A multi-center collaborative network for emergency medicine research at Kaiser Permanente.
Selected Projects
The KP CREST Network in Northern California is not alone in doing emergency medicine research in Kaiser Permanente. Clinical researchers in KP Southern California are also active. We in CREST collaborate with our SoCAL counterparts on studies and also team up to produce quarterly reports that highlight recent projects and publications from both halves of the Golden State. You can read the latest report here.
Improving Risk Stratification for Acute Heart Failure
Building and Testing a Machine-learning Platform for Personalized, Accurate, Real-time Risk Prediction
Investigators: Dana Sax (Oakland), Jamal Rana (Oakland), Dustin Mark (Oakland), Dustin Ballard (San Rafael), Vincent Liu (DOR), Mary Reed (DOR) and the KP CREST Network
In this project, we leverage the power of innovative machine learning approaches and the data available in our comprehensive electronic health record to develop a highly accurate acute heart failure ED risk stratification tool. Our short-term goal is to implement this tool for bedside, real-time decision support to make better informed hospitalization decisions for ED patients with heart failure. We will identify multilevel barriers and facilitators to dissemination and
implementation, develop tailored implementation strategies, and conduct a multi-center trial to evaluate its implementation. Our long-term goal is to develop a scalable real-time clinical decision support process and systemic implementation strategy for other disease processes.
Leveraging EMS Data for Improved ED Triage
Incorporating EMS Data into ED Risk Stratification tools; A Triage-Based Proof of Concept Study
Investigators: Dustin Ballard (San Rafael), Dana Sax (Oakland), Mary Reed (DOR), David R Vinson (Sacramento), and the KP CREST Network
This study addresses the challenge of improving ED triage accuracy for patients arriving via ambulance or other EMS transport. Current triage tools often result in significant under- or over-triage, particularly for EMS patients, and contribute to operational inefficiencies such as ambulance patient offload delays (APOD). By integrating pre-hospital EMS data—including vital signs, interventions, and diagnostics—into ED risk stratification tools, this project aims to enhance triage performance and streamline ED throughput. Preliminary findings suggest that leveraging EMS data can improve predictive models for patient acuity, with potential applications across all Kaiser Permanente EDs. Outcomes of interest include triage accuracy, APOD metrics, and patient outcomes, with findings expected to inform regional and statewide ED operations and quality improvement initiatives.
Pediatric Respiratory Illnesses in the Post-COVID Era
Illness Epidemiology and ED Care and Outcomes
Investigators: Mamata Kene (San Leandro), Tara Greenhow (San Francisco), Dana Sax (Oakland), Mary Reed (DOR), and the KP CREST Network
This longitudinal cohort study explores the changing epidemiology and outcomes of pediatric respiratory illnesses in the wake of the COVID-19 pandemic. By analyzing ED visits from 2018 to 2024, the study examines trends in hospitalization, serious illness, and ED utilization during the pre-pandemic, pandemic, and post-pandemic periods. It investigates the impact of mitigation measures like mask-wearing and social distancing on respiratory disease patterns and the resurgence of illnesses like RSV in 2022. Findings aim to inform surge planning, resource allocation, and prevention strategies, providing actionable insights for clinicians and health systems facing future respiratory illness surges.
Pregnancy-Associated Pulmonary Embolism
Assessment of Outpatient Management Strategies
Investigators: David R. Vinson (Sacramento), Mary Reed (DOR), Nareg Roubinian (Oakland), and the KP CREST Network
This data-only cohort study evaluates the safety, prevalence, and eligibility criteria for outpatient management of pregnancy-associated pulmonary embolism (PE). Despite professional society endorsements, the practice remains largely unexplored in clinical literature. Leveraging data from over a decade of cases across Kaiser Permanente Northern California, this study will analyze the prevalence of outpatient PE management, identify patient- and facility-level factors for eligibility, and assess safety outcomes within seven days of discharge. Findings aim to pave the way for a shift in PE care for pregnant and postpartum patients, addressing a significant gap in current medical knowledge and informing global guidelines.
INF STEWARD (Infant Fever: Standardizing Emergency Work-ups Around Risk Data)
Management of the febrile infant in the Emergency Department
Investigators: Dustin Ballard (San Rafael), Tara Greenhow (San Francisco), David R. Vinson (Sacramento/Roseville), Mary Reed (DOR) and the KP CREST Network
Including collaboration with investigators in Kaiser Southern California, this study aims to improve the evaluation and management of febrile infants, drawing from risk data to standardize practice. The multiphase project aims to assess current patterns of care, introduce guideline-based clinical decision support, and integrate shared decision-making tools. By addressing variability in diagnostics, unnecessary testing, and hospitalizations, the study seeks to enhance clinical outcomes, reduce resource utilization, and promote evidence-based care. Investigators continue to build upon and modify algorithms to optimize care for the febrile infant.
"Cold Drink Heart"
Characterizing Patients with Cold Drink-triggered Atrial Fibrillation: A Survey Study
Investigators: David R Vinson (Sacramento), Adina Rauchwerger (DOR), Jennifer Zhang (DOR), Daniel DiLena (DOR), and the KP CREST Network
Atrial fibrillation (AF) is the most treated cardiac arrhythmia and is associated with an increased risk for heart failure, stroke, myocardial infarction, chronic kidney disease, dementia, and death. Cold drink or food-triggered AF was first reported in the literature 30 years ago, but little progress has been made in understanding this association. We are conducting a patient survey study to characterize patients who have experienced "Cold Drink Heart." Although this study is currently underway, it is evident that avoidance of known AF triggers can help to reduce or even eliminate recurrent episodes.
Syncope
Improving management of emergency department patients with unexplained syncope: prospective validation of the Canadian Syncope Risk Score
Investigators: David R. Vinson (Sacramento), Mary Reed (DOR), and the KP CREST Network
Syncope, defined as a temporary loss of consciousness and muscle tone with complete, spontaneous recovery, accounts for approximately 1% of all ED visits across the US. A small percentage of unexplained syncope cases are the result of a much more serious underlying issue, such as ventricular arrythmia or pulmonary embolism. This project seeks to build and validate a clinical decision support system in order to better stratify syncope patients by their risk for an adverse event. This tool, which has been embedded in RISTRA (our risk stratification software) is currently undergoing prospective validation at several KPNC faclities.
Post-acute sequelae of COVID (PASC)
Incidence, characteristics and predictors of emergency department visits associated with post-acute sequelae SARS-CoV-2 infection (PASC) among adults
Investigators: Dustin Ballard (San Rafael), Mary Reed (DOR), Jacek Skarbinski (DOR), Edward Durant (Manteca), David R. Vinson (Sacramento), Marcos Siqueiros (Santa Clara), Sri Madhavi Cholleti (Campbell), Edward Durant (Modesto) and the KP CREST Network
It is well established that COVID-19 symptoms can persist long after initial SARS-CoV-2 infection. The most common of these "post-acute sequelae of COVID" include loss of taste or smell, fatigue, and anxiety and depression. This project seeks to characterize all patients with a PASC-related ED visit and establish associations between specific patient characteristics and a greater likelihood of developing PASC.
RISTRiage
Standardizing ER triage across KPNC emergency departments: understanding the problem of under-triage and its implications on patient safety
Investigators: Dana Sax (Oakland), Mary Reed (DOR), and the KP CREST Network
The most commonly used ED triage system in the United States, and throughout KPNC, is the Emergency Severity Index. Unfortunately, studies have shown that triage assignments are often subjective and have poor inter-rater reliability. Mis-triage, both under- and over-triage, leads to delays in care and may introduce bias into providers’ evaluations, with serious potential patient safety concerns. We assessed the frequency of under-triage and its potential patient safety implications across all 21 KPNC EDs. We also identified factors that predict this under-recognition of patient acuity or resource needs.
RISTRA-ACS
Validation of a Novel Algorithm for Emergency Department Patients with Possible Acute Coronary Syndromes Following the Transition to a High-Sensitivity Troponin Assay
Investigators: Dustin Mark, Jie Huang, Keane Lee, Dana Sax, Mamata Kene, Dustin Ballard, David R. Vinson, Mary Reed
High sensitivity troponin (hsTn) assays are being increasingly adopted throughout the US, replacing the conventional troponin (cTn) assay. This increase in analytic precision at the patient evaluation stage provides a higher level of diagnostic sensitivity for myocardial injury. An assessment of the predictive accuracy of the KPNC hsTn diagnostic algorithm among ED patients found a 99.8% negative predictive value for 30-day myocardial infarction while identifying 70% of patients for expedited discharge.
