A retrospective cohort analysis was performed on all patients undergoing pancreatic surgery at the First Affiliated Hospital of Nanjing Medical University, a Grade-A tertiary hospital with 4500 beds. In this study, when the information system generates a clinician's order, the clinician has the option to specify the purpose of the medication, whether it is therapeutic or preventive. The empirical use of medication is a therapeutic purpose and is outside the scope of this study.

Patients undergoing the first pancreatic surgery from January 2022 to December 2022 were included in our study. The flowchart of this study is displayed in Fig. 1. A total of 14 patients who did not receive pancreatic surgery in the surgical records were excluded from 1087 patients. Finally, a total of 1073 patients were included and allocated into the non-AP after 48 h postoperatively group (n = 963; patients did not receive any AP after 48 h to 30 days postoperatively) and the AP after 48 h postoperatively group (n = 110; patients received AP at least one dose after 48 h to 30 days postoperatively). Two groups are distinguished in Fig. 2.

Flowchart of patient exclusion and inclusion

Range distinction between AP after 48 h postoperatively group and non-AP after 48 h postoperatively group

The Xinglin Real-Time Nosocomial Infection System and iih System were used to collect the demographic data of patients, including patient characteristics (age, sex, diabetes, and length of hospital stay), surgical variables [surgical category, surgical approach, surgical time, emergency, American Society of Anesthesiology (ASA) score (according to the principles of surgical risk assessment, I and II, score 0; III–V, score 1), National Nosocomial Infections Surveillance (NNIS) score, surgeon, inpatient department, and intraoperative blood loss], and Antibiotic (any AP, intraoperative redosing, AP within 48 h postoperatively, and AP days).

Outcomes included SSIs (the primary outcome) within 30 days and other HAIs (the secondary outcome) [postoperative sepsis, postoperative pneumonia, pelvic and abdominal tissue infections (infections occur more than 30 days after surgery), and urinary tract infections, and others that occurred during the hospital stay after pancreatic surgery]. HAIs were diagnosed according to the Diagnostic Criteria of Nosocomial Infection (Trial) issued by the Ministry of Health in 2001 [23].

Data were summarized as mean ± standard deviation (SD) or frequencies (percentages), as appropriate. The Chi-square or Fisher’s exact test and t-test or Mann–Whitney U test were utilized for descriptive statistics, as appropriate.

Continuous variables were classified into two categorical variables based on the SSI-risk age (65 years), the length of hospital stay of patients receiving pancreatic surgery (14 days), the 75% time of pancreatic surgery (4.88 h), and the days of prophylaxis (one day), respectively. Intraoperative blood loss was categorized into three categorical variables.

Univariate and multivariate logistic regression analyses were performed to assess the associations between AP after 48 h postoperatively and HAIs. The adjusted covariates in model 1 were age (continuous), sex (male or female), and diabetes (yes or no). In model 2, the adjusted covariates were age (continuous), sex (male or female), diabetes (yes or no), surgical category (pancreaticoduodenectomy, distal pancreatectomy, and others), surgical approach (non-endoscopic surgery and endoscopic surgery), surgical time (continuous), emergency (yes or no), ASA score (0 and 1), NNIS score 1 (0 and 1 point) and 2 (2 and 3 points), inpatient department (pancreatic center and nonpancreatic center), surgeon (Doctor 1, Doctor 2, Doctor 3, Doctor 4, Doctor 5, Doctor 6, and others), intraoperative blood loss (continuous), any AP (yes or no), intraoperative redosing (required but no redosing, required and redosing, or not required), AP within 48 h postoperatively (yes or no), and AP days (continuous).

A series of analyses were conducted to examine whether there was effect modification by age, sex, diabetes, surgical category, ASA score (0 and 1), NNIS score 1 (0 and 1 point) and 2 (2 and 3 points), inpatient department, surgical time, intraoperative blood loss, intraoperative redosing, and AP within 48 h postoperatively. For these analyses, we included an interaction term in the primary model between AP after 48 h postoperatively and these variables.

Analyses were performed using Statistical Product and Service Solutions (SPSS) (version 23.0; IBM Corp. Armonk, NY, USA), R software (version 3.6.0; R Core Team), EmpowerStats (www.empowerstates.com), and Graph Pad Prism 8.0 (San Diego, CA, USA). A P value of < 0.05 indicated a statistically significant difference.