Purpose: We investigated the prognostic significance of postoperative infections for the outcome
of 192 patients with colon cancer.
Methods: The 5-year survival rates were analyzed by the Kaplan-Meier technique. Univariate
and multivariate analyses were done to evaluate prognostic variables using Cox's proportional
hazard model.
Results: Forty-three patients developed deep incisional or organ/space surgical site infections.
The groups with and without infection were comparable. Multivariate analysis showed
that only Dukes' stage (p = 0.048) and postoperative infection (p = 0.011) were independently
associated with outcome. In patients with infective complications, the survival rate was significantly
lower than in subjects without infection (log rank p = 0.0004).
Conclusions: These results stress the importance of evaluating variables other than the classical
tumor stage in predicting long-term cancer outcome.
