Background
Surgical site infections (SSIs) are a major postoperative complication after abdominal surgery, especially in the colorectal field [
1]. With a reported incidence of over 20%, SSIs significantly increase the length of stay (LOS), readmission rate, expenses, and mortality [
2,
3]. Therefore, the identification of an effective method of reducing SSI incidence is critically important [
4]. Colonic bacterial florae are considered to be the major cause of SSIs after elective colorectal procedures, but the most effective means of decreasing this bacterial load remains under debate [
5]. Preoperative mechanical bowel preparation (MBP) was first utilized by surgeons, as it can theoretically remove stool content and associated bacterial load within the bowel and surgical field, thus reducing the risk of SSIs [
6]. More recently, as antibiotics have come to be widely utilized, the preoperative administration of unabsorbed oral antibiotics (OAs) in combination with MBP was widely conducted [
2,
7].
Multiple trials have been performed to explore the best bowel preparation strategies, but their results remain controversial [
8‐
10]. Since 2005, several RCTs and meta-analyses have demonstrated MBP alone was not associated with a reduced incidence of SSIs related to patients that did not undergo MBP, whereas MBP patients exhibited paradoxical increases in postoperative ileus, anastomotic leakage, and other complications [
11‐
14]. Recently, the merit of OA and MBP has been rediscovered in several related retrospective studies, which demonstrated a significant decrease in the rate of SSIs [
15‐
17]. However, as information in these trials were exacted from national databases without any detailed matching between patient groups, the existence of bias in these trials may affect the validity of their results. Furthermore, none of these studies assessed the relative prophylactic effects of the novel MBP mode in right or left-side colorectal surgery. Herein, we report on our experiences in a single-center comparison of MBP + OA with MBP alone, assessing the rates of prophylactic combinations between groups via propensity score matching and stratification.
Method
Study population
This retrospective study was approved by the Ethics committee of the Third Affiliated Hospital of Sun Yat-sen University. Eligible patients were identified by searching the database of the Gastrointestinal Surgery Centre, Third Affiliated Hospital of Sun Yat-sen University from 2011 to 2017. Patient inclusion criteria were as follows: (1) patient underwent elective colorectal resection to treat a malignancy, (2) patient baseline characteristics and operative information were available, and (3) MBP was performed before surgery, with or without OA.
Patient exclusion criteria were as follows: (1) emergency surgery; (2) MBP was not conducted due to ileus or patient refusal; (3) enough data was not available; (4) colorectal resection was performed due to benign disease; (5) the procedure was accompanied by other procedures that had the potential to contaminate the incision, such as cholecystectomy or appendectomy; and (6) patients underwent neoadjuvant radiotherapy before surgery.
The primary and secondary aims of the study have been stated in the latest manuscript. The primary aim was to evaluate prophylactic function of preoperative OA combined with MBP vs MBP alone in postoperative SSI incidence. The secondary aim was to explore the potential benefit on length of hospital stays of OA + MBP mode compared with simple MBP.
Application of preoperative antibiotics was under surgeons’ decisions; no patients withdrew during the study period. Either polyethylene glycol or magnesium sulfate was adopted as a laxative 1 day before surgery. Clyster was conducted on surgery morning. Streptomycin 1 g plus metronidazole 0.2 g was prescribed 3 times a day for 3 days before surgery in the OA + MBP group patients.
Intravenous antibiotic prophylaxis was based on local guidelines and resistance profiles: most of the patients received cefmetazole 2 g intravenous drip 30 min before incision and once every 12 h until 48 h after surgery. Patients with penicillin or cephalosporin allergy were given clindamycin 0.6 g twice a day. If the surgical procedure lasted more than 180 min, a booster dose of antibiotic was administered.
Outcomes
Based on the preparation procedures employed, patients were divided into a mechanical preparation plus oral antibiotics group (MBP + OA group) and a simple MBP group. The following demographic, clinical, and pathological information were extracted from the database: age, gender, BMI, comorbidities, American Society of Anesthesiologists (ASA) score, operative duration, laparoscopic or laparotomy approach, surgical site, neoadjuvant chemotherapy, combination with multi-organ resection, TNM stage, and preoperative serum albumin level. Outcomes of interest were length of hospital stay (LOS), expense, and rates of postoperative complications, which included anastomotic leakage, SSIs, postoperative ileus, respiratory/urinary infection, deep vein thrombosis (DVT), and postoperative Clostridium difficile infection (CDI).
Statistical analysis
Frequencies were presented for categorical variables, and means ± standard deviation were given for continuous variables. Pearson’s χ2 or Fisher’s exact tests were used to analyze categorical variables. Student’s t tests were used for analyzing normally distributed data; otherwise, Mann-Whitney U tests were used for continuous variables. Propensity score matching was performed for minimizing confounding based on TNM stage, laparoscopic or laparotomy approach, ASA score, gender, BMI, and neoadjuvant chemotherapy. A multivariate logistic regression model was used to identify independent SSI risk factors, and a stepwise forward method was used for variable selection (inclusion p < 0.05; permanence p < 0.1). The fit for this logistic regression was tested with the Hosmer and Lemeshow test. All data analyses were performed with SPSS v22 (Armonk, NY: IBM Corp).
Discussion
Surgical site infection is one of the most common complications after colorectal surgery, substantially increasing patient morbidity and expenses [
1‐
3]. With the large burden of bacteria in the bowel, elective colorectal resections are associated with particularly high rates of SSIs [
2,
4]. Bowel preparation modes prior to elective colorectal surgery have been varied for decades and aim to reduce the SSIs [
8‐
14,
18].
MBP was initially performed preoperatively with the goal of reducing bacterial burden and human fecal content and to thereby decrease SSI rates [
6]. However, as mentioned above, subsequent research demonstrated that MBP alone failed to achieve this objective [
2,
6], instead causing paradoxical complications [
11‐
14]. It has been proposed that when implemented in concert with OA administration, the MBP-mediated reduction in bacterial burden may guarantee better OA delivery to the entire length of the colon, improving prophylactic activity [
19‐
22]. Furthermore, with the advent of the ERAS era, surgeons have sought to minimize perioperative physiologic perturbations, leading to increasing concern regarding and abandonment of the use of MBP or OA. One such concern is that the combinational preparation may prolong preoperative hospital stays and expenses, in addition to causing increased patient discomfort and reduced compliance. As such, there is a need to determine whether the combination of MBP + OA yields better patient outcomes. Recently, the combination of OA and MBP has been evaluated in several retrospective studies which demonstrated a significant decrease in the rate of SSIs [
15‐
17]. However, dietary structure, BMI, lifestyle, and colonic flora differ between people from Eastern and Western nations. Furthermore, no previous studies have evaluated the value of OA in a site-specific manner in the colon/rectum. As such, we performed a propensity matching retrospective study with subgroup analyses in order to further evaluate the prophylactic value of OA.
The current study revealed that the application of MBP + OA can significantly decrease the overall incidence of SSIs (10.59% vs 16.56%, p = 0.03) and expenses (56.74 ± 16.60 vs 66.73 ± 25.66 kRMB, p < 0.05) relative to MBP alone in patients undergoing elective colorectal resection. However, postoperative LOS was longer in the MBP + OA group (10.10 ± 5.19 days vs 9.20 ± 5.01 days, p = 0.03). The incidence of anastomotic fistula, postoperative ileus, urinary infection, Clostridium difficile infection, pulmonary infection, hemorrhage, DVT, and 30-day readmission was comparable in both groups. Owing to the retrospective nature of this analysis, several patient baseline characteristics were different between groups, including serum albumin, surgical approach, neoadjuvant chemotherapy history, and rectal resection proportion, potentially confounding our results. As such, a propensity score matching analysis was conducted to normalize patient groups according to TNM stage, surgical approach, ASA score, gender, BMI, tumor location, and neoadjuvant chemotherapy history with a 1:1 ratio. This led us to analyze a total of 428 patients in the final data analysis, which revealed that overall SSI incidence (7.01% vs 15.89%, p = 0.004), superficial SSI incidence (2.34% vs 7.01%, p = 0.03), and hospitalization expense (56.98 ± 16.58 vs 65.27 ± 20.13 kRMB, p < 0.05) were significantly higher in the MBP alone group. Postoperative LOS was comparable between both groups, potentially because patients with superficial SSI were first discharged and undergoing outpatient wound care, potentially influencing the LOS results in our center.
Previous studies have indicated that in the right hemi-colon, the concentration of bacteria ranges from 10
6 to 10
7 bacteria/g of stool content, whereas these numbers rise to 10
11–10
12 bacteria/g in the rectosigmoid region [
6,
19]. Several studies have demonstrated a lower risk of SSIs for right colon resections [
23,
24]. Therefore, the proposal to forgo the use of OA prior to right hemicolectomy was raised in the ERAS era, without formal demonstration of the outcomes of such an approach. Hence, in the present study, we performed for the first time a stratified propensity analysis of patient outcomes for right hemicolectomy and left-side (descending, sigmoid colon, and rectum) colorectal resection subgroups. The preventative function was especially prominent in left-side patients, in whom the incidence of overall SSIs (7.01% vs 15.89%,
p = 0.004), superficial SSI (2.34% vs 7.01%,
p = 0.03), and hospitalization expenses (56.98 ± 16.58 vs 65.27 ± 20.13 kRMB,
p < 0.05) were significantly higher relative to the MBP alone group. However, similar improvements in outcomes upon OA administration were not observed in the right hemicolectomy subgroup. These findings are consistent with previous proposals, underscoring the need to tailor bowel preparation strategies based on the surgical site in a given patient.
Our logistic regression model demonstrated that SSI incidence was associated with age, surgical duration, and the application of OA. Together our findings demonstrate the key value of OA in bowel preparation for patients undergoing colorectal surgery, suggesting that individual preoperative evaluation may help avoid unnecessary bowel preparation and minimize postoperative morbidity.
There are several strengths to our study. Our detailed database provided us with a sufficient sample size to analyze the association between bowel preparation mode and postoperative outcomes. The propensity analysis helped minimize the bias in the baseline characteristics of enrolled patients and thus enhanced the generalizability of our findings. However, several limitations still exist in our trial. First, this was a single-institution study, which limits its external validity. Second, the bowel preparation modes were reviewed through the database and medical records, and as such, the compliance and quality could not be evaluated. Third, as in other retrospective studies, historical bias may still exist despite our propensity analysis.
With increasing adoption of ERAS, fewer patients will undergo bowel preparations. However, as this was a retrospective analysis, all reviewed patients were from an era prior to ERAS application. Therefore, comparisons between no bowel preparation, simple MBP, OA + MBP, and simple OA could not be conducted. We are now in the process of conducting a randomized controlled trial of the effects of MBP + OA compared with MBP alone (NCT03856671). With the application of ERAS, future assessments of patients who receive no MBP or simple OA will be conducted, thereby helping to overcome the limitations of the present study.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.