Modified Bedside Index for severity in acute pancreatitis (BISAP) score validation in the national inpatient sample database

Abstract

PurposeThe aim of this study was to build and validate modified score to be used in the healthcare cost and utilization project databases for further classification of acute pancreatitis (AP). Materials and methodsThe National Inpatient Sample database for the years 2016-2019 was queried for all primary adult discharge diagnoses of AP. An mBISAP score system was created utilizing the ICD-10CM codes for pleural effusion, encephalopathy, acute kidney injury, systemic inflammatory response, and age >60. Each was assigned a 1-point score. A multivariable regression analysis was built to test for mortality. Sensitivity and specificity analyses were performed for mortality. ResultsA total of 1,160,869 primary discharges for AP were identified between 2016 and 2019. The pooled mortality rate was: 0.1%, 0.5%, 2.9%, 12.7%, 30.9% and 17.8% (P ​< ​0.01), respectively for scores 0 to 5. Multivariable regression analysis showed increasing odds of mortality with each one-point increment: mBISAP score of 1 (adjusted odds ratio [aOR] 6.67; 95% confidence interval [CI] 4.69-9.48), score of 2 (aOR 37.87; 95% CI 26.05- 55.03), score of 3 (aOR 189.38; 95% CI 127.47-281.38), score of 4 (aOR 535.38; 95% CI 331.74-864.02), score of 5 (aOR 184.38; 95% CI 53.91-630.60). Using a cut-off of ≥3, sensitivity and specificity analyses reported 27.0% and 97.7%, respectively, with an area under the curve (AUC) of 0.811. ConclusionIn this 4-year retrospective study of a US representative database, an mBISAP score was constructed showing increasing odds of mortality with each 1-point increase and a specificity of 97.7% for a cut-off of ≥3.