Predicting infected bile among patients undergoing percutaneous cholecystostomy

• Published in: Cardiovascular and interventional radiology 2005-05, Vol.28 (3), p.319-325
• Main Authors: Beardsley, Shannon L, Shlansky-Goldberg, Richard D, Patel, Aalpen, Freiman, David B, Soulen, Michael C, Stavropoulos, S William, Clark, Timothy W I
• Format: Article
• Language: English
• Subjects: Adult; Aged; Aged, 80 and over; Bacterial Infections - diagnosis; Bacterial Infections - diagnostic imaging; Bacterial Infections - therapy; BILE; Bile - microbiology; BILIARY TRACT; BLOOD CELLS; CALCULI; CANDIDA; Cholecystitis - diagnostic imaging; Cholecystitis - microbiology; Cholecystitis - therapy; Cholecystostomy - methods; E coli; Exudates and Transudates; Female; Forecasting; Gallbladder - diagnostic imaging; Gallbladder diseases; Gallstones - microbiology; Gram-Negative Bacterial Infections - diagnosis; Humans; Infections; KLEBSIELLA; Leukocyte Count; Male; Middle Aged; MORTALITY; PATIENTS; PSEUDOMONAS; Radiography, Interventional; RADIOLOGY AND NUCLEAR MEDICINE; Radionuclide Imaging; Retrospective Studies; REVIEWS; Tomography, X-Ray Computed; Ultrasonography, Interventional
• ISSN: 0174-1551; 1432-086X
• DOI: 10.1007/s00270-003-0260-1

Patients may not achieve a clinical benefit after percutaneous cholecystostomy due to the inherent difficulty in identifying patients who truly have infected gallbladders. We attempted to identify imaging and biochemical parameters which would help to predict which patients have infected gallbladders. A retrospective review was performed of 52 patients undergoing percutaneous cholecystostomy for clinical suspicion of acute cholecystitis in whom bile culture results were available. Multiple imaging and biochemical variables were examined alone and in combination as predictors of infected bile, using logistic regression. Of the 52 patients, 25 (48%) had infected bile. Organisms cultured included Enterococcus, Enterobacter, Klebsiella, Pseudomonas, E. coli, Citrobacter and Candida. No biochemical parameters were significantly predictive of infected bile; white blood cell count >15,000 was weakly associated with greater odds of infected bile (odds ratio 2.0, p = NS). The presence of gallstones, sludge, gallbladder wall thickening and pericholecystic fluid by ultrasound or CT were not predictive of infected bile, alone or in combination, although a trend was observed among patients with CT findings of acute cholecystitis toward a higher 30-day mortality. Radionuclide scans were performed in 31% of patients; all were positive and 66% of these patients had infected bile. Since no patient who underwent a radionuclide scan had a negative study, this variable could not be entered into the regression model due to collinearity. No single CT or ultrasound imaging variable was predictive of infected bile, and only a weak association of white blood cell count with infected bile was seen. No other biochemical parameters had any association with infected bile. The ability of radionuclide scanning to predict infected bile was higher than that of ultrasound or CT. This study illustrates the continued challenge to identify bacterial cholecystitis among patients referred for percutaneous cholecystostomy.