The Role of Bowel Preparation in Colorectal Surgery

Results of the 2012-2015 ACS-NSQIP Data

Aaron L. Klinger, MD; Heather Green, MS; Dominique J. Monlezun, MD, PhD, MPH; David Beck, MD; Brian Kann, MD; Herschel D. Vargas, MD; Charles Whitlow, MD; David Margolin, MD


Annals of Surgery. 2019;269(4):671-677. 

In This Article


This study is a retrospective case-control study of the prospectively collected American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) Colectomy Targeted database from 2012 to 2015. Inclusion criteria were patients undergoing elective colorectal resection without concomitant stoma creation from 2012 to 2015 (CPT codes 44140, 44145, 44160, 44204, 44205, and 44207). Exclusion criteria included patients with known infection or sepsis at the time of operation, American Society of Anesthesiologist (ASA) Class V patients, emergency operations, and obstructive cancers. Further exclusions were made for those without antibiotic or MBP data and those with no recorded values for postoperative incisional SSI. NSQIP differentiates between superficial and deep SSI; for the purpose of this study SSI refers to any superficial or deep SSI.

The primary endpoints were a −3% independent likelihood postoperative outcomes. Sample size calculations indicated that 1873 subjects were required to assess these endpoints with 80% power and 5% error (using postestimation predictive margins for 2 sample proportions test after fully adjusted regression models). Descriptive statistics were performed for the entire sample. Bivariable analysis based on overall SSI (deep and superficial) was conducted with independent sample t test comparing means and Wilcoxon rank sum tests comparing medians for continuous variables as appropriate, and Pearson chi square test or Fisher exact test comparing proportions for categorical variables as appropriate. To provide causal inference and so approximate results from a randomized clinical trial of similar endpoint and patient sample, doubly robust propensity score adjusted multivariable logistic regression was then conducted for the study endpoints with adjustments for patient demographic, disease, and risk stratification variables significant in bivariable analysis or previously documented in the literature for being clinically and/or statistically meaningful. This propensity score-adjusted regression was augmented with a modified forward and backward stepwise regression that assessed model inclusion of every variable in the bivariable analysis.

Following the final model recommended by both stepwise algorithms, variables were added or deleted based upon prior published research, clinical intuition, and statistical model fit. The propensity score was constructed based on the likelihood of receiving both mechanical and antibiotic bowel prep using statistically and/or clinically important variables from the bivariable analysis, and balance was verified based on the final number of blocks identified (7 total). A modified doubly robust propensity score method was used at this stage with variables for consideration in the propensity score from the bivariable analysis being also used in the final regression model for the study endpoints. Stratified analysis was then conducted for colon and rectal surgery patients separately. Predictive margins were calculated for the fully adjusted final regression model to investigate significant predictors. The final regression models were reviewed by an academic physician and academic biostatistician/data scientist to ensure support by substantive clinical and statistical theory and evidence. Correlation matrix and variance inflation factor were used to ensure no multicollinearity in the final models. Hosmer-Lemeshow goodness-of-fit test was also conducted to determine whether the final models fit the data well. All regression estimates with 95% confidence intervals are reported as fully adjusted results. Statistical significance was set at 2-tailed P value <0.05. All analyses were conducted using STATA 14.2 (STATACorp, College Station, TX).