Gabriel J. Escobar, MD

gabriel.escobar@kp.org

Gabriel J. Escobar, MD, is a retired research scientist at the Kaiser Permanente Northern California Division of Research. He was director of the Division of Research Systems Research Initiative (a research program focusing on adult hospital processes and outcomes, informatics, and program evaluation); and Regional Director for Hospital Operations Research for Kaiser Permanente Northern California, in which capacity he worked to improve Kaiser Permanente’s internal reporting and quality measurement capabilities. Dr. Escobar received his medical degree from Yale University School of Medicine; completed his pediatrics residency at University of California, San Francisco; and was a Robert Wood Johnson Clinical Scholar at Stanford University School of Medicine. He is board certified in Clinical Informatics.

Dr. Escobar’s research interests included risk adjustment, predictive modeling, severity-of-illness scoring, the use of comprehensive inpatient and outpatient electronic medical records for health services research, and the use of real-time decision support tools that are embedded in the electronic record. Between 1991 and 2001, Dr. Escobar developed a research program in neonatology for Kaiser Permanente Northern California. In 2009 he turned over the neonatology research program to Michael Kuzniewicz, MD, at the Division of Research so that he could focus on the work of the Systems Research Initiative. Dr. Escobar also worked as a pediatric hospitalist at the Kaiser Permanente Walnut Creek, Antioch, and Oakland Medical Centers between 1984 and 2016.

Studies

COVID-19: Antecedents to Hospitalization and Deterioration

This study will characterize and compare antecedents to hospitalization and in-hospital deterioration among patients with COVID-19 and comparator patients with influenza. We will quantify pre-existing comorbidities, medication adherence, and temporal factors prior to hospitalization with COVID-19 or influenza. We will also characterize the antecedents and time course of infections in these 2 groups of patients and quantify the relative contribution of 4 key groups of predictors for non-elective hospitalization and adverse outcomes (critical illness, death): demographics, pre-existing comorbidities, medication adherence, and time to treatment for patients who went to the emergency department.

Investigator: Escobar, Gabriel

Funder: AstraZeneca PLC

Tools for Outpatient and Population Management of SARS-COV-2 Infection (TOPS2)

This project aims to improve The Permanente Medical Group’s capability for outpatient and population management of adults with COVID-19. It will identify the spatial, temporal, and racial distribution of coronavirus infections; characterize the clinical progression of coronavirus infection; and synthesize these findings to develop both simple predictive models and job aids that can help clinicians in the outpatient setting and complex risk adjustment models for population management. The tools we develop will help general internists and family practitioners predict the likely temporal and clinical trajectory of patients initally seen as outpatients or in the emergency department.

Investigator: Escobar, Gabriel

Funder: TPMG Delivery Science Projects Program

Hospital Alternatives (HALT)

The SRI team will (a) assess results of previous internal and external deployments of alternatives to hospitalization, (b) define strategies to quantify the potential pool of Kaiser Permanente Northern California patients who would be eligible for such alternatives, (c) develop appropriate predictive models to support planned and possible deployment strategies, (d) identify synergies with other SRI projects, and (e) determine sequencing of efforts from a compliance standpoint.

Investigator: Escobar, Gabriel

Funder: Kaiser Foundation Hospitals

Development, Reevaluation, and Implementation of Outcome/Efficiency Measures for Hospital and Eligible Clinicians

The goal of this project is to assist with measure development and testing for the Hospital-Wide Mortality (HWM) hybrid and HWM claims-only measures. This includes providing consulting support to the Yale Center for Outcomes Research and Evaluation team and defining key data elements.

Investigator: Escobar, Gabriel

Funder: Centers for Medicare and Medicaid Services

It Takes a Network: A Network Science- and Command Center-Based Approach to Hospital Error Detection, Prevention, and Mitigation

This project aims to develop a set of algorithms for generic error detection, prevention, and mitigation in the hospital setting. The goal is to develop algorithms that can operate in real time so as to support an existing infrastructure (the eHospital Safety Net, a Kaiser Permanente Northern California system that involves trained registered nurses who monitor the inpatient hospital record in real time, remotely). To develop these algorithms, network science, neural networks, and machine learning will be employed to identify discrete, transient (temporal frame of one to two hours) patient clusters at high risk for errors and process failures. This project builds on work conducted by the same team (assignment of severity of illness scores in real time, and quantification of in-hospital ecological markers such as occupancy and hourly change in census), as well as existing systems such as the Hospital Throughput Monitor. It expands on this work by developing an approach to quantify cognitive load on networks of caregivers and patients as well as hospital units. The project involves a collaboration with Stanford University.

Investigator: Escobar, Gabriel

Funder: Kaiser Foundation Hospitals

Social and Functional Predictors of Rehospitalization: YCLS and PROMIS (SDH2)

This project aims to determine whether fields included in the National Institutes of Health’s Patient Reported Outcomes Measurement System (PROMIS) questionnaires for cognition and physical function, as well as those in the Your Current Life Situation (YCLS) questionnaire can enhance our ability to predict rehospitalization. 1600 patients will be interviewed shortly before their hospital discharge from the Kaiser Permanente Northern California Oakland, San Leandro, and Walnut Creek Hospitals. Patients at low or medium risk for rehospitalization will be oversampled. Data from the interviews will be combined with an existing real time rehospitalization score in use in all Kaiser Permanente Northern California hospitals. If successful, the project would permit better targeting of patients in need of support following hospitalization.

Investigator: Escobar, Gabriel

Funder: Kaiser Permanente Program Offices

Does Information about Members’ Social Determinants of Health Improve Prediction of Risk for Hospitalization, Emergency Department Visits, and Outpatient Utilization?

The new Kaiser Permanente Your Current Life Situation (YCLS) screening questionnaire will be mailed to a random sample of 7,000 adult members of Kaiser Permanente Northern California (KPNC), oversampling patients at low to medium risk for hospitalization. After receiving responses, YCLS data will be linked with existing variables known to be predictors of utilization, including age, sex, COPS2 (COmorbidity Point Score, version 2), abLAPS (abbreviated Laboratory-based Acute Physiology Score), other composite indices currently available in KPNC electronic databases, and U.S. census-derived socioeconomic status data. The linked data will be analyzed to determine the extent to which the YCLS data improves prediction of hospitalization, emergency department use, and total outpatient utilization, and, if improvement in prediction is found after incorporating YCLS data, which specific items from the YCLS have the greatest impact. Responses to the YCLS survey, weighted to the adult membership, will also be used to estimate prevalence of KPNC adult members’ social and economic needs.

Investigator: Gordon, Nancy; Escobar, Gabriel

Funder: Kaiser Permanente Program Offices

Social and Functional Predictors of Rehospitalization: YCLS and PROMIS

This study aims to determine whether specific items from two questionnaires currently in use at Kaiser Permanente Northern California (the Your Current Life Situation questionnaire and selected functional status questionnaires from the National Institutes of Health’s Patient Reported Outcomes Measurement System) could enhance the ability to predict a patient’s risk of rehospitalization. To achieve this goal, approximately 2,000 patients will be approached on the day of discharge at three hospitals with a target goal of enrolling 1,600. Data from the questionnaires will be combined with other electronic data in order to modify existing algorithms used in the current Kaiser Permanente Northern California real-time rehospitalization prediction model.

Investigator: Escobar, Gabriel

Funder: Kaiser Permanente Program Offices

Characterizing and Predicting Adverse Drug Events Outside the Hospital

This study aims to employ machine learning methods to develop predictive models for adverse drug events, or ADEs, that result in patients using the emergency department and/or being hospitalized (non-nosocomial ADEs, NNADEs). Based on work with a pharmacist expert panel that will include examination of literature as well as KPNC ADE data, six candidate drugs that are currently not the focus of preventive efforts will be selected. A population database that will include geographic information system data as well as ADEs will then be built. Finally, using machine learning methods, the investigators will attempt to develop predictive models for ADEs involving the six selected candidate drugs.

Investigator: Escobar, Gabriel

Funder: Gordon and Betty Moore Foundation

Early Detection of Obstetric Complications: Development of Predictive Models Suitable for Real-Time Surveillance During Labor, Delivery, and the Postpartum Period

This is a feasibility study to determine whether data from Kaiser Permanente HealthConnect® can be harnessed to predict obstetric complications in real time. The long-term goal is to develop algorithms that could be instantiated in the electronic record to provide clinicians with discrete probability estimates with lead times of 6 to 12 hours. If successful, the algorithms would be instantiated using methods similar to those employed with the Advanced Alert Monitor, the early warning system currently in operation at Kaiser Permanente Northern California.

Investigator: Escobar, Gabriel

Funder: Kaiser Permanente Program Offices

Recovery or Disability? Trajectories After Pneumonia: Integrating 12 Months of Frequent Patient-Reported Measures with EMR Data

This is a pilot study to determine the feasibility of conducting a prospective study that aims to characterize longitudinal trajectories among patients with pneumonia. To achieve this goal, we will attempt to enroll 150 pneumonia patients at the time of hospital discharge, with a goal of retaining 100. Enrolled patients will be interviewed in the hospital, within 1 week of hospital discharge, and then again at 1.5, 3, and 4.5 months post-discharge. Our long-term objectives are to characterize post-discharge trajectories, with special emphasis on the impact of rehospitalization on a patient’s subsequent trajectory. The co-principal investigator will be Theodore Iwashyna, MD, PhD, of the University of Michigan.

Investigator: Escobar, Gabriel

Funder: National Institute on Aging

Respiratory Syncytial Virus Infection in Preterm Infants: Six Seasons; Experience in an Integrated Health Care Delivery System

We will study the impact of the change in the American Academy of Pediatrics’s palivizumab recommendations in a large cohort of infants (300,000) of all gestations over a 7-year period (2010- 2016). We will quantify and characterize the following over this period: 1) occurrence of bronchiolitis in children 24 months of age; 2) testing patterns for Respiratory Syncytial Virus (RSV); 3) RSV immunoprophylaxis utilization; and 4) health care utilization among infants and children with RSV infection, with particular emphasis on hospitalization and use of intensive care.

Investigator: Escobar, Gabriel

Funder: AstraZeneca PLC

Default Palliative Care Consultation for Seriously Ill Hospitalized Patients

This project aims to test the impact of changing the default mode for palliative care consultation in hospitalized patients from an “opt in” to an “opt out” model. This will be tested in 10-12 Ascension Health hospitals using a stepped wedge design. No KP patients are involved, and Dr. Escobar’s primary role will be to assist in the UH2 (planning phase), which has the following aims: develop electronic medical record algorithms to identify patients meeting palliative care triggers; validate data capture of all proposed outcomes in a sample of de-identified patients; validate methods for tracking adherence to electronically generated palliative care orders; establish mechanisms and materials for notifying patients and providers about the research; complete regulatory compliance components, including defining strategy for IRB approval, trial registry, and DSMB assembly.

Investigator: Escobar, Gabriel

Funder: National Institute on Aging

Advance Alert Monitor Operations

The Systems Research Initiative will provide project management and operational consultation for the spread of the Advance Alert Monitor within KPNC between 10/1/16 and 9/30/18. During this time Division of Research will complete the following tasks: continue the support of AAM initiation in all KPNC hospitals; troubleshoot transition from Q6 hours to Q1 hours alarm firing; provide parameters, guidelines, variables, and oversight for quality control algorithms; recalibrate the AAM model twice (once during first year and once during second year); develop documentation for the AAM algorithm and its components; develop documentation for model recalibration; collaborate with KPIT to develop documentation of the Java application; collaborate to hand off operation of AAM to appropriate entities TBD; manual chart abstraction for evaluation of adherence to Life Care Planning program; and preliminary and final outcomes analysis, including submission of manuscripts for publication.

Investigator: Escobar, Gabriel

Funder: Kaiser Foundation Health Plan

Development, Reevaluation and Implementation of Outcome/Efficiency Measures

This project aims to test the feasibility of employing clinical data from comprehensive electronic medical record systems (such as the Epic EMR) for developing, testing and validating potential publicly-reported quality, process, outcomes and risk-adjustment measures on hospitalized adult patients.

Investigator: Escobar, Gabriel

Funder: Centers for Medicare and Medicaid Services

Early Detection, Prevention and Mitigation of Impending Physiologic Deterioration in Hospitalized Patients Outside the ICU: A Pilot Project

An electronic alert system targeting adult patients in 3 hospital units (ward, telemetry, and transitional care unit) will be deployed at two KPNC hospitals (South San Francisco and Sacramento). The alert system provides early (12 hours) warning about a patient’s impending risk of deterioration as well as global risk around the time of admission to the hospital. It will be coupled with changes in clinician workflow (e.g., use of an automated checklist for patients who have triggered an alert) as well as frequent feedback on hospital adverse events from the DOR. To evaluate the pilot intervention, a difference in differences analysis will be conducted in which adverse outcome rates in the two pilot sites will be compared against the remaining 19 KPNC hospitals. Data from the pilot will be used to estimate duration of a stepped wedge cluster randomized trial involving all remaining KPNC hospitals.

Investigator: Escobar, Gabriel

Funder: Gordon and Betty Moore Foundation

CAREER: Managing Patient Flows with Congestion Effects

This project will employ simulation methods to analyze emergency department, hospital, and ICU data, including physiologic markers of severity of illness, with a view to identifying specific patient flow, hospital and ED occupancy states, and processes of care (e.g., early goal-directed therapy for sepsis) associated with improved patient outcomes (e.g., death, unplanned transfer to intensive care) as well as improved system states (e.g., absence of ED overflow conditions, absence of census-driven discharges from hospital units).

Investigator: Escobar, Gabriel

Funder: National Science Foundation

Development of Predictive Models for Recurrent Clostridium Difficile Infection (rCDI)

This project will examine predictors of and outcomes after recurrent Clostridium difficile infection in adults. Predictors will include age, sex, diagnoses, severity of illness, hospital length of stay, and antibiotic exposure. Outcomes will include hospital utilization, particularly use of intensive care, mortality, and other major morbidity (e.g., major bowel surgery). The initial cohort will consist of adults 18 years of age or older who (a) were members on 1/1/02, (b) did not have any C difficile infection between 9/1/01 and 12/31/01, and (c) developed C difficile infection on or after 1/1/02. These patients will then be divided into those that did and did not have a recurrent (2nd) C difficile infection.

Investigator: Escobar, Gabriel

Funder: Merck ＆ Company, Inc.

Publications

Evaluation of Incident 7-Day Infection and Sepsis Hospitalizations in an Integrated Health System

Authors: Liu, Vincent X; Manickam, Raj N; Greene, John D; Schuler, Alejandro; Kipnis, Patricia; Bhimarao, Meghana; Barreda, Fernando; Escobar, Gabriel J

Ann Am Thorac Soc. 2022 05;19(5):781-789.

Gabriel J. Escobar, MD

Studies

COVID-19: Antecedents to Hospitalization and Deterioration

Tools for Outpatient and Population Management of SARS-COV-2 Infection (TOPS2)

Hospital Alternatives (HALT)

Development, Reevaluation, and Implementation of Outcome/Efficiency Measures for Hospital and Eligible Clinicians

It Takes a Network: A Network Science- and Command Center-Based Approach to Hospital Error Detection, Prevention, and Mitigation

Social and Functional Predictors of Rehospitalization: YCLS and PROMIS (SDH2)

Does Information about Members’ Social Determinants of Health Improve Prediction of Risk for Hospitalization, Emergency Department Visits, and Outpatient Utilization?

Social and Functional Predictors of Rehospitalization: YCLS and PROMIS

Characterizing and Predicting Adverse Drug Events Outside the Hospital

Early Detection of Obstetric Complications: Development of Predictive Models Suitable for Real-Time Surveillance During Labor, Delivery, and the Postpartum Period

Recovery or Disability? Trajectories After Pneumonia: Integrating 12 Months of Frequent Patient-Reported Measures with EMR Data

Respiratory Syncytial Virus Infection in Preterm Infants: Six Seasons; Experience in an Integrated Health Care Delivery System

Default Palliative Care Consultation for Seriously Ill Hospitalized Patients

Advance Alert Monitor Operations

Development, Reevaluation and Implementation of Outcome/Efficiency Measures

Early Detection, Prevention and Mitigation of Impending Physiologic Deterioration in Hospitalized Patients Outside the ICU: A Pilot Project

CAREER: Managing Patient Flows with Congestion Effects

Development of Predictive Models for Recurrent Clostridium Difficile Infection (rCDI)

Publications

Evaluation of Incident 7-Day Infection and Sepsis Hospitalizations in an Integrated Health System

Developing a shared sepsis data infrastructure: a systematic review and concept map to FHIR

Association of Intensive Care Unit Admission and Outcomes in Sepsis and Acute Respiratory Failure

Performance of Predictive Models for 30-day Hospitalization and Mortality after COVID-19 Infection

Hospital-specific Template Matching for Benchmarking Performance in a Diverse Multihospital System

Predictive Models for In-Hospital Deterioration in Ward Patients

Validation of Postpartum Hemorrhage Admission Risk Factor Stratification in a Large Obstetrics Population

Evaluation of an intervention targeted with predictive analytics to prevent readmissions in an integrated health system: observational study

AUTOMATED ALERTS IN OBSTETRICS

Development of a healthcare system COVID Hotspotting Score in California: an observational study with prospective validation

Equitably Allocating Resources During Crises: Racial Differences in Mortality Prediction Models

Racial Disparities in COVID-19 Testing and Outcomes : Retrospective Cohort Study in an Integrated Health System

The Presentation, Pace, and Profile of Infection and Sepsis Patients Hospitalized Through the Emergency Department: An Exploratory Analysis

Automated Identification of Adults at Risk for In-Hospital Clinical Deterioration. Reply

PREDICTION OF OBSTETRIC AND FETAL COMPLICATIONS USING AUTOMATED ELECTRONIC HEALTH RECORD DATA

Development and Validation of an Obstetric Comorbidity Risk Score for Clinical Use

Predicting preventable hospital readmissions with causal machine learning

Automated Identification of Adults at Risk for In-Hospital Clinical Deterioration

Expanding Post-discharge Readmission Metrics in Patients with Chronic Obstructive Pulmonary Disease

Association of a Novel Index of Hospital Capacity Strain with Admission to Intensive Care Units

Assessment of Value of Neighborhood Socioeconomic Status in Models That Use Electronic Health Record Data to Predict Health Care Use Rates and Mortality

Developing a template matching algorithm for benchmarking hospital performance in a diverse, integrated healthcare system

Characteristics of Hospitalized Adults With COVID-19 in an Integrated Health Care System in California

The Impact of Step-Down Unit Care on Patient Outcomes After ICU Discharge

Comparison of Early Warning Scoring Systems for Hospitalized Patients With and Without Infection at Risk for In-Hospital Mortality and Transfer to the Intensive Care Unit

Red blood cell transfusions for emergency department patients with gastrointestinal bleeding within an integrated health system

What Do We Do After the Pilot Is Done? Implementation of a Hospital Early Warning System at Scale

Goldilocks, the Three Bears and Intensive Care Unit Utilization: Delivering Enough Intensive Care But Not Too Much. A Narrative Review

Association Between Peripheral Blood Oxygen Saturation (SpO2)/Fraction of Inspired Oxygen (FiO2) Ratio Time at Risk and Hospital Mortality in Mechanically Ventilated Patients

Multiyear Rehospitalization Rates and Hospital Outcomes in an Integrated Health Care System

Assessing clinical heterogeneity in sepsis through treatment patterns and machine learning

Effect of donor, component and recipient characteristics on hemoglobin increments following red blood cell transfusion

Neonatal Sepsis Evaluation: Facing the Certainty of Uncertainty

The Impact of Pharmacy-specific Predictors on the Performance of 30-Day Readmission Risk Prediction Models

Statistical Modeling and Aggregate-Weighted Scoring Systems in Prediction of Mortality and ICU Transfer: A Systematic Review

Automated Early Detection of Obstetric Complications: Theoretical and Methodological Considerations

Long-Term Outcomes Among Patients Discharged From the Hospital With Moderate Anemia: A Retrospective Cohort Study

Associations of ICU Capacity Strain with Disposition and Outcomes of Patients with Sepsis Presenting to the Emergency Department

Prevalence of infant bronchiolitis-coded healthcare encounters attributable to RSV

Risk Adjustment for Hospital Characteristics Reduces Unexplained Hospital Variation in Pressure Injury Risk

The Impact of Acute Organ Dysfunction on Long-Term Survival in Sepsis

Healthcare Utilization and Infection in the Week Prior to Sepsis Hospitalization

Accuracy of international classification of diseases, ninth revision, codes for postpartum hemorrhage among women undergoing cesarean delivery

Incorporating Longitudinal Comorbidity and Acute Physiology Data in Template Matching for Assessing Hospital Quality: An Exploratory Study in an Integrated Health Care Delivery System

Effectiveness of Respiratory Syncytial Virus Immunoprophylaxis on Bronchiolitis Hospitalizations among High-risk Infants

Hospital-Acquired Pressure Injury: Risk-Adjusted Comparisons in an Integrated Healthcare Delivery System

Comparing Hospital Processes and Outcomes in California Medicare Beneficiaries: Simulation Prompts Reconsideration

Identifying Distinct Subgroups of ICU Patients: A Machine Learning Approach

Reply to the Timing of Early Antibiotics and Hospital Mortality in Sepsis – Playing Devil’s Advocate

The Timing of Early Antibiotics and Hospital Mortality in Sepsis

Prediction of Recurrent Clostridium Difficile Infection Using Comprehensive Electronic Medical Records in an Integrated Healthcare Delivery System

Adaptation of the Acute Organ Failure Score for Use in a Medicare Population

Individual and health system variation in rehospitalizations the year after pneumonia

A Quantitative, Risk-Based Approach to the Management of Neonatal Early-Onset Sepsis

Patterns and predictors of severe postpartum anemia after Cesarean section

Implementation of a multicenter performance improvement program for early detection and treatment of severe sepsis in general medical-surgical wards

Piloting electronic medical record-based early detection of inpatient deterioration in community hospitals

Early detection, prevention, and mitigation of critical illness outside intensive care settings

Data that drive: Closing the loop in the learning hospital system