Business intelligence for detecting possible surgical site infections from post-cesarean section operation with a focus on antibiotic prescriptions in Ramathibodi Hospital, Thailand

Abstract

Objective: To evaluate the effectiveness of postcaesarean infection surveillance using the Power Business Intelligence (BI) program, focusing on antibiotic prescriptions. Second, to compare the workload between the traditional and new approaches. Design: A diagnostic accuracy and workload evaluation. Setting: A tertiary care university hospital in metropolitan Bangkok, Thailand. Participants: All patients who underwent cesarean section between January 1, 2019, and September 30, 2020. Method: ICD-10 diagnoses, microbiological cultures, and postcesarean section antibiotic prescriptions in 3,243 medical records were captured by the Power BI program to detect surgical site infections (SSIs). All cases underwent conventional surveillance, which independently performed by infection control nurses. All patients were under surveillance until 45 days after surgery to capture delayed SSI diagnosis. SSIs were compared with sensitivity and specificity used to evaluate the new method. The Wilcoxon signed-rank test was employed to compare workload differences between the two methods in a paired-sample design. Results: The findings demonstrated the high sensitivity (100%) (95% CI: 66.4–100%) and specificity (93%) (95% CI: 90.5–95.4%) of the Power BI method when focusing on antibiotic prescriptions between 8- and 45-days postoperation. Additionally, the Power BI infection monitoring system significantly reduced the number of cases requiring review from 452 to 39 patients (a 91% reduction), indicating a substantial decrease in workload after implementation (P < .001). Conclusion: This antibiotic prescription-based, semi-automated surveillance program significantly reduced workload, demonstrating its potential to enhance infection monitoring in postcesarean section cases.