Infected versus sterile abdominal fluid collections in postoperative CT: a scoring system based on clinical and imaging findings.

Abstract

Characterization of intraabdominal fluid collections as postoperative complication is a challenging task. The aim was to develop and validate a new score to differentiate infected from sterile postoperative abdominal fluid collections and to compare it with a published score. From May to November 2015, all patients with postoperative CT and C-reactive protein (CRP) 24 hours before CT-guided drainage were retrospectively included (Group A). HU, gas entrapment and wall enhancement of fluid collections were evaluated in the CT. All parameters were correlated with microbiology. To validate the score and to compare it with a published score, a second patient cohort was retrospectively recruited (Group B; January 2013-April 2015; December 2015-September 2016). In Group A (50 patients), univariate analysis confirmed that the four parameters were significantly associated with infected fluid collections. Based on binary logistic regression analysis, a score from 0 to 11 was developed (CRP </≥ 150 mg/l: 0/4 points; HU </≥ 20: 0/2 points; wall enhancement no/yes: 0/2 points; gas entrapment no/yes: 0/3 points). The best cutoff to diagnose infected fluid collections was ≥ 5 points (sensitivity 85%, specificity 79%, PPV 82%, NPV 79%). In Group B (425 patients), this score yielded sensitivity, specificity, PPV and NPV of 93%, 80%, 90% and 86%, respectively. For the published score, values were in the same range (93%, 44%, 77%, 77%). The score provides good discrimination between infected and sterile postoperative abdominal fluid collections. It yields comparable accuracy as the published score.