Can yeast isolation in peritoneal fluid be predicted in intensive care unit patients with peritonitis?* Hervé Dupont, MD; Agnes Bourichon; Catherine Paugam-Burtz; Jean Mantz; Jean-Marie Desmonts D espite constant improvement in the standard of intensive care unit (ICU) care over the last 20 yrs, mortality attrib- uted to severe intraabdominal sepsis re- mains as high as 40% (1). A prominent feature of peritonitis requiring intensive care is the frequency of polymicrobial flora recovered from peritoneal fluid. The presence of intraperitoneal yeast is de- tected in 5% to 22% of cases in this setting (1– 4). Interestingly, the presence of yeast as a causative agent of peritonitis has been reported to be associated with an increased overall mortality rate as high as 60% to 70% (5). The reason for this extremely high mortality associated with fungal peritonitis has not yet been elucidated. Inadequate early antimicrobial ther- apy has been associated with a poorer outcome in critically ill patients with peritonitis (1). In the ICU, 17% of noso- comial isolates are yeasts, as reported in the European Prevalence of Infection in Intensive Care study (6). However, only half of the patients were treated by anti- fungal therapy in this study, essentially because of the lack of consensus concern- ing the pathogenicity of yeast recovered from bacteriologic samples. This is par- ticularly true when yeast is isolated from the urinary tract or specimens are ob- tained from protected pulmonary sam- ples. Similarly, the pathogenicity of yeast in peritoneal fluid also remains contro- versial (7). In the recently published practice guidelines for the treatment of candidiasis, antifungal therapy was rec- ommended for documented peritoneal di- alysis catheter yeast infections and peri- tonitis related to intraabdominal leakage of fecal material. Surgical repair and drainage are also recommended in this case (8). Although risk factors for candidemia have been extensively described (9, 10), no data are available yet on the risk factors for developing fungal peritoni- tis. The aim of this study was first to determine independent risk factors of yeast isolation in peritoneal fluid of ICU patients with severe peritonitis and sec- ond to generate and validate a predic- tive score of yeast isolation in this patient subpopulation. MATERIALS AND METHODS Patient Selection. A retrospective study in- cluding all adult patients with a diagnosis of peritonitis requiring admission to a surgical ICU for organ failure was performed between 1994 and 1999. All consecutive patients in whom peritonitis was diagnosed were prospec- tively included in a database and their charts were reviewed retrospectively. Peritonitis was diagnosed on the basis of both macroscopic findings and positive culture of peritoneal fluid collected during an operation. The exclu- sion criteria were acute infected pancreatitis, acute perforation caused by trauma 6 hrs, and primary peritonitis, such as dialysis cath- eter infection and ascites infection. A prospec- tive validation cohort of ICU patients with the same diagnosis was also evaluated in 2000. Surgery. The surgical management of se- vere peritonitis was based on three fundamen- tal principles (11). The first one was the elim- ination of the source of infection, the second one the reduction of bacterial inoculum of the peritoneal cavity, and the third one the pre- vention of persistent or recurrent intraab- dominal infection. The source control was performed through midline laparotomy for all patients. Primary anastomosis was not per- formed and enterostomy preferred. The sec- *See also p. 976. From the Anesthésie Réanimation Chirurgicale, Hopital Bichat-Claude Bernard, Paris, France. Copyright © 2003 by Lippincott Williams & Wilkins DOI: 10.1097/01.CCM.0000053525.49267.77 Objective: To generate and validate a predictive score of yeast isolation based on independent risk factors of yeast isolation in intensive care unit patients with peritonitis. Design: Retrospective cohort study to determine independent risk factors of yeast isolation, generation of the score, and vali- dation in a prospective cohort of patients with peritonitis. Setting: Tertiary-care, university-affiliated hospital. Patients: Two hundred twenty-one patients with peritonitis hospitalized in a surgical intensive care unit between 1994 and 1999 for the retrospective cohort and 57 patients in the prospec- tive cohort (2000). Measurements and Main Results: Four independent risk fac- tors of yeast isolation in peritoneal fluid (similar odds ratio) were found in the retrospective cohort: female gender, upper gastro- intestinal tract origin of peritonitis, intraoperative cardiovascular failure, and previous antimicrobial therapy at least 48 hrs before the onset of peritonitis. A score based on the number of risk factors was constructed (grade A  zero or one risk factor, grade B  at least two risk factors, grade C  at least three risk factors, and grade D  four risk factors), and validated in the prospective cohort. For a grade C score, sensitivity was 84%, specificity was 50%, positive and negative predictive values were 67% and 72%, respectively, and overall accuracy was 71%. Conclusions: Four independent risk factors of yeast isolation in the peritoneal fluid were identified in critically ill surgical patients with peritonitis. The presence of at least three of these factors (grade C score) was associated with a high rate of yeast detection. This approach could be helpful to initiate early antifungal therapy in this patient population. (Crit Care Med 2003; 31:752–757) KEY WORDS: peritonitis; Candida; risk factors; score; intensive care 752 Crit Care Med 2003 Vol. 31, No. 3