Statistical Analysis
Continuous data are presented as mean ± standard deviation for Gaussian
variables or median [interquartile range (IQR)] for non-gaussian
variables and all categorical data as number (percentage). Normality was
assessed using the Kolmogorov-Smirnov test. Normally distributed
continuous data and categorical data were compared with Pearson’s
Chi-squared test or Fisher’s exact test when 25% of available data
points had expected values <5. Non-Gaussian distributions were
evaluated using Mann-Whitney U test.
Multivariable Cox Proportional Hazards modeling was used to identify
predictors for mortality. Competing risk methods by Fine and Grey
modeling was used to model all-cause readmission. Multivariable logistic
regression was used to identify risk-adjusted predictors for GI
complications. Co-variables were first assessed by univariable model
with a threshold of P<0.2 for inclusion into the multivariable
model.
Propensity score matching on a 1:1 basis using nearest neighbor matching
without replacement and caliper setting of 0.2 of the standard deviation
of the logit propensity score. Patients were matched on preoperative
demographics, comorbidities, and operative characteristics. This
resulted in 224 cases, 112 with and 112 without postoperative GI
complications.