Correlation between the Predictive Accuracy of Computed Tomography Severity Index and Clinical Metrics in Acute Pancreatitis at a Tertiary Care Hospital Lahore

Abstract

Acute pancreatitis was a common clinical emergency and presents with a vast spectrum of severity and clinical outcomes. The Computed Tomography Severity Index (CTSI) was widely used to evaluate extent of pancreatic inflammation and necrosis. Objective: To compare the CTSI with the clinical severity of acute Pancreatitis in local settings. Methods: It was a retrospective cohort study done on 136 cases diagnosed with acute pancreatitis between 2017 to 2023 at Lahore General Hospital, Lahore, Pakistan.  Patients received contrast-based Computed Tomography (CT) within 30 days of onset. CTSI scores were independently assessed by two experienced radiologists. Clinical severity was categorized as mild, moderate and severe pancreatitis. Statistical analysis was done with SPSS 26.0 which involved descriptive, correlational statistics, sensitivity and specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), predictive Accuracy along with ROC curve analysis and Cohen’s kappa statistic. Results: The patients were 74.3% males and had a median age of 51 years. CTSI demonstrated 79.37% sensitivity, 83.56% specificity, 80.65% PPV and 82.43% NPV in determining clinical severity as mild, moderate and severe with a predictive accuracy of 81.62%. Cohen’s kappa of 0.72 reflected substantial agreements between the CTSI and clinical severity assessment.   Under the ROC Curve (AUC) the area was 0.87, showing an excellent diagnostic performance. Conclusion: CTSI provides a moderate to fair agreement with clinical severity assessments in acute pancreatitis. It effectively differentiates between mild, moderate and severe cases, supporting its assessment and management.