Applied Clinical Informatics最新文献

In dentistry, social determinants of health (SDoH) are potentially recorded in the clinical notes of electronic dental records. The objective of this study was to examine the availability of SDoH data in dental clinical notes and evaluate natural language processing methods to extract SDoH from dental clinical notes.A set of 1,000 dental clinical notes was sampled from a dataset of 105,311 patient visits to a dental clinic and manually annotated for information pertaining to sugar, tobacco, alcohol, methamphetamine, housing, and employment. Annotations included temporality, dose, type, duration, and frequency where appropriate. Experiments were to compare extraction using fine-tuned pretrained language models (PLMs) with a rule-based approach. Performance was measured by F1-score.For identifying SDoH, the best-performing PLM method produced F1-scores of 0.75 (sugar), 0.69 (tobacco), 0.67 (alcohol), 0.42 (housing), and 0 (employment). The rule-based method produced F1-scores of 0.70 (sugar), 0.69 (tobacco), 0.53 (alcohol), 0.44 (housing), and 0 (employment). The overall difference between PLMs and rule-based methods was F1-score of 0.04 (95% confidence interval -0.01, 0.09). SDoH were relatively rare in dental clinical notes, from sugar (9.1%), tobacco (3.9%), alcohol (1.2%), housing (1.2%), employment (0.2%), and methamphetamine use (0%).The main challenge of extracting SDoH information from dental clinical notes was the frequency with which they are recorded, and the brevity and inconsistency where they are recorded. Improved surveillance likely needs new ways to standardize how SDoHs are reported in dental clinical notes.

The purpose of this systematic literature review is to critically evaluate the use of mathematical and simulation models within emergency departments (EDs) and assess their potential to improve the quality of care. This review emphasizes the critical need for quality enhancement in health care systems, with a specific focus on EDs.This review incorporates studies that have investigated the quality of care provided in ED settings, employing assorted mathematical and simulation models for adult populations. Based on the selected studies, a narrative approach was used to synthesize the findings, focusing on outcome classification, simulation, and modelling. There are six outcome dimensions: safety, effectiveness, patient-centeredness, timeliness, efficiency, and equity.This review analyzed 112 studies, uncovering a distinct focus on a set of key performance measures within ED operations, accounting for 222 instances across these studies. Measures assessing timeliness were most frequent, occurring 111 times, indicative of a strong emphasis on operational efficiency aspects such as waiting times and patient flow. A total of 75 examinations were conducted on efficiency-related measures, with a specific focus on identifying and addressing operational bottlenecks and optimizing resource utilization. On the other hand, safety, patient-centeredness, and effectiveness were not as commonly represented, with only 3, 4, and 29 instances, respectively.This review highlights the considerable potential of mathematical and simulation models to enhance ED operations, particularly regarding timeliness and efficiency. However, aspects such as patient safety, effectiveness, and patient-centeredness were underrepresented, while equity was absent across the studies, indicating a clear need for further research. These findings emphasize the importance of adopting a more thorough approach to evaluating and improving the quality of emergency care. Future research should also concentrate on refining data management practices, incorporating observational studies, and exploring various simulation tools to develop a more balanced and inclusive understanding of these models' applications.

Emergency department (ED) handoff to inpatient teams is a potential source of error. Generative artificial intelligence (AI) has shown promise in succinctly summarizing large quantities of clinical data and may help improve ED handoff.Our objectives were to: (1) evaluate the accuracy, clinical utility, and safety of AI-generated ED-to-inpatient handoff summaries; (2) identify patient and visit characteristics influencing summary effectiveness; and (3) characterize potential error patterns to inform implementation strategies.This exploratory study evaluated AI-generated handoff summaries at an urban academic ED (February-April 2024). A Health Insurance Portability and Accountability Act-compliant GPT-4 model generated summaries aligned with the IPASS framework; ED providers assessed summary accuracy, usefulness, and safety through on-shift surveys.Among 50 cases, median quality and usefulness scores were 4/5 (standard error = 0.13). Safety concerns arose in 6% of cases, with issues including data omissions and mischaracterizations. Consultation status significantly affected usefulness scores (p < 0.05). Omissions of relevant medications, laboratory results, and other essential details were noted (n = 6), and emergency medicine clinicians disagreed with some AI characterizations of patient stability, vitals, and workup (n = 8). The most common response was positive impressions of the technology incorporated into the handoff process (n = 11).This exploratory provider-in-the-loop model demonstrated clinical acceptability and highlighted areas for refinement. Future studies should incorporate recipient perspectives and examine clinical outcomes to scale and optimize AI implementation.

Electronic health record (EHR) patient portal messaging has become an essential tool for patient-clinician communication by improving accessibility to primary care. While messaging is beneficial for patients, it can increase clinicians' workloads. Female clinicians receive a greater number of EHR messaging, resulting in an increased workload.This evaluation explores the factors in clinician gender disparity in EHR messaging burden.The first phase of the evaluation included a retrospective analysis of the messages to 267 primary care clinicians in the University of Vermont Health Network (UVMHN). The second phase analyzed patient demographics and panel complexity. Statistical analysis was performed across all categories of patient care-generated messages to primary care clinicians and subsequently on all messages across the UVMHN.Female clinicians received significantly more patient-initiated medical advice request messages than their male counterparts (68.28 vs. 49.22 messages/month, p = 0.005) and spent more time managing messages (1.85 vs. 1.35 minute/day, p = 0.006). Despite this increased workload, response times remained similar between genders. Female clinicians have a higher proportion of female patients, and analysis of all messages sent across the organization demonstrated that female patient care produces more messages than male patient care (59 vs. 52 messages/female vs. male, p = 0.001). Panels size and complexity were similar for both male and female providers.These findings highlight an unequal messaging burden for female clinicians in primary care specialties of internal and family medicine, largely due to patient demographics. Patient panel complexity as defined by UVMHN and clinician full-time equivalent were similar between genders. Disparities in message volumes appear to be driven primarily by patient communication behavior differences between genders rather than differences in workload allocation. These findings likely contribute to increased burnout risk among female clinicians. Addressing this imbalance through workflow optimization and artificial intelligence-driven message triage systems may help to mitigate the burden on female clinicians and promote greater equity in primary care.

Ensuring clinician safety in health care settings is critical, particularly regarding exposure to hazardous drugs and bodily fluids, which can be carcinogenic, teratogenic, genotoxic, or cause organ toxicity at low doses. At SickKids a safety issue arose when a clinician was unknowingly exposed to hazardous bodily fluids due to inadequate communication of a patient's hazardous medication status.This clinical decision support (CDS) redesign aimed to reduce alert fatigue while ensuring timely team awareness to minimize hazardous bodily fluid exposure risk. This case study aims to explore how redesigning a CDS system addressed the dual challenge of maintaining safety communication while minimizing alert fatigue and improving workflow integration.In 2018, a biohazardous bodily fluids alert was introduced within the hospital's electronic patient record (EPR) to raise awareness. However, its frequent and disruptive nature resulted in a 0% alert action rate and 89 unactionable clinician hours over a 90-day period. Feedback collected over 42 months revealed clinician frustration and desensitization due to the alert's timing and frequency. Using a human-centered design approach, the alert was redesigned from an interruptive pop-up to a passive notification embedded within the patient's storyboard.The redesigned alert allowed clinicians to review hazardous status information without immediate interruptions, reducing workflow disruption while maintaining its critical safety function. This approach effectively balanced safety communication with clinicians' need for efficient workflows, addressing the root cause of alert fatigue.This case study highlights the importance of ongoing CDS evaluation and redesign to enhance clinician safety, minimize alert fatigue, and improve workflow integration. Future evaluations will assess the redesign's effect on personal protective equipment compliance and clinician burnout.

Hypertension is a significant contributor to cardiovascular disease, yet evidence-based blood pressure (BP) control practices are inconsistently applied. The Collaboration Oriented Approach to Controlling High Blood Pressure (COACH) is a digital clinical decision support tool designed to improve BP self-management and support clinician workflows. While the patient perspective on COACH has been evaluated in a separate study, this study evaluates organizational readiness for COACH implementation across three health systems using the Consolidated Framework for Implementation Research (CFIR).This study aimed to assess preimplementation facilitators and barriers for COACH, focusing on organizational readiness and modifiable factors influencing scalability.Qualitative interviews were conducted with 72 care team members from nine primary care clinics across three health systems using Epic or Oracle electronic health records. Data were analyzed using CFIR domains: innovation, inner setting, outer setting, individuals, and implementation process. Subdomains were rated from -2 (barrier) to +2 (facilitator).Overall, 79% of CFIR domain scores were positive, suggesting strong readiness for COACH implementation. The innovation domain scored 80% positive, highlighting COACH's user-friendly design, robust evidence base, and perceived advantages over current workflows. The inner setting domain showed 85% positive scores, driven by strong leadership, established infrastructures for patient-centered care, and high motivation for quality improvement. The outer setting domain scored 70% positive, reflecting barriers such as reimbursement policies, resource limitations, and staffing shortages. Participants noted the importance of continued leadership engagement, team-based support, and addressing workload challenges for sustainable implementation.The study demonstrates high organizational readiness for COACH, with critical barriers in reimbursement and resources that must be addressed for successful adoption. Findings underscore COACH's potential to enhance clinical decision-making and patient engagement. Future research should explore long-term impacts on care delivery and outcomes, informing broader adoption of digital health interventions in clinical practice.

The Health Level 7 (HL7) Electronic Health Record Workgroup identified home medication list reconciliation as a prime opportunity to improve patient safety and reduce clinician burden. We developed a platform-neutral, Fast Healthcare Interoperability Resources (FHIR)-enabled reference model and demonstration wireframe to articulate the concepts of an interoperable, patient-centric home medication list management ecosystem.Four principal artifacts describe the reference model: (1) a conceptual (high-level) model, (2) a data architecture (detailed) model including representations of the interactions among actors, workflows, data, and functionality, (3) a functionality (style) guide describing expected system behaviors, and (4) a high-fidelity, end-to-end wireframe. The wireframe was constructed using JavaScript, Bootstrap Studio, and FHIR to maximize code modularity, device compatibility, and interoperability.The conceptual and architecture models capture the complex interplay of actors and data occurring among healthcare providers, information systems, and patients, positioning the patient at the center of home medication list management. The style guide reflects functionality requirements. The wireframe demonstrates the use of FHIR for data interoperability while representing patient and clinician interactions that reduce burden. The wireframe accesses standardized data elements via FHIR calls to an EHR sandbox and integrates RxNorm content to improve usability and associated medication metadata. Finally, the wireframe generates a FHIR patient-reconciled medication list data package and printable lists that can be shared with the clinician to facilitate outpatient medication reconciliation.This proof-of-concept highlights the potential of FHIR to facilitate patient-facing medication list management and provides a reference framework for developers.

Electronic health records (EHRs) have revolutionized clinical practice, but clinicians and institutions have not yet fully optimized their use. Inconsistent documentation of ophthalmic test results can increase potential medicolegal risks if providers bill for tests without properly documenting clinical interpretations.To address this, we developed and implemented a logic tool in Epic (Epic Systems, Verona, Wisconsin, United States) that prompts clinicians to document diagnostic test interpretations as discrete data before closing the patient chart.We implemented a "Close Encounter Warning" using logic rules to redirect clinicians to the Imaging and Procedures section of the Epic chart for documenting test interpretations. The implementation only allows clinicians to finalize each outpatient encounter's charting as closed if the logic rules confirm that no unsigned test results remain. The logic rules were revised many times to accommodate the unique workflow of the Ophthalmology department and to consider the roles of fellows, residents, and staff who also work with encounter charting. We implemented the initial logic rule on October 23, 21 and the final iteration on February8, 22. To evaluate the impact, we compared the number of closed charts containing unresulted diagnostic tests from October 2017 to December 2024.Before we implemented the logic rules, clinicians closed an average of 897.1 charts per month with unresulted diagnostic images (median: 916, interquartile range [IQR]: 170, 5.78% of all outpatient encounters). After implementation, this number dropped to 8.3 per month (median: 8, IQR: 5.75, 0.05% of all outpatient encounters), a 108% reduction (p < 0.001).The Close Encounter Warning logic rules significantly reduced the number of Imaging and Procedure-type diagnostic tests lacking final attending signatures in the Ophthalmology department. By implementing this EHR change, we successfully minimized potential medicolegal liability for our clinicians and institution.

Traditional early warning systems (EWS) have shown uncertain efficacy in real-world settings. More recently, machine learning models like the Epic Deterioration Index (DTI) have been developed, promising greater accuracy. Recognizing the potential of DTI, but also the pervasive issue of alert fatigue with interruptive (i.e., pop-up) EWS alerts, our institution implemented a DTI-enabled EWS with passive alerts (colored icons visible in prespecified locations within the electronic health record). We hypothesized that our intervention would reduce the time to treatment for deteriorating patients.We piloted a DTI-enabled EWS in a 30-bed intermediate care unit at a large academic medical center. DTI scores, alert icons, and vital signs appeared on a custom Patient List interface. In the event of an alert, charge nurses were expected to conduct a bedside assessment and escalate care as necessary. We compared the 111-day pre- and postimplementation periods, with alert-to-action time as the primary outcome. Secondary outcomes included mortality, length of stay, ICU transfer, documentation rate, and provider acceptance.Among 301 patients with an elevated-risk score (156 pre- and 145 postimplementation), we found no significant differences in alert-to-action time (469 vs. 359 minutes before alert; p = 0.96), with provider actions typically occurring several hours before the alert in both periods. There were no significant differences in mortality (10.3% vs. 13.1%; p = 0.56), length of stay (15.7 vs. 12.8 days; p = 0.23), or ICU transfer (8.3% vs. 6.2%; p = 0.63). Charge nurses documented acknowledgment of the alert in 18.6% of cases, and acceptance was poor. Most nurses expressed a preference for interruptive alerts and more prominent DTI display locations.In this single-unit pilot, passive DTI-enabled EWS alerts did not improve time to intervention or clinical outcomes. High-risk DTI scores often occurred after clinical deterioration had already been recognized.

Chemoprophylaxis reduces the risk of hospital-acquired venous thromboembolism (VTE), but is not reliably ordered. Our institution created a clinical decision support (CDS) interruptive alert to remind clinicians to order VTE chemoprophylaxis when it is missing for qualifying inpatients. Unfortunately, this alert has required repeated modifications to ensure accurate logic, and continues to generate negative feedback from users.This study aimed to describe multiple failures in the development and postdeployment optimization of this interruptive alert, and our lessons learned.This study analyzed the number of times this alert fired over 6 months of testing and 16 months of deployment, and changes in either the frequency of alert firing or the frequency of the alert being dismissed without orders placed with iterative changes in the alert logic. Feedback about this alert was compiled and classified into common themes.The initial alert fired an average of 11,154 times per week when tested silently, prompting significant refinements before release. The alert shown to users fired an average of 53.8 times per 1,000 patient days in the first 6 months of the study period. Despite postlaunch improvements, this rose to 61 alerts per 1,000 patient days in the final 6 months of the study. Modifications also did not cause a significant decrease in how frequently the alert was dismissed without further action being taken (88%). Review of narrative feedback and its classification highlights "wrong person" receiving the alert being by far the most prevalent cause for negative submitted user feedback (nearly 50%), despite efforts to develop logic that limits firing to the patient's primary team.Changes to this VTE alert were summarized as failures to meet the "five rights" of CDS. Alerts for high-priority safety issues require persistent feedback-driven improvement, particularly when there is poor performance or negative user experience.