Association model data learned from clinicians stratified by patient mortality outcomes at a Tertiary Academic Center.
Ontology highlight
ABSTRACT:
SUBMITTER: Wang JK
PROVIDER: S-EPMC6247447 | biostudies-literature | 2018 Dec
REPOSITORIES: biostudies-literature
Publications
Association model data learned from clinicians stratified by patient mortality outcomes at a Tertiary Academic Center.
Wang Jason K JK Hom Jason J Balasubramanian Santhosh S Chen Jonathan H JH
Data in brief 20181102
In this data article, we learn clinical order patterns from inpatient electronic health record (EHR) data at a tertiary academic center from three different cohorts of providers: (1) Clinicians with lower-than-expected patient mortality rates, (2) clinicians with higher-than-expected patient mortality rates, and (3) an unfiltered population of clinicians. We extract and make public these order patterns learned from each clinician cohort associated with six common admission diagnoses (e.g. pneumo ...[more]
Similar Datasets
Project description:ObjectiveEvaluate the quality of clinical order practice patterns machine-learned from clinician cohorts stratified by patient mortality outcomes.Materials and methodsInpatient electronic health records from 2010 to 2013 were extracted from a tertiary academic hospital. Clinicians (n = 1822) were stratified into low-mortality (21.8%, n = 397) and high-mortality (6.0%, n = 110) extremes using a two-sided P-value score quantifying deviation of observed vs. expected 30-day patient mortality rates. Three patient cohorts were assembled: patients seen by low-mortality clinicians, high-mortality clinicians, and an unfiltered crowd of all clinicians (n = 1046, 1046, and 5230 post-propensity score matching, respectively). Predicted order lists were automatically generated from recommender system algorithms trained on each patient cohort and evaluated against (i) real-world practice patterns reflected in patient cases with better-than-expected mortality outcomes and (ii) reference standards derived from clinical practice guidelines.ResultsAcross six common admission diagnoses, order lists learned from the crowd demonstrated the greatest alignment with guideline references (AUROC range = 0.86-0.91), performing on par or better than those learned from low-mortality clinicians (0.79-0.84, P < 10-5) or manually-authored hospital order sets (0.65-0.77, P < 10-3). The same trend was observed in evaluating model predictions against better-than-expected patient cases, with the crowd model (AUROC mean = 0.91) outperforming the low-mortality model (0.87, P < 10-16) and order set benchmarks (0.78, P < 10-35).DiscussionWhether machine-learning models are trained on all clinicians or a subset of experts illustrates a bias-variance tradeoff in data usage. Defining robust metrics to assess quality based on internal (e.g. practice patterns from better-than-expected patient cases) or external reference standards (e.g. clinical practice guidelines) is critical to assess decision support content.ConclusionLearning relevant decision support content from all clinicians is as, if not more, robust than learning from a select subgroup of clinicians favored by patient outcomes.
| S-EPMC6250126 | biostudies-literature
Project description:BackgroundThis study describes the program and learning outcomes of a telehealth skills curriculum based on the Association of American Medical Colleges (AAMC) telehealth competencies for clerkship-level medical students.MethodsA total of 133 third- and fourth-year medical students in a required family medicine clerkship at Stanford University School of Medicine participated in a telehealth curriculum, including a telehealth workshop, site-specific telehealth clinical encounters, and telemedicine objective structured clinical examinations (teleOSCEs) between July 2020 and August 2021. Their workshop communication and physical examination competencies were assessed in two teleOSCEs utilizing a novel telehealth assessment tool. Students' attitudes, skills, and self-efficacy were assessed through voluntary pre-clerkship, post-workshop, and post-OSCE surveys.DiscussionMost learners reported low confidence in their telehealth physical examinations [n = 79, mean = 1.6 (scale 0-5, 5 = very confident, SD = 1.0)], which improved post-workshop [n = 69, 3.3 (0.9), p < 0.001]; almost all (97%, 70/72) felt the workshop prepared them to see patients in the clinic. In formative OSCEs, learners demonstrated appropriate "webside manner" (communication scores 94-99%, four items) but did not confirm confidentiality (21%) or review limitations of the visit (35%). In a low back pain OSCE, most learners assessed pain location (90%) and range of motion (87%); nearly half (48%) omitted strength testing.ConclusionOur telehealth curriculum demonstrated that telehealth competencies can be taught and assessed in medical student education. Improvement in self-efficacy scores suggests that an 80-min workshop can prepare students to see patients in the clinical setting. Assessment of OSCE data informs opportunities for growth for further development in the curriculum, including addressing visit limitations and confidentiality in telehealth visits.
| S-EPMC10577422 | biostudies-literature
Project description:ObjectiveThe objective of this paper is to share challenges, recommendations, and lessons learned regarding the development and implementation of a Patient Safety Learning Laboratory (PSLL) project, an innovative and complex intervention comprised of a suite of Health Information Technology (HIT) tools integrated with a newly implemented Electronic Health Record (EHR) vendor system in the acute care setting at a large academic center.Materials and methodsThe PSLL Administrative Core engaged stakeholders and study personnel throughout all phases of the project: problem analysis, design, development, implementation, and evaluation. Implementation challenges and recommendations were derived from direct observations and the collective experience of PSLL study personnel.ResultsThe PSLL intervention was implemented on 12 inpatient units during the 18-month study period, potentially impacting 12,628 patient admissions. Challenges to implementation included stakeholder engagement, project scope/complexity, technology/governance, and team structure. Recommendations to address each of these challenges were generated, some enacted during the trial, others as lessons learned for future iterative refinements of the intervention and its implementation.ConclusionDesigning, implementing, and evaluating a suite of tools integrated within a vendor EHR to improve patient safety has a variety of challenges. Keys to success include continuous stakeholder engagement, involvement of systems and human factors engineers within a multidisciplinary team, an iterative approach to user-centered design, and a willingness to think outside of current workflows and processes to change health system culture around adverse event prevention.
| S-EPMC7647251 | biostudies-literature
Project description:There is an increasing demand for interventional endoscopic services and the need to develop efficient endoscopic units. The aim of this study was to analyze performance data and define metrics to improve efficiency in a single academic interventional endoscopy center. ]The prospective operations performance data (6-month period) of our interventional endoscopy unit (EU) was analyzed. First-case start time (FIRST) delay was defined as any time the first patient of the day entered the endoscopy room after the scheduled time. Non-endoscopy time (NET) and total time (TT) were defined as non-procedural and total time elapsed in the EU, respectively. Time-interval between successive patients (TISP) was defined as the time from one patient departure from the room until the time of arrival of the next patient in the room.A total of 1421 patients underwent 1635 endoscopic procedures. FIRST was delayed (54.2 % cases) by 13.6 min (range 1 - 53), but started within 15 min of the scheduled time in 85 % of the cases. NET accounted for 9.1 hours (67.2 %) of 13.5 hours TT/day. TISP (37.1 min, range 5 - 125) comprised 54.2 % of the NET, and was delayed (> 30 min) in 49.8 % of cases. "Patient flow" processes (registration, admission, transportation, scheduling) accounted for 50.1 % of TISP delays.Delays in NET, specifically TISP, rather than FIRST, were identified as a cause for decreased efficiency. "Patient flow" processes were the main reasons for delays in TISP. This study identifies potential process measures that can be used as benchmarks to improve efficiency in the EU.
| S-EPMC4751008 | biostudies-other
Project description:BACKGROUND:The Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey is the first publicly reported nationwide survey to evaluate and compare hospitals. Increasing patient satisfaction is an important goal as it aims to achieve a more effective and efficient healthcare delivery system. In this study, we develop and apply an integrative, data-driven approach to identify clinical risk factors that associate with patient satisfaction outcomes. METHODS:We included 1,771 unique adult patients who completed the HCAHPS survey and were discharged from the inpatient Medicine service from 2010 to 2012. We collected 266 clinical features including patient demographics, lab measurements, medications, disease categories, and procedures. We developed and applied a data-driven approach to identify risk factors that associate with patient satisfaction outcomes. FINDINGS:We identify 102 significant risk factors associating with 18 surveyed questions. The most significantly recurrent clinical risk factors were: self-evaluation of health, education level, Asian, White, treatment in BMT oncology division, being prescribed a new medication. Patients who were prescribed pregabalin were less satisfied particularly in relation to communication with nurses and pain management. Explanation of medication usage was associated with communication with nurses (q = 0.001); however, explanation of medication side effects was associated with communication with doctors (q = 0.003). Overall hospital rating was associated with hospital environment, communication with doctors, and communication about medicines. However, patient likelihood to recommend hospital was associated with hospital environment, communication about medicines, pain management, and communication with nurse. CONCLUSIONS:Our study identified a number of putatively novel clinical risk factors for patient satisfaction that suggest new opportunities to better understand and manage patient satisfaction. Hospitals can use a data-driven approach to identify clinical risk factors for poor patient satisfaction to support development of specific interventions to improve patients' experience of care.
| S-EPMC4881910 | biostudies-literature
Project description:IntroductionThe COVID-19 pandemic has prompted global research efforts to reduce infection impact, highlighting the potential of cross-disciplinary collaboration to enhance research quality and efficiency.MethodsAt the FMUSP-HC academic health system, we implemented innovative flow management routines for collecting, organizing and analyzing demographic data, COVID-related data and biological materials from over 4,500 patients with confirmed SARS-CoV-2 infection hospitalized from 2020 to 2022. This strategy was mainly planned in three areas: organizing a database with data from the hospitalizations; setting-up a multidisciplinary taskforce to conduct follow-up assessments after discharge; and organizing a biobank. Additionally, a COVID-19 curated collection was created within the institutional digital library of academic papers to map the research output.ResultsOver the course of the experience, the possible benefits and challenges of this type of research support approach were identified and discussed, leading to a set of recommended strategies to enhance collaboration within the research institution. Demographic and clinical data from COVID-19 hospitalizations were compiled in a database including adults and a minority of children and adolescents with laboratory confirmed COVID-19, covering 2020-2022, with approximately 350 fields per patient. To date, this database has been used in 16 published studies. Additionally, we assessed 700 adults 6 to 11 months after hospitalization through comprehensive, multidisciplinary in-person evaluations; this database, comprising around 2000 fields per subject, was used in 15 publications. Furthermore, thousands of blood samples collected during the acute phase and follow-up assessments remain stored for future investigations. To date, more than 3,700 aliquots have been used in ongoing research investigating various aspects of COVID-19. Lastly, the mapping of the overall research output revealed that between 2020 and 2022 our academic system produced 1,394 scientific articles on COVID-19.DiscussionResearch is a crucial component of an effective epidemic response, and the preparation process should include a well-defined plan for organizing and sharing resources. The initiatives described in the present paper were successful in our aim to foster large-scale research in our institution. Although a single model may not be appropriate for all contexts, cross-disciplinary collaboration and open data sharing should make health research systems more efficient to generate the best evidence.
| S-EPMC10927964 | biostudies-literature
Project description:ObjectivesThis quality improvement project sought to enhance clinical information sharing for interhospital transfers to an inpatient hepatology service comprised of internal medicine resident frontline providers (housestaff) with the specific aims of making housestaff aware of 100% of incoming transfers and providing timely access to clinical summaries.InterventionsIn February 2020, an email notification system to senior medicine residents responsible for patient triage shared planned arrival time for patients pending transfer. In July 2020, a clinical data repository ("Transfer Log") updated daily by accepting providers (attending physicians and subspecialty fellows) became available to senior medicine residents responsible for triage.MethodsLikert scale surveys were administered to housestaff before email intervention (pre) and after transfer log intervention (post). The time from patient arrival to team assignment (TTA) in the electronic medical record was used as a proxy for time to patient assessment and was measured pre- and postinterventions; >2 hours to TTA was considered an extreme delay.ResultsHousestaff reported frequency of access to clinical information as follows: preinterventions 4/31 (13%) sometimes/very often and 27/31 (87%) never/rarely; postinterventions 11/26 (42%) sometimes/very often and 15/26 (58%) never/rarely (p = 0.02). Preinterventions 12/39 (31%) felt "not at all prepared" versus 27/39 (69%) "somewhat" or "adequately"; postinterventions 2/24 (8%) felt "not at all prepared" versus 22/24 (92%) somewhat/adequately prepared (p = 0.06). There was a significant difference in mean TTA between pre- and posttransfer log groups (62 vs. 40 minutes, p = 0.01) and a significant reduction in patients with extreme delays in TTA post-email (18/180 pre-email vs. 7/174 post-email, p = 0.04).ConclusionEarly notification and increased access to clinical information were associated with better sense of preparedness for admitting housestaff, reduction in TTA, and reduced frequency of extreme delays in team assignment.
| S-EPMC9433165 | biostudies-literature
Project description:ICU is a multifaceted organization where multiple teams care for critically ill patients. In the current era, collaboration between teams and efficient workflows form the backbone of value-based care. Geographical cohorting is a widespread model for hospitalist rounding, but its role in ICUs is unclear. This study evaluates the outcomes of geographical cohorting in a large ICU of an Academic Health Center.DesignThis is a retrospective analysis of quality metrics collected 12 months pre- and post-implementation of geographical cohorting.SettingA total of 130 bedded ICU at tertiary academic health center in Midwest.PatientsAll patients admitted to the ICU.InterventionsOur institution piloted the geographical cohorting model for critical care physician rounding on September 1, 2018.MeasurementsThe quality metrics were categorized as ICU harm events and ICU hospital metrics. Team of critical care providers were surveyed 12 months after implementation.Main resultsThe critical care utilization in the pre- and post-implementation numbers were similar for patient days (pre = 34,839, post = 35,155), central-line days (pre = 17,648, post = 19,224), and Foley catheter days (pre = 18,292, post = 17,364). The ICU length of stay was similar (4.9 d) in both pre- and post-intervention periods. Significant reduction in the incidence of Clostridium difficile infection (relative risk, -0.50; 95% CI, 0.25-0.96; p = 0.039), hospital-acquired pressure injury (relative risk, -0.60; 95% CI, 0.39-0.92; p = 0.020), central line-associated bloodstream infection incidence (relative risk, -0.19; 95% CI, 0.05-0.52; p = 0.008), and catheter-associated urinary tract infection (relative risk, -0.52; 95% CI, 0.29-0.93; p = 0.027). Healthcare providers perceived optimal utilization of their time, reduced interruptions, and improved coordination of care with geographical rounding.ConclusionsGeographical cohorting improves coordination of care, physician workflow, and critical care quality metrics in very large ICUs.
| S-EPMC7523794 | biostudies-literature
Project description:Patient outcomes with hospitalist care have been studied in many settings, yet little is known about how hospitalist care interacts with trainee care to affect patient outcomes in teaching hospitals.The aim of this study was to compare patient outcomes between hospitalist-preceptors and hospitalists working alone (isolating the effect of housestaff involvement), and between hospitalist-preceptors and academician-preceptors (isolating the effect of attending type, given housestaff involvement).A four-year retrospective cohort study of patients (n?=?13,313) admitted to all internal medicine services at an academic medical center from July 2008 to June 2012.Using generalized estimating equations, we measured readmission within 30 days, hospital length of stay, cost of the index hospitalization, and cumulative cost including readmissions within 30 days.In the adjusted models, 30-day readmission odds were higher for academic-preceptors (OR, 1.14 [95% CI, 1.03?-?1.26]) and hospitalist-preceptors (OR, 1.10 [95% CI, 1.002?-?1.21]) than for hospitalists working alone. Compared with hospitalists working alone, academic-preceptors were associated with shorter length of stay (mean difference, 0.27 days [95% CI, 0.18?-?0.38]), lower index hospitalization costs (mean difference, $386 [95% CI, $192?-?$576]), but similar cumulative inpatient costs within 30 days of discharge. Compared with hospitalists working alone, hospitalist-preceptors were associated with shorter length of stay (mean difference, 0.34 days [95% CI, 0.26?-?0.42]), lower index hospitalization cost (mean difference, $570 [95% CI, $378?-?$760]), and a trend toward lower cumulative cost (mean difference, $1347 [95% CI, $254?-?$2,816]).Preceptor-led medicine services were associated with more readmissions within 30 days, shorter lengths of stay, and lower index admission-associated costs. However, when considering cumulative hospitalization costs, patients discharged by academician-preceptors incurred the highest cost and hospitalist-preceptors incurred the lowest cost.
| S-EPMC4242879 | biostudies-other
Project description:Hereditary leiomyomatosis and renal cell cancer (HLRCC) is an autosomal dominant condition attributed to pathogenic variants in fumarate hydratase (FH) and presents with cutaneous leiomyomas (CLMs), uterine leiomyomas (ULMs) and renal cell cancer (RCC). The objective of this study was to characterize the spectrum of clinical and genetic findings in HLRCC at a large academic tertiary care referral center with a focus on dermatologic manifestations. Fifty-seven patients, 41 female and 16 male, with 27 unique pathogenic or likely-pathogenic FH variants were identified from 38 families. Mean age of HLRCC diagnosis was 44.4 years (range 8-82). CLMs were the primary reason for referral in 49.1% (n=28). CLMs were present in 43/56 patients who underwent full skin examination. Three of these 56 patients were diagnosed with cutaneous leiomyosarcoma. Incidence of ULMs was 37/41 female patients; no uterine leiomyosarcomas were observed. RCC was observed in 6/57 patients (mean age of diagnosis: 47.3 years (range 28-79)). CLMs predated RCC in the 3 patients diagnosed with both. Dermatologists have an opportunity to recognize cutaneous manifestations of HLRCC, including cutaneous leiomyomas and rarely cutaneous leiomyosarcomas, and refer for genetic evaluation to provide definitive diagnosis. Identification of HLRCC can promote family cascade testing and screening for RCC.
| S-EPMC9906083 | biostudies-literature