study aimed to evaluate and compare the accuracy and response consistency of three leading LLMs-ChatGPT-5, Gemini 2.5, and Grok 4-on Allergy/Immunology United States Medical Licensing Examination (USMLE) Step 1-style questions under different prompt conditions. Abstract:METHODS: Thirty-five USMLE Step 1-style questions were selected. Questions were presented to each model in two formats: single-question prompts and a combined prompt containing all questions. Fifteen trials were conducted for each format per model. Performance was assessed using mean accuracy, and variability was measured using Shannon entropy. Mixed-effects models tested the effects of model, prompt condition, and question difficulty. Abstract:RESULTS: Overall accuracy differed significantly (p < 0.001), with Gemini (80.7%) and Grok (80.5%) achieving higher mean scores than ChatGPT (74.3%). Single-item prompts yielded superior performance, with Grok (93.1%) and Gemini (90.9%) demonstrating the highest accuracy. Transitioning to a combined prompt significantly reduced accuracy for all models. Accuracy also decreased with increasing question difficulty for all models. Grok demonstrated superior reliability, maintaining the lowest overall response entropy, whereas ChatGPT exhibited the highest variability. Abstract:CONCLUSION: On Allergy/Immunology Step 1-style questions, Gemini and Grok demonstrated higher accuracy than ChatGPT, although their overall accuracies remained approximately 81%. Grok offered the most consistent performance. All models demonstrated substantial sensitivity to prompt complexity and inherent performance limitations. These findings underscore the importance of prompt optimization and support the supplementary role of these models in medical education.
Abstract:BACKGROUND: Ambient clinical documentation tools are increasingly used to reduce administrative burden and improve provider experience. However, evidence describing their effects at scale across large, multisite health systems remains limited. Abstract:OBJECTIVES: The objective of this study is to evaluate the effect of Dragon Ambient eXperience (DAX) Copilot on documentation efficiency, note composition practices, and provider well-being across a multistate health system. Abstract:METHODS: Due to rolling registration and training, we conducted a retrospective index-date-aligned pre-post analysis of objective electronic health record activity metrics, supplemented by a postimplementation survey. Physicians and advanced practice providers in ambulatory settings between June 2024 and November 2025 were included. Epic Signal data were analyzed for each provider over the 8 months before and after DAX activation. Metrics included active time in notes per appointment, proportion of note content originating from manual typing, copy/paste, or conventional voice recognition, and average note length. Surveys were administered 45 days postactivation to assess satisfaction, burnout, intent to remain with the organization, and work-life balance. Abstract:RESULTS: A total of 210 providers met the inclusion criteria, defined as generating ≥25% of total note content using ambient voice technology and having at least 16 months of Signal Data (8 months pre- and 8 months postintervention). Compared with the preimplementation period, active note time decreased significantly (5.54-4.09 min per appointment; 26.2% reduction; p < 0.001). Despite a +28.8% increase in note length, manual documentation behaviors declined substantially, including typing (-51.7%), copy/paste (-42.7%), and conventional voice recognition (-67.8%). Survey responses (n = 233; 26.4% response rate) indicated improvements across domains, with most respondents reporting reduced burnout, increased job satisfaction, improved work-life balance, and a strong preference to continue using DAX. Abstract:CONCLUSION: Deployment of ambient artificial intelligence documentation technology at scale was associated with improvements in documentation efficiency, substantial reductions in manual input, and improved provider well-being, supporting its potential to meaningfully alleviate documentation burden.
Background:Health-related social needs (HRSN) significantly influence health outcomes, yet healthcare organizations face persistent challenges tracking referrals to community-based organizations and interpreting referral outcomes across fragmented clinical and social care systems. Prior studies report low referral fulfillment rates, but much of this evidence derives from single organizations, manual data collection, or incentivized documentation workflows, limiting insight into how information infrastructure shapes what is observable at scale. Objective:This study aimed to evaluate how HRSN referral initiation, documented outcomes, and time-to-fulfillment are captured when electronic health record (EHR) data and closed-loop HRSN platform data are integrated through a statewide health information exchange, and to identify informatics strengths and limitations affecting cross-system observability of referral processes. Methods:We conducted a retrospective observational study of 1,628 adult patients from a federally qualified health center and three health systems. Linked EHR and HRSN platform data were used to examine referral initiation, documented fulfillment, and time to fulfillment across resource types. Results:Patients received an average of two referrals, with approximately 80% addressing basic needs. Fewer than 3% were documented as fulfilled, reflecting constrained documentation workflows rather than confirmed absence of service delivery. Short-term needs were more likely to transition to documented outcome states and had shorter documented time-to-fulfillment, whereas long-term needs-particularly housing-exhibited wide variability and documentation patterns consistent with workflow-driven closure behavior. Conclusion:Instead of examining referral fulfillment as a performance metric, the findings suggest how socio-technical infrastructure, workflow design, and documentation incentives affect what becomes observable in secondary use of HRSN data.
Objectives:This study evaluated the effect of a Spatial Awareness Integrated Electronic Health Record (SAI-EHR) prototype on efficiency, usability, and cognitive workload among intensive care nurses performing standardized documentation tasks, compared with a traditional linear flowsheet-style interface (LID-EHR). Methods:A randomized, two-period crossover simulation was conducted with 36 intensive care unit (ICU) nurses who completed two standardized scenarios. Primary outcomes were documentation efficiency (time to complete a fixed task), usability (System Usability Scale [SUS]), and cognitive workload (NASA Task Load Index [NASA-TLX]). Secondary outcomes included accuracy and Technology Acceptance Model (TAM) measures of Perceived Usefulness (TAM-PU) and Behavioral Intention (TAM-BI). Analyses used linear mixed-effects models. Within-participant effect sizes were calculated using Cohen's d. Results:SAI-EHR reduced documentation time by 177 seconds (27% faster; p < 0.001, d = 0.88) per task. NASA-TLX workload scores were 51.3% lower with SAI-EHR compared with the LID-EHR (18.44 vs. 37.82; p < 0.001, d = 0.73). Accuracy was 6.3% higher with SAI-EHR (98.70 vs. 92.86%; p < 0.001, d = 0.68). SUS and perceived usefulness scores favored SAI-EHR but did not differ significantly (SUS: 77.99 vs. 71.60; TAM-PU: 3.72 vs. 3.38). TAM-BI was substantially higher for SAI-EHR (25.5% increase; Δ = 0.96; 95% confidence interval [CI]: 0.55-1.36; p < 0.001, d = 0.72). Lower cognitive workload was associated with a stronger intention to adopt (r = -0.64; p < 0.001). Conclusion:An SAI-EHR prototype mirroring ICU room layout enabled faster, more accurate documentation with lower perceived workload than a linear flowsheet interface. Stated intention to adopt was higher and inversely associated with workload; differences in SUS and TAM perceived usefulness scores favored the spatial interface but did not reach statistical significance in this sample. Findings support spatial awareness integration as a workload-reducing design principle, while formal usability and usefulness comparisons warrant a larger-sample evaluation.
Objective:In our recent study we showed that GPT's ability to perform OPS-code extraction from operational reports was equivalent to coding neurosurgeons. In this study we aim to evaluate the effect of context enhancement on GPT-5's code extraction abilities. Methods:We provided OpenAIs GPT-5 with 100 operational reports (OR) of patients that underwent meningioma surgery. A chat prompt was generated instructing GPT to generate the correct OPS coding. Five groups were formed and provided with different context-enhancing modalities: 1. No additional context (GPT-5s), 2. the current OPS-catalogue (GPT-5o), 3. specified rules on Meningioma coding (GPT-5r), 4. code-triggering sample phrases based on 50 additional ORs (GPT-5e) and 5. all enhancements combined (GPT-5c). We analyzed code-extraction abilities, mistakes and hallucinations. Results:Non context enhanced GPT-5s showed the lowest rate of correct coding (44%) and highest number of hallucinations (105) and total mistakes (132). GPT-5o showed identical accuracy to GPT-5s (44%, p = 1.0), but fewer hallucinations (51, p = 0.002) and mistakes (78, p = 0.008). All other models were significantly superior to GPT-5s and GPT-5o in accuracy (GPT-5r 79%, GPT-5e 70%, GPT-5c 86%, p < 0.001), hallucinations (GPT-5r 8, GPT-5e 2, GPT-5c 6, p < 0.001) and mistakes (GPT-5r 21, GPT-5e 35, GPT-5c 14, p < 0.001). Highest coding accuracy was achieved by GPT-5c (GPT-5c vs GPT-5r, p = 0.143, GPT-5c vs GPT5e p = 0.008). GPT-5r and GPT-5e performed equally regarding accuracy, GPT-5r produced fewer mistakes (p = 0.034), while GPT-5e hallucinated less (p = 0.057). Conclusion:We show, that specific context enhancement significantly improves GPT-5s ability to perform OPS-code extraction from operational reports, while significantly lowering mistake- and hallucination rates.
Objectives:Hospital artificial intelligence (AI) is increasingly embedded in electronic health record workflows, cloud inference pipelines, imaging, medication review, triage, early warning, documentation, and operational management. This paper proposes a risk-tiered governance framework and implementation pathway for hospital AI applications that are integrated into clinical and operational workflows. Methods:We conducted a narrative review and framework synthesis of peer-reviewed evidence, reporting guidelines, regulatory and policy sources, implementation studies, and applied governance case reports relevant to hospital AI. Sources were used according to their evidence type: systematic and scoping reviews identified recurring risks and barriers; empirical studies and trials illustrated bounded implementation patterns; case reports informed organizational design; and reporting or regulatory frameworks informed documentation, validation, change control, and oversight elements. Results:The proposed framework has four components. First, a use-case inventory tags AI applications by decision influence and workflow coupling. Second, a six-domain risk taxonomy addresses clinical safety, privacy and data security, ethics and fairness, transparency and explainability, system stability and resilience, and compliance and accountability. Third, a four-tier risk scheme links the level of oversight to potential harm, automation, reversibility, and operational coupling. Fourth, a governance architecture assigns responsibilities to an AI governance committee, clinical owners, risk control functions, and independent assurance. A patient safety-oriented lifecycle pathway is proposed across initiation, local validation, shadow mode, controlled go-live, monitoring, change control, retirement, and organizational learning. To support feasible adoption, we also provide barriers, first steps, and a maturity-based ramp-up model. Conclusion:Hospital AI governance is best framed as proportional lifecycle control rather than one-time go-live approval. The proposed framework is an adaptable scaffold, not a universal mandate. It is intended to help clinical informatics leaders, quality and safety teams, and hospital executives prioritize oversight according to risk, resources, and organizational maturity.
Objectives:Clinical documentation consumes substantial clinician time, potentially detracting from patient care. Generative artificial intelligence (AI) may support drafting discharge summaries and patient referral documents, but feasibility in non-Western-language oncology settings using real-world electronic health record (EHR) data remains insufficiently evaluated. This study assessed feasibility in a Japanese cancer hospital using an enterprise AI system. Methods:Medical records from 61 consenting adult patients at Chiba Cancer Center were analyzed. Although the plan aimed at comprehensive EHR data, actual input was limited to extractable text (physician notes, nursing records); structured laboratory data and imaging, endoscopy, and pathology reports were not directly used, and existing summaries and external referrals were excluded to avoid information leakage. Data were converted to JavaScript Object Notation; GaiXer generated 31 discharge summaries and 30 referral documents. Four evaluators scored them; ≥80/100 was an exploratory threshold for draft-level practical utility. Feedback drove one refinement cycle. Results:Generated documents scored approximately 60 to 70. A score ≥80 was reached by 9 of 31 discharge summaries in each evaluation; for referrals, none reached the threshold initially, whereas 5 of 30 did after refinement. Discharge summary scores did not substantially improve; referral scores did. Raw percent agreement among three nonphysician evaluators was high, although chance-corrected agreement varied. Wilcoxon signed-rank tests showed no significant change for discharge summaries (p = 0.866) but significant improvement for referrals (p = 0.006). Conclusion:This feasibility study suggests AI may support drafting these documents in a secure environment using real-world Japanese EHR data, although the generated documents did not consistently reach the predefined threshold for draft-level utility. Findings should not be interpreted as demonstrating workload reduction or maximum performance under ideal data conditions. Future studies should evaluate larger datasets, multiple institutions and models, blinded evaluations, actual editing time, clinician acceptance, and workflow impact.
Background:Interactive digital self-management (IDSM) interventions may help individuals with chronic respiratory disease self-manage their condition. Objective:This review aimed to summarize the characteristics and effectiveness of IDSM interventions for this population. Methods:Reviewers systematically screened titles, abstracts, and full texts of randomized controlled trials (RCTs) examining applications or wearable devices for IDSM in adults with chronic respiratory diseases. Study and intervention characteristics and outcomes were extracted. Meta-analysis was conducted in a subset of studies. Results:The search generated 109,900 unique records and 34 relevant RCTs were identified. These studies evaluated 2,320 individuals with asthma, 1,766 with chronic obstructive pulmonary disease (COPD; FEV1 % predicted 36 to 69), and 33 with cystic fibrosis. In asthma, median (range) intervention length was 12 (1-52) weeks. In COPD, median (range) intervention length was 24 (2-48) weeks. In COPD, IDSM interventions showed greater effect than control for health status (COPD Assessment Test score mean difference [95% CI]: -1.6 [-3.2, 0.0], p < 0.05, very low quality of evidence), physical activity (steps per day: 937.9 [95% CI: 311.5, 1,564.4], p < 0.05, very low quality of evidence), and exercise capacity (six-minute walk test distance mean difference [95% CI]: 5.6 [1.7, 9.4], p < 0.05, low quality of evidence). In asthma, no statistically significant effect was observed for asthma control or quality of life and only one RCT in cystic fibrosis was identified. Conclusion:Interventions were heterogeneous and yielded very low to moderate quality evidence. In COPD, IDSM interventions were associated with statistically significant improvements in health status, physical activity, and exercise capacity; however, findings should be interpreted cautiously due to low or very low certainty of evidence. Evidence in asthma and cystic fibrosis remains limited and inconsistent. At present, IDSM tools should be considered adjunctive options rather than replacements for established self-management strategies.
Objective:This study aimed to examine perspectives on challenges, design requirements, and implementation considerations for developing a granular, patient-facing consent tool that enables meaningful control over sensitive health data sharing in real-world healthcare settings. Methods:Guided by the National Science Foundation I-Corps framework and conducted in collaboration with Shift-a national consortium for patient-driven sharing of health information-we conducted semi-structured interviews with 16 expert stakeholders representing clinical, health IT, governance, and policy domains. Transcripts were analyzed using thematic analysis with combined deductive and inductive coding to identify key challenges, user needs, and system constraints. Results:Participants most frequently emphasized the need for patient-centered, trauma-informed user experiences that recognize prior experiences of stigma, discrimination, surveillance, or harm associated with disclosure of sensitive health information, and that use plain language and education to support meaningful consent (75%). Lack of trust, stigma, and workflow burden were major barriers to sensitive data sharing (63%), alongside technical limitations in EHRs that prevent reliable segmentation and enforcement of preferences (56%). Granular, computable consent-allowing patients' choices-was viewed as essential (50%), though legal and policy fragmentation complicates implementation (50%). Additional considerations included proxy and adolescent-aware configurations (44%) and interoperability standards such as Fast Healthcare Interoperability Resources (FHIR) Consent and Data Segmentation for Privacy (19%). Conclusion:Findings suggest that effective consent management requires alignment between patient-centered design, technical computability, and governance. Enforceable consent integrated into routine workflows and paired with transparency may reduce stigma-driven nondisclosure while supporting clinical safety. Granular, patient-directed consent can translate privacy expectations into practice. Embedding computable consent within interoperable standards and trauma-informed design offers a path toward more trustworthy, equitable, and safe health data sharing.
Background:Inappropriate antibiotic prescribing contributes to antimicrobial resistance. Antimicrobial stewardship (AMS) programs increasingly rely on electronic health record (EHR)-based clinical decision support (CDS) alerts to guide evidence-based care, yet poorly designed alerts can increase cognitive burden and contribute to burnout. Usability testing offers a structured approach to improving alert design and adoption. Objectives:This study aimed to evaluate and iteratively refine interruptive CDS alerts developed to support appropriate antibiotic prescribing for pediatric acute respiratory tract infections (ARTIs) within an urban academic health system as part of a larger implementation project. Methods:Twelve providers participated in four waves of virtual usability testing using scripted clinical scenarios reflecting common workflows. Participants interacted with three clinical alerts and one design-focused scenario. Quantitative outcomes included task effectiveness, efficiency, satisfaction, and perceived difficulty. Qualitative feedback was gathered via think-aloud techniques and synthesized using a "rainbow spreadsheet." Iterative modifications were implemented between waves until thematic saturation was achieved. Results:Task completion across scenarios was high (83.3%). Effectiveness and efficiency improved over waves, and satisfaction increased when alerts embedded actionable tasks such as ordering tests, adding diagnoses, or switching antibiotics. Participants emphasized the importance of clear, concise text and intuitive defaults. Including links to evidence-based guidelines improved trust. Alert refinements reduced cognitive load and improved workflow alignment. Conclusion:Rapid, iterative usability testing substantially improved the clarity and workflow integration of interruptive CDS alerts. Incorporating usability evaluation into CDS stewardship efforts can enhance adoption, reduce alert fatigue, and strengthen implementation of evidence-based practices.
Background:China's hospital-integrated Internet Hospital model embeds virtual outpatient care within hospital information systems, pharmacy workflows, and regulatory oversight structures. However, encounter-level evidence on real-world utilization, operational efficiency, online prescribing, and repeat use remains limited. Objectives:The objective of this study is to describe the first-year utilization, service efficiency, prescribing safety, and continuity-of-care potential of a hospital-integrated Internet Hospital in Henan province, China. Methods:We conducted a retrospective observational study of all online consultation orders generated from November 1, 2023, to October 31, 2024, at a large tertiary hospital. Deidentified data from consultation logs, prescribing modules, and pharmacy systems were descriptively analyzed. Key measures included order outcomes, consultation type and modality, patient-level repeat use, response time, online prescribing, prescription review outcomes, and ratings. Results:A total of 90,006 online consultation orders were generated; 54,389 (60.4%) were completed, 32,720 (36.4%) were cancelled, and 2,897 (3.2%) expired because payment was not completed. Most cancellations occurred before physician acceptance. Text-based consultations accounted for 83,833 orders (93.1%). Virtual Consultations accounted for 78,313 orders (87.0%) and did not permit prescribing; Virtual Follow-up Consultations accounted for 11,693 orders (13.0%) and had a higher completion rate (74.3 vs. 58.4%). Among 49,724 unique patients, 16,554 (33.3%) generated two or more orders. Median response time was 93.4 minutes (interquartile range: 12.4-345.5). Prescriptions were generated in 491 orders (0.55%); 359 patients paid for medication orders, and 118 prescriptions required physician modification after initial automated review. No legally restricted controlled drugs were prescribed online. Ratings were submitted for 5,820 orders (6.5%). Conclusion:The platform primarily served as a hospital-governed digital entry point for consultation and triage, with a controlled follow-up pathway for selected prescription renewal. It demonstrated feasibility but also highlighted preacceptance cancellations, long-tailed response times, limited older-adult use, and expert-heavy workload as priorities for optimization.
Objectives:Patient portal use has steadily increased across most populations. Prior, now dated, studies indicated lower adoption rates among Spanish- versus English- speaking patients. This study compared patient portal activation and use patterns between Spanish- and English-speaking patients. Methods:This retrospective cohort study was conducted at three North Texas health systems using the MyChart patient portal (Epic Systems Co.) and included patients ≥ 18 years with ≥1 completed clinician encounter between April 5, 2021 and April 4, 2022. The primary activation outcome was the baseline MyChart account activation rate. The secondary activation outcome was the MyChart account activation rate within the following year among patients without an account at baseline. The primary use outcome was the rate of patients logging in. Secondary use outcomes included rates of results review, notes review, and message initiation in the following year. We also evaluated the rates of proxy account use and mobile app use. We fit multivariable logistic regression models adjusting for health system, age, sex, comorbidity count, and the number of prior-year encounters. Results:Spanish speakers represented 128,338 of 1,550,220 (8.3%) patients. Spanish speakers had lower odds of having an activated account at baseline (adjusted odds ratio [aOR], 0.39 [0.39-0.40]) or activating one in the following year (aOR, 0.68 [0.65-0.71]). Spanish speakers also had lower odds of logging in (aOR, 0.62 [0.61-0.63]), reviewing results (aOR, 0.79 [0.76-0.81]), reviewing notes (aOR, 0.87 [0.84-0.89]), or sending messages (aOR, 0.41 [0.40-0.42]). More Spanish than English speakers used the mobile app (59 vs. 50%). There were intersite differences in the rate of proxy account use. Conclusion:Given lower levels of portal activation and use among Spanish-speaking patients, strategies are needed to identify and address barriers to activation and use. Qualitative studies could delineate these barriers and potential mitigating strategies.
Background Despite advances in interoperability, patient-supplied medical histories commonly use paper-based forms. Dissatisfaction with this inefficiency has led to the "Kill the Clipboard" initiative from the Centers for Medicare & Medicaid Services to advance digital alternatives. During the coronavirus disease 2019 (COVID-19) pandemic, patient-controlled vaccination record sharing demonstrated that Quick Response (QR) codes and digital data standards provide an effective option. Building on this, several nations are using three standards to securely view and exchange medical data: Fast Healthcare Interoperability Resources, the International Patient Summary, and SMART Health Links. Objectives This case report explores the use of QR codes and medical data standards in patient-mediated workflows to share medical histories, describing implementation experiences from four such implementations. Methods We describe standards and developer community experience with global initiatives and implementations across the United States, Canada, and the Hajj pilgrimage, which all support the use of QR codes in patient-mediated data exchange. Results We report on technical readiness, patient adoption, and reported improvements to emergency preparedness using QR codes to share patient summaries. Conclusion Advancing patient-mediated exchange with these technologies aligns with initiatives to "Kill the Clipboard." Preliminary evidence suggests this novel approach can better inform care. These four case studies demonstrate the viability of this approach and the lessons learned from this digital transition.
Objectives:This study aimed to describe the usage and perceptions of physicians and nurses in primary and specialty care on the acceptability, usability, and quality of artificial intelligence (AI)-generated draft replies to patient portal messages. Methods:We conducted a pilot study of 80 users across seven clinics within a large academic health system in Pennsylvania. Epic's Augmented Response Technology (ART) used a large language model to present clinicians with draft replies to patient messages. To measure usage, we calculated the frequency with which replies were initiated with ART-generated drafts, as well as the overlap between drafts and replies sent to patients. To measure acceptability, usability, and quality, we developed and disseminated a 15-item survey. Results:Voluntary pilot participants included 80 users (36 physicians, 25 nurses [registered nurses/licensed practice nurses], 10 nurse practitioners, 9 medical assistants). The mean rate of starting with an ART draft was 20.2%, and 40% of replies exhibited little to no editing of the draft. Of the 52 survey respondents, 66% (n = 33) agreed that AI-generated drafts were useful, and 46% (n = 23) agreed that ART improved the quality of replies to patient messages. Across survey questions, a higher percentage of nurses reported favorable opinions of the technology than did physicians. In free-response questions, the most common themes were that the message content was helpful (28.3%, n = 13) and that it reduced perceived effort or time (23.9%, n = 11), but that drafts occasionally contained incorrect or inappropriate content (34.9%, n = 15) or too often suggested an appointment (27.9%, n = 12). Conclusion:In this voluntary pilot study, AI-generated draft replies to patient portal messages were generally rated as acceptable and useful, with nurses and primary care clinicians more often reporting acceptance of the technology than physicians or specialty care clinicians. Scaling efforts should continue to monitor acceptability and effects on cognitive burden, in addition to safety.
Background Ambulatory central line-associated bloodstream infections (A-CLABSIs) cause significant morbidity, are costly, and are a relatively new patient safety target. To calculate A-CLABSI rates, the total ambulatory line-day denominator must be known; however, few hospitals can measure this accurately. Access to this data is vital to the quality improvement process. Objectives The project aimed to automate the collection of central line (CL) information from the electronic health record (EHR) and the calculation of ambulatory CL days to within a 5% variance rate compared with the manual process to monitor and improve A-CLABSI rates. Methods Existing documentation processes were analyzed, and appropriate fields were determined for CL data capture. Multiple revisions were needed to hard-wire accurate EHR documentation in identified areas. An algorithm was developed to identify and electronically track CL insertions/removals and calculate days. We completed 9,533 manual chart reviews, correcting retrospective data errors in documentation (EID), and implemented a process to address EID in real time to ensure accurate data and sustainability. Results The EHR automated process identified an adjusted average of 98% of patients with a CL compared with the manual process with a variance of 0.3% between June 2024 and May 2025. The inpatient and ambulatory variance rates were 3.50 and 5.2%, respectively. The automated process identified an adjusted average of 96% total CL days compared with the manual process, with an average variance rate of 3.2% . Inpatient and ambulatory CL day calculations had variance rates of 0.7 and 4.5% respectively, compared with the manual process. The average EID rate for all CLs was 12.8%. Conclusion Through real-time electronic data capture, a more efficient and sustainable process for maintaining accurate CL data and calculating A-CLABSI rates was developed.
Objective:Intravenous (IV) and oral acetaminophen have equivalent bioavailability and effectiveness, but IV formulations are significantly more expensive. Clinical decision support (CDS) tools can encourage cost-effective prescribing. We evaluated the effectiveness and acceptability of different CDS strategies on prescribing patterns of acetaminophen. Methods:We conducted a two-arm randomized controlled trial with a historical comparator at our academic medical center, including 23,676 acetaminophen orders for 13,139 adult inpatients between July 2023 and November 2024. In the preintervention period, providers received a single-click interruptive alert when ordering IV acetaminophen. In the postperiod, providers were randomized to either a multiclick interruptive alert or no alert. The primary outcome was alert effectiveness; secondary outcomes included acceptability and durability. Results:The mean patient age was 57 years; 64% were female, 62% White, and 87% English-speaking. Most orders were placed for medicine (32%), primarily by residents (61%). Preintervention, 9.4% of orders were for IV acetaminophen, compared with 11.2% in the multiclick interruptive group and 12.1% in controls (p = 0.153). Among 624 interruptive alerts followed by orders, 24.5% switched to oral. Interrupted time series analysis showed no significant change in IV order trends. Conclusion:Approximately one-quarter of multiclick alerts prompted a switch to oral acetaminophen, but overall effectiveness was no greater than the control or single-click alert. Findings suggest that increasing alert disruptiveness may add cognitive burden without meaningful gains in effectiveness. A multiclick interruptive alert did not significantly change IV acetaminophen prescribing. Effective CDS strategies may require minimally disruptive designs integrated with education.
