Predictors and Outcomes of Acute Respiratory Failure Amongst the Patients Hospitalized With Acute Pancreatitis

Abstract

SESSION TITLE: Acute Lung Injury & Respiratory Failure SESSION TYPE: Original Investigation Slide PRESENTED ON: Sunday, October 29, 2017 at 01:30 PM - 03:00 PM PURPOSE: Acute respiratory failure (ARF) is one of the most common causes of mortality in patients hospitalized with acute pancreatitis. This study aims to analyze the prevalence of ARF in acute pancreatitis and determine the predictors of mortality in nationally representative sample of hospitals in the United States. METHODS: We queried the Nationwide Inpatient Sample (NIS) database from 2012 to 2014 to identify patients admitted with primary diagnosis of acute pancreatitis using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. After excluding patients with missing information on age, gender and inpatient mortality, we used ICD-9 codes to classify patients with ARF. Bivariate group comparison between patients with and without ARF with the use of weighted Chi-squared analysis for categorical data and weighted Student t test for continuous data was conducted. Multiple logistic regression analysis was used to obtain adjusted odds ratios (aOR) for predictors of mortality amongst the patients with ARF by accounting for potential confounders. All statistical tests were performed with a significance level set at 0.05. RESULTS: A total of 813,120 hospitalized patients with acute pancreatitis were identified, of which 21415 (2.63 %) had ARF. Mean age of the study population with ARF was 58.7 years with 41.3% females and 73.65% whites. Mean length of stay for patients with ARF was 16.16 days vs. 4.36 days (p<0.0001) in non-ARF patients and the mean cost of hospitalization was $46,057 in ARF vs. $8498 in non-ARF patients. ARF was associated with significantly higher mortality (17.04% vs. 0.3%, p<0.0001) even after adjusting for the confounders with aOR: 7.73 (CI: 6.01-9.94, p<0.0001). Multivariate analysis for predictors of mortality in ARF revealed age >80 (aOR: 8.55 CI: 5.23-13.95, p<0.0001), charlson comorbidity index >=3 (aOR 1.41 CI: 1.11-1.81, p = 0.0059), septic shock (aOR 3.09 CI: 2.46-3.88, p<0.0001), acute kidney injury(AKI) (aOR 2.29 CI: 1.88-2.79, p<0.0001), disseminated intravascular coagulation (DIC) (aOR 4.99 CI: 2.74-9.10,p<0.0001), patients on mechanical ventilator (aOR 2.85 CI: 2.32-3.49, p<0.0001) and need for vasopressors (aOR 2.01 CI: 1.29-3.12, p = 0.0021) while systemic inflammatory response syndrome (SIRS) (aOR 0.73 CI: 0.51-1.05, p =0.09) and gender (aOR 0.84 CI: 0.69-1.01, p =0.059) were statistically non significant predictors of mortality. CONCLUSIONS: ARF is associated with seven times higher inpatient mortality amongst patients with acute pancreatitis. Our study shows that elderly patients, AKI, mechanical ventilation, DIC, septic shock and vasopressor use are predictors of higher mortality in ARF patients while SIRS and gender are not. CLINICAL IMPLICATIONS: Acute respiratory failure in acute pancreatitis remains a major challenge that requires substantial hospital resources and is associated with higher mortality. Early recognition of predictors of worse outcomes may be vital in improving mortality of acute pancreatitis patients with ARF. DISCLOSURE: The following authors have nothing to disclose: Sajin Karakattu, Kalpit Devani, Chakradhar Reddy, Girendra Hoskere No Product/Research Disclosure Information