In-hospital cardiac arrest (IHCA) is a significant public health problem, afflicting over 200,000 patients in the United States annually with a mortality rate of approximately 80%. The majority of these patients show signs of clinical deterioration in the hours before the event. This has led to the development of vital sign-based early warning scores designed to detect high-risk patients before IHCA to trigger life-saving interventions. However, the vast majority of these risk scores were created subjectively in individual hospitals and have shown limited accuracy for detecting adverse outcomes. Developing an accurate risk score to detect patients at highest risk of IHCA is essential to decreasing preventable in-hospital death. In my prior work, I completed several studies investigating the accuracy of vital signs for predicting IHCA. These studies, previous literature, and my preliminary data have resulted in the following conclusions: 1) statistically developed risk scores are more accurate than previously published risk scores, 2) multicenter data is needed to create the most accurate and generalizable risk score, 3) additional data, such as laboratory results, will likely improve the accuracy of risk scores, and 4) a cutting-edge method for developing prediction models, called machine learning, may result in more accurate risk scores. Importantly, significant improvement in accuracy leads to better identification of patients at highest risk of IHCA and decreased resource utilization. Therefore, in this grant proposal I aim to develop and validate IHCA prediction models using different statistical techniques in a multicenter database and then estimate the impact of the most accurate risk score using simulation studies. I will do this by firt developing prediction models using classic survival analysis methods (Aim 1a) and machine learning methods, such as neural networks and decision trees (Aim 1b). Then, I will compare the models I develop to the most accurate previously published risk scores in Aim 2. Finally, I will investigate the impact of the most accurate model from Aim 2 on patient outcomes using simulation modeling (Aim 3). Completion of this proposal will result in a validated IHCA risk score that can be implemented in the electronic health record to trigger life- saving interventions to decrease preventable in-hospital death. In addition, this career development award will provide critical data to inform future R01-level awards, including a clinical trial to investigate he impact of the developed prediction model on patient outcomes. I will complete this project under the direct supervision of my mentor (Dr. David Meltzer), co-mentor (Dr. Dana Edelson), and the rest of my advisory team (Drs. Jesse Hall, Robert Gibbons, and Michael Kattan). Together, this multidisciplinary team brings nationally renowned expertise in in-hospital cardiac arrest, outcomes research, critical care, and clinical prediction modeling. In addition, they serve as Chairs of the Section of Hospital Medicine (Dr. Meltzer), Section of Pulmonary and Critical Care (Dr. Hall), and Quantitative Health Sciences at the Cleveland Clinic (Dr. Kattan), and Directors of the Center for Health and the Social Sciences (Dr. Meltzer), Center for Health Statistics (Dr. Gibbons), and Clinical Research for the Emergency Resuscitation Center (Dr. Edelson). The mentorship, expertise, and resources that they provide will ensure my success as I grow into an independent physician-scientist. My career goal is to become an independent critical care outcomes researcher with a focus on developing prediction models for clinical deterioration that will improve patient outcomes. To accomplish this long-term goal, I have three short-term goals: (1) to gain expertise in the development and implementation of clinical prediction models, (2) to create an IHCA prediction model that will identify high-risk patients on the wards to trigger life-saving interventions, and (3) to gain expertise in simulation modeling in order to study the impact of the developed prediction model. To accomplish these goals, I will build upon the foundation I developed when earning my Master's Degree in Public Health and during my initial training in the PhD program in the Department of Health Studies. Although my training to date has provided me with a strong background in epidemiology and biostatistics, further advanced training in biostatistics is crucial for my development into a successful independent researcher. An integrated program of didactic coursework, seminars, research activities, and conference participation will span the duration of the award. By accomplishing my three short- term goals, I will develop unique skills that will allow me to become a successful independent researcher. Specifically, the expertise I will gain in prediction model development, implementation, and simulation modeling can be applied not only to IHCA research but also to other areas of critical care medicine. In addition, completion of these goals will result in a validated IHCA prediction model that I will study in future implementation and cost-effectiveness studies and will serve as a basis for future R01-level grant submissions.
Over 200,000 in-hospital cardiac arrests occur in the United States each year, and studies suggest that many of these events may be preventable if the clinical warning signs can be identified and acted upon quickly. However, the vast majority of tools used to identify patients at high risk of cardiac arrest were created subjectively and have limited accuracy. Development of a statistically derived risk tool is essential to detect at- risk patients accurately and early in order to provide the best opportunity to improve patient outcomes and reduce preventable in-hospital death.
|Sanchez-Pinto, L Nelson; Venable, Laura Ruth; Fahrenbach, John et al. (2018) Comparison of variable selection methods for clinical predictive modeling. Int J Med Inform 116:10-17|
|Lyons, Patrick G; Edelson, Dana P; Churpek, Matthew M (2018) Rapid response systems. Resuscitation 128:191-197|
|Rojas, Juan C; Carey, Kyle A; Edelson, Dana P et al. (2018) Predicting Intensive Care Unit Readmission with Machine Learning Using Electronic Health Record Data. Ann Am Thorac Soc 15:846-853|
|Blagev, Denitza P; Collingridge, Dave S; Rea, Susan et al. (2018) The Laboratory-Based Intermountain Validated Exacerbation (LIVE) Score Identifies Chronic Obstructive Pulmonary Disease Patients at High Mortality Risk. Front Med (Lausanne) 5:173|
|Sanchez-Pinto, L Nelson; Luo, Yuan; Churpek, Matthew M (2018) Big Data and Data Science in Critical Care. Chest 154:1239-1248|
|Churpek, Matthew M; Snyder, Ashley; Twu, Nicole M et al. (2018) Accuracy Comparisons between Manual and Automated Respiratory Rate for Detecting Clinical Deterioration in Ward Patients. J Hosp Med 13:486-487|
|Shappell, Claire; Snyder, Ashley; Edelson, Dana P et al. (2018) Predictors of In-Hospital Mortality After Rapid Response Team Calls in a 274 Hospital Nationwide Sample. Crit Care Med 46:1041-1048|
|Han, Xuan; Edelson, Dana P; Snyder, Ashley et al. (2018) Implications of Centers for Medicare & Medicaid Services Severe Sepsis and Septic Shock Early Management Bundle and Initial Lactate Measurement on the Management of Sepsis. Chest 154:302-308|
|Parker, William F; Anderson, Allen S; Hedeker, Donald et al. (2018) Geographic Variation in the Treatment of U.S. Adult Heart Transplant Candidates. J Am Coll Cardiol 71:1715-1725|
|Adegunsoye, Ayodeji; Oldham, Justin M; Chung, Jonathan H et al. (2018) Phenotypic Clusters Predict Outcomes in a Longitudinal Interstitial Lung Disease Cohort. Chest 153:349-360|
Showing the most recent 10 out of 38 publications