53 - Development and Validation of a Mortality Risk Score for Acute Pancreatitis Patients in the ICU

Award: ACG Auxiliary Award (Trainee)

Hassam Ali, MD¹, Vishali Moond, MD², Fnu vikash, MD³, Dushyant S. Dahiya, MD⁴, Manesh Kumar Gangwani, MD⁵, Waqas Rasheed, MD⁶, Amir Sohail, MD⁷, Umar Hayat, MD⁸, Mona A.. Ali, MD⁹, Pratik Patel, MD¹⁰, Syed Muhammad Ali Najafi, MBBS¹¹, Fouad Jaber, MD¹², Douglas Adler, MD^13
1East Carolina University Brody School of Medicine, Greenville, NC; ²Saint Peter's University Hospital, New Brunswick, NJ; ³Saint Peter's University Hospital/Robert Wood Johnson Medical School, New Brunswick, NJ; ⁴The University of Kansas School of Medicine, Kansas City, KS; ⁵University of Toledo, Toledo, OH; ⁶University of Kentucky, Lexington, KY; ⁷NYU Langone Health, Long Island, NY; ⁸Geisinger Wyoming Valley Medical Center, Wilkes-Barre, PA; ⁹Mansoura University, Mansoura, Ad Daqahliyah, Egypt; ¹⁰Mather Hospital/Hofstra University Zucker School of Medicine, Port Jefferson, NY; ¹¹Services Institute of Medical Sciences, Lahore, Punjab, Pakistan; ¹²University of Missouri - Kansas City School of Medicine, Kansas City, MO; ¹³Center for Advanced Therapeutic (CATE), Centura Health, Porter Adventist Hospital, Peak Gastroenterology, Denver, CO

Introduction: Predicting inpatient mortality for acute pancreatitis (AP) patients admitted to the intensive care unit (ICU) is crucial for optimizing treatment strategies. This study aims to develop a concise and accurate risk score model to predict 7-day and 30-day mortality in AP patients in ICU settings.

Methods: We retrospectively analyzed data from 382 AP patients admitted to the ICU. Twenty-eight variables were initially considered and reduced using LASSO regression, followed by backward elimination in multivariate models. The final model included four key predictors: INR, albumin, creatinine, and lactic acid. The model's predictive performance was evaluated using 5-fold cross-validation, yielding mean AUC values for both 7-day and 30-day mortality.

Results: The overall mortality rate was 23.2% (n=88). The LASSO regression identified 13 significant predictors, which were further refined to four key predictors in the final model. The model demonstrated a mean AUC of 0.93 (SD: 0.045) for 7-day mortality and 0.83 (SD: 0.052) for 30-day mortality. At the 7-day mark, the model exhibited a sensitivity of 78.4% and a specificity of 76.4%. For 30-day mortality, sensitivity and specificity were similarly robust. Compared to the BISAP and APACHE-II scores, our model showed superior predictive performance, with AUCs of 0.94 and 0.85 for 7-day and 30-day mortality, respectively, versus 0.66 and 0.62 for BISAP, and 0.79 and 0.71 for APACHE-II (Figure 1). High-risk patients, identified by our model, had significantly higher mortality rates, underscoring the model’s clinical utility in early identification of high-risk AP patients for timely interventions.

Discussion: Our risk score model, based on routinely available laboratory parameters, provides a reliable tool for predicting mortality in AP patients admitted to the ICU. It offers a significant improvement over current scoring systems, facilitating better resource allocation and timely interventions.




Disclosures:


Hassam Ali, MD¹, Vishali Moond, MD², Fnu vikash, MD³, Dushyant S. Dahiya, MD⁴, Manesh Kumar Gangwani, MD⁵, Waqas Rasheed, MD⁶, Amir Sohail, MD⁷, Umar Hayat, MD⁸, Mona A.. Ali, MD⁹, Pratik Patel, MD¹⁰, Syed Muhammad Ali Najafi, MBBS¹¹, Fouad Jaber, MD¹², Douglas Adler, MD¹³, 53, Development and Validation of a Mortality Risk Score for Acute Pancreatitis Patients in the ICU, ACG 2024 Annual Scientific Meeting Abstracts. Philadelphia, PA: American College of Gastroenterology.