Surgical site infection rates are higher among elective colorectal surgery patients receiving non–beta-lactam–based antimicrobial prophylaxis: A retrospective chart review

Abstract

We compared the odds of acquiring surgical site infection (SSI) for clean-contaminated colorectal surgeries between intravenous β-lactam–based prophylaxis (BLP) versus alternative antimicrobial prophylaxis (AAP). We calculated the odds of acquiring an SSI using logistic regression; adjusted odds ratios (ORs) with 95% confidence intervals (CIs) are reported. Increased odds of SSI were detected with AAP versus BLP (OR, 2.15; 95% CI, 1.33–3.50; P = .002).

Surgical site infection (SSI) affects 4%–17% of elective colorectal surgery patients. ^1,2 Robust evidence supports appropriate, timely administration of intravenous β-lactam–based antimicrobial prophylaxis (BLP) to reduce SSI risk. ^1,2 Guidelines recommend cefazolin with metronidazole as first line prophylaxis for clean-contaminated, elective colorectal surgery procedures. ^3,4 Alternative antimicrobial prophylaxis (AAP) is recommended for patients with penicillin allergy, but the equivalence of AAP versus BLP remains unproven. Approximately 10% of patients report penicillin allergy, and most receive AAP, although true allergy is rare. ^5,6

The objective of this study was to determine whether the odds of SSI differ between adult elective colorectal surgery patients receiving intravenous BLP and AAP. If a difference is identified, this study could challenge the notion of equivalence of BLP and AAP in preventing SSI following such procedures.

Methods

This retrospective cohort study reviewed elective colorectal surgeries between October 2009 and April 2016 in a tertiary-care hospital setting in Ontario, Canada. The ethics board approved this study. Eligible procedures included adult patients aged ≥18 years, elective clean-contaminated colorectal surgeries, traceable 30-day follow-up, and administration of guideline-recommended preoperative intravenous antimicrobial prophylaxis. ^3,4

The outcome variable was SSI within 30 postoperative days, according to the Centers for Disease Control (CDC) definition. ⁷ Chart review was performed by 6 trained abstractors. The explanatory variable was administration of BLP or AAP. Abstracted covariates included age, sex, body mass index (BMI), reported penicillin allergy, timing of antimicrobial prophylaxis administration, appropriate antibiotic redosing, perioperative normothermia, skin preparation, American Society of Anesthesiologists (ASA) class, and operative variables (ie, indication, duration, and approach). ^1–4 SSI-related outcomes included length of stay (LOS), readmission, reoperation, unplanned physician assessment, outpatient nursing care, and mortality. Clinical criteria were used for SSI diagnosis. According to the CDC, bacteriological data were not required for diagnosis. ⁷ Left-sided colon and rectal surgeries received mechanical bowel preparation. ⁸

Data were collected using Research Electronic Data Capture (REDCap) version 6.4.2 software. Full second reviews of 10% of eligible charts were performed to verify data abstraction quality. Missing data points prompted focused review to optimize data completeness. Counts and percentages describe nominal data, and medians with minimum and maximum are reported for quantitative data. Univariable analyses of patient and perioperative characteristics were performed comparing patients with and without an SSI. Covariates known to influence SSI risk were selected a priori for multivariable analyses. ^1–4 One a priori variable, the National Nosocomial Infections Surveillance (NNIS) risk index, is an aggregate score (ie, 0–3) of known SSI risk factors: ASA class, wound classification, and operative duration. ⁹ Descriptive statistics, univariable analyses, and multivariable analyses were performed. Statistical significance was set at P < .05. Odds ratios (ORs) with 95% confidence intervals (CIs) and P values are reported. SPSS version 25 software (IBM, Armonk, NY) was used for these analyses.

Results

Patient characteristics

Health records identified 2,029 bowel procedures. From these, 969 were excluded because they were urgent, emergent, contaminated, dirty, or not colorectal; 89 were excluded due to nontraceable 30-day follow-up or because of nonverifiable antimicrobial administration; 5 occurred prior to October 1, 2009. In total, 966 adult, clean-contaminated, elective colorectal surgeries involving guideline-recommended intravenous antimicrobial prophylaxis were included. Cefazolin and metronidazole, dosed per guidelines, comprised BLP. AAP patients received metronidazole plus ciprofloxacin (86.5%) or gentamicin (13.5%). Vancomycin was used in 3.6% of these patients.

We detected no differences in clinically relevant covariates among patients receiving BLP and AAP, aside from penicillin allergy status; 113 patients (11.7%) reported penicillin allergy, 111 patients (11.5%) received AAP. A collinear relationship between patients with penicillin allergy and those receiving AAP was noted, but it did not affect the multivariable analysis, which excluded allergy status. The most common indication for surgery was neoplasm (n = 711, 73.6%), followed by inflammatory bowel disease (n = 112, 11.6%), diverticular disease (n = 77, 8.0%), rectal prolapse, constipation or motility disorder (n = 29, 3.0%), other diagnosis (n = 27, 2.8%), or prior emergency surgery (n = 10, 1.0%).

Postoperative SSIs

SSIs occurred in 163 patients who underwent eligible procedures (16.9%); superficial SSIs occurred in 77 of these patients (8.0%); abscess or anastomotic leak occurred in 75 patients (7.8%); and wound dehiscence occurred in 11 patients (1.1%). Readmission was needed by 56 patients (5.8%), and reoperation was performed on 18 patients (1.9%). Organ-space SSIs occurred in 2 (0.6%) of the 6 patients who died in this study cohort.

Table 1 summarizes univariable analyses comparing patients with and without an SSI. Unadjusted analyses suggested increased odds of SSI with BMI ≥30 kg/m ² , penicillin allergy, AAP administration, duration >4 hours, open or converted-open approach, ASA class ≥3, and an NNIS score of 2. Table 2 summarizes multivariable analyses; 888 patients had complete data sets for all a priori variables. Antibiotic prophylaxis type, BMI, NNIS score, and approach were independent predictors of SSI. Administration of AAP was associated with higher odds of SSI than BLP (adjusted OR, 2.15; 95% CI, 1.33–3.50; P = .002).

Table 1. Univariable Analysis of Baseline Characteristics by Postoperative SSI in 966 Patients

Note. SSI, surgical site infection; CI, confidence interval; OR, operating room; HIV, human immunodeficiency virus; ASA, American Society of Anesthesiologists.

^a Bold P value indicates statistical significance.

Table 2. Stepwise Multivariable Analysis Using A Priori Variables by Postoperative SSI

Note. SSI, surgical site infection; CI, confidence interval; BMI, body mass index; BLP, β-lactam–based prophylaxis; AAP, alternative antimicrobial prophylaxis; NNIS, National Nosocomial Infection Surveillance. Bold P value indicates statistical significance. Hosmer-Lemeshow P = .754. A priori variables included age quartiles (referent, 1), sex (referent, male), NNIS (referent, 0; includes operative duration, ASA class and wound classification), BMI (referent, <30), antibiotic type (referent, PBAP), antibiotic timing (referent, correct), antibiotic redosed (referent, yes), approach (referent, laparoscopic), normothermia maintained (referent, yes).

Outcomes in patients who developed SSIs

Most SSIs manifested as superficial incisional SSIs (77 patients, 47.2% of SSIs). Patients with SSIs had longer postoperative LOS (8 vs 6 days; P = .003) and required more postdischarge physician assessments, outpatient nursing care, readmissions, and reoperations (all P < .001). Interestingly, AAP was associated with significantly more postdischarge physician assessments (P = .028), outpatient nursing care (P = .001), and readmissions (P = .049).

Discussion

Following elective colorectal surgery, SSIs confer significant morbidity and mortality despite provision of guideline-recommended intravenous antimicrobial prophylaxis. ^1–4 The equivalent efficacy of BLP and AAP in preventing SSI remains unproven. ¹⁰ These results suggest that BLP and AAP may not confer equivalent efficacy in preventing SSI in this population. When controlling for covariates that influence SSI incidence, AAP was associated with a higher odds of SSI than BLP, disproportionately impacting patients claiming penicillin allergy, who often receive AAP. ^5,6,10

The varied elective colorectal surgery patient population was well represented in this study. ^2,4,10 Consistent with guideline-recommended antimicrobial prophylaxis, the most common BLP applied in this study was cefazolin with metronidazole. ^3,4 The most common AAP regimens applied were metronidazole with ciprofloxacin, or metronidazole with gentamicin.

Multivariable analyses included a priori covariates known to be associated with SSI risk and often included in SSI-prevention care bundles. ^1–4 The covariates identified in multivariable analyses as independent risk factors for SSI (BMI, ASA class, approach, and duration) are recognized SSI risk factors. ^1–4 Two are nonmodifiable (BMI and ASA class), but duration and approach could be targeted to reduce SSI risk locally.

This study had several limitations. The study design did not facilitate causal conclusions. It had inadequate power for subgroup analyses as well as uncertain generalizability beyond the elective colorectal surgery population. Also, the data collection period of the study did not align with the current resurrection of the oral antimicrobial preparation debate. ⁸ These study results must be interpreted in the context of local antibiograms, which predict superior coverage of BLP over AAP due to the relative resistance of Escherichia coli and methicillin-susceptible Staphylococcus aureus (MSSA) to ciprofloxacin locally. This condition highlights the importance of applying local antibiograms in parallel with guidelines to determine appropriate regional prophylaxis recommendations. ^3–6 Finally, standard CDC SSI definitions permit inherent subjectivity in diagnosing SSI, introducing unmeasured variability.

In conclusion, AAP in elective colorectal surgery was associated with a significantly greater odds of SSI than BLP in this study. SSI was associated with longer postoperative LOS and more unplanned physician visits, outpatient nursing care, and readmissions. Our findings challenge the notion of equivalence between guideline-recommended BLP and AAP in elective colorectal surgery. These findings suggest that SSI rates in elective colorectal surgery could be decreased by optimizing the proportion of eligible patients receiving BLP.