Sailhamer EA, Carson K, Chang Y, Zacharias N, Spaniolas K, Tabbara M, Alam HB, DeMoya MA, Velmahos GC. Fulminant Clostridium difficile ColitisPatterns of Care and Predictors of Mortality. Arch Surg. 2009;144(5):433-439. doi:10.1001/archsurg.2009.51
Copyright 2009 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2009
There exist predictors of mortality and the need for colectomy among patients with fulminant Clostridium difficile colitis.
Academic tertiary referral center.
We reviewed the records of 4796 inpatients diagnosed as having C difficile colitis from January 1, 1996, to December 31, 2007, and identified 199 (4.1%) with fulminant C difficile colitis, as defined by the need for colectomy or admission to the intensive care unit for C difficile colitis.
Main Outcome Measures
Risk of inpatient mortality was determined by multivariate analysis according to clinical predictors, colectomy, and medical team.
The inhospital mortality rate for fulminant C difficile colitis was 34.7%. Independent predictors of mortality included the following: (1) age of 70 years or older, (2) severe leukocytosis or leukopenia (white blood cell count, ≥35 000/μL or <4000/μL) or bandemia (neutrophil bands, ≥10%), and (3) cardiorespiratory failure (intubation or vasopressors). When all 3 factors were present, the mortality rate was 57.1%; when all 3 were absent, the mortality rate was 0%. Patients who underwent colectomy had a trend toward decreased mortality rates (odds ratio, 0.49; 95% confidence interval, 0.21-1.1; P = .08). Among patients admitted primarily for fulminant C difficile colitis, care in the surgical department compared with the nonsurgical department resulted in a higher rate of operation (85.1% vs 11.2%; P < .001) and lower mortality rates (12.8% vs 39.3%; P = .001). Patients admitted directly to the surgical department had a shorter mean (SD) interval from admission to operation (0 vs 1.7 [2.8] days; P = .001).
Despite awareness and treatment, fulminant C difficile colitis remains a highly lethal disease. Reliable predictors of mortality exist and should be used to prompt aggressive surgical intervention. Survival rates are higher in patients who were cared for by surgical vs nonsurgical departments, possibly because of more frequent and earlier operations.
Clostridium difficile colitis is the most common nosocomial infection of the gastrointestinal tract, affecting more than 3 million patients per year in the United States, and is associated with significant morbidity and mortality.1- 3Clostridium difficile is an anaerobic, spore-forming, gram-positive bacillus that in pathogenic strains produces 2 important exotoxins that cause diarrhea and colitis. When established in the colon, it causes a spectrum of disease severity, ranging from asymptomatic carriage to severe diarrhea, pseudomembranous colitis, toxic megacolon, colonic perforation, or death. Fulminant C difficile colitis has been broadly defined as C difficile colitis with significant systemic toxic effects and shock, resulting in need for colectomy or death, and occurs in approximately 3% to 5% of patients with C difficile colitis, with a marked increase in severity and mortality during the last several years.4- 6 The objectives of this study are to analyze characteristics of and treatment patterns in patients with fulminant C difficile colitis and to identify predictors of mortality.
All patients with C difficile colitis admitted at the Massachusetts General Hospital during a 12-year period (January 1, 1996, to December 31, 2007) were identified by the discharge diagnostic code for C difficile or pseudomembranous colitis (International Classification of Diseases, Ninth Revision [ICD-9] code 008.45).7 We defined fulminant C difficile colitis as the need for intensive care unit (ICU) admission or colectomy owing to C difficile colitis. We included only those patients who were admitted to the ICU primarily for C difficile colitis or who were already in the ICU for other reasons but developed C difficile colitis that required a prolonged ICU stay. Patients with C difficile colitis who were in the ICU for unrelated reasons were excluded. In unclear cases, 3 independent investigators (E.A.S., K.S., K.C.) reviewed the medical records to form consensus. The diagnosis of fulminant C difficile colitis was confirmed by toxin immunoassay (Premier Toxin A/B test; Meridian Diagnostics, Cincinnati, Ohio), endoscopic findings, or histopathologic analysis by surgical specimen or autopsy or by the appropriate clinical context in the setting of recent documented C difficile colitis. Primary outcome was inpatient mortality, and secondary outcome was the need for colectomy. Some colectomy patients have been previously described.8
We collected data on demographics; comorbidities (diabetes mellitus; immunosuppression; cardiovascular, pulmonary, and renal disease; and cancer); history of hospitalization and antibiotic therapy during the 2 months before the current admission; recurrent C difficile colitis in the 6 months before the current admission; vital signs; symptoms; abdominal examination findings; mental status; body temperature; white blood cell (WBC) count; radiographic or endoscopic findings; medical (antibiotics, intubation, and vasopressors) and surgical (type of operation) intervention; interval between diagnosis and surgical intervention; surgical findings; sequential organ failure assessment score9; ICU and hospital length of stay; mortality; and primary department that provided care.
Patients who survived were compared with those who died. Categorical data were compared by the Fisher exact or χ2 test and continuous data by the Student t test. Pertinent continuous variables were dichotomized at predefined, clinically meaningful cutoff values. Variables with P < .20 in the univariate analysis or considered to be clinically important were entered into multiple logistic regression models to identify independent predictors of colectomy and mortality. Variables with P < .05 were retained in the model as significant predictors, and variables that produced an effect size change greater than 20% were retained in the model as significant confounders. The odds ratios (ORs) and 95% confidence intervals (CIs) were reported for each independent predictor. A 2-tailed P≤ .05 was considered statistically significant in all comparisons. All data are presented as mean (SD), unless otherwise specified. Approval was obtained from our institutional review board.
Of 532 810 patients admitted to our hospital during the 12-year study period, 4796 had C difficile colitis (incidence, 0.9%), and 199 patients had fulminant C difficile colitis (4.1% of patients with C difficile colitis). The incidence of C difficile colitis increased from 0.65% in 1996 to 1.2% in 2007, and the incidence of fulminant C difficile colitis increased from 2.5 per 10 000 admissions in 1996 to a peak of 6.3 per 10 000 admissions in 2005 (P < .001 for both). The mean (SD) age of patients with fulminant C difficile colitis was 68.3 (15.6) years, and 48.2% were male. Sixty-three patients (31.7%) developed fulminant C difficile colitis during an admission for another disease, whereas fulminant C difficile colitis was the primary reason for admission in 136 (68.3%). Most (86.9%) had been hospitalized in the previous 2 months, and 66 patients (33.2%) were postoperative (11 gastrointestinal and 55 nongastrointestinal operations). Major comorbidities included diabetes (29.6%), renal insufficiency (25.1%), cancer (25.6%), heart disease (51.3%), and chronic obstructive pulmonary disease (26.6%). One hundred patients (50.3%) were immunosuppressed (42.2% because of long-term steroid therapy, 15.1% because of chemotherapy, 4.0% because of immunosuppressive medications for organ transplantation, and 1.5% because of human immunodeficiency virus or AIDS). One hundred eighty-two patients (91.5%) had known antibiotic exposure in the previous 2 months. The most common preceding antibiotics were fluoroquinolones (39.0%), cephalosporins (35.7%), penicillins (15.4%), and clindamycin (11.5%).
Some degree of abdominal tenderness was found in all patients who could be examined; 35 (17.6%) had no diarrhea. The mean (SD) peak WBC count was 31 900/μL (20 900/μL), with 18.1% (15.2%) neutrophil bands. One hundred twenty patients (60.3%) required cardiorespiratory support (intubation in 85 and vasopressors in 92). The mean (SD) sequential organ failure assessment score was 5.6 (3.4).
A stool toxin immunoassay was performed in 194 patients, with a positive result in 179 (92.3%). In the 15 patients with a false-negative toxin assay result, the diagnosis of C difficile colitis was made by endoscopy in 6, computed tomography (CT) or recent positive toxin assay result with the appropriate clinical context in 7, and pathologic examination of the specimen in 2. All 23 patients who underwent endoscopy had findings consistent with C difficile colitis. Of 150 patients who underwent CT during the acute phase of fulminant C difficile colitis, abnormal findings were detected in 140 (93.3%), including colonic wall thickening in 126 (84.0%), colonic dilation in 34 (22.7%), free fluid in 79 (52.7%), and free air in 6 (4.0%). All colon pathologic specimens obtained from the operation confirmed the diagnosis of C difficile colitis.
Seventy-six patients (38.2%) were cared for by the general surgical department, and the remaining 123 patients (61.8%) were cared for by a nonsurgical department (medicine, neurology, neurosurgery, gynecology, or pediatrics). Medical treatment included oral metronidazole in 161 (80.9%), oral vancomycin in 88 (44.2%), and both in 79 (39.7%). Seventy-eight patients (39.2%) underwent an operation, including subtotal colectomy with diverting ileostomy in 69 (88.5%), hemicolectomy in 6 (7.7%), and exploratory laparotomy without bowel resection in 3 (3.8%). In the last 3 cases, the colon was found to be dilated intraoperatively but colectomy was not performed. The stool toxin was positive for C difficile postoperatively, and 2 of the 3 patients died. At laparotomy, 44 patients (58.7%) had only nonspecific findings (edema, free fluid, or dilation). Of the remaining patients, 30 (40%) had colonic ischemia or infarction, and 12 (16%) had perforation.
Table 1 compares patients treated nonoperatively with those who underwent colectomy. Patients were more likely to undergo colectomy if they presented with abnormal abdominal examination results (tenderness, distension, and peritonitis), had a higher peak WBC count or bandemia, had abnormal CT results (thickening, dilation, ascites, or perforation), or required intubation. A history of C difficile colitis in the recent past (recurrent C difficile) colitis was protective. Independent predictors of colectomy include abdominal distention (OR, 4.0; 95% CI, 2.0-8.4; P <.001), peritonitis (OR, 4.3; 95% CI, 1.6-11.5; P = .004), positive CT results (OR, 4.3; 95% CI, 1.8-10.7; P = .001), and no prior history of C difficile colitis (OR, 2.3; 95% CI, 1.0-5.0; P = .046).
The overall inpatient mortality rate was 34.7% (69/199).Table 2 compares those who survived with those who died. The independent predictors of mortality are given in Table 3: age of 70 years or older, WBC of 35 000/μL or higher or less than 4000/μL, neutrophil bands of 10% or more, need for vasopressors, and intubation. Notably, patients who underwent colectomy had a trend toward decreased mortality rates (OR, 0.49; 95% CI, 0.21-1.1; P = .08) after controlling for other predictors of mortality. Treatment with oral vancomycin was also associated with decreased mortality (OR, 0.25; 95% CI, 0.11-0.55; P <.001). The C statistic for the model with the predictors listed in Table 3 was 0.84.
These clinical predictors were further combined into 3 groups, easily usable at the bedside: (1) age of 70 years or older, (2) marked leukocytosis or leukopenia (WBC count, ≥35 000/μL or <4000/μL; or neutrophil bands, ≥10%), and (3) cardiopulmonary support (vasopressor agents and/or intubation).Table 4 gives the relative contribution of each of these 3 combined predictors to mortality. When all 3 predictors were present, 57.1% of patients died; when all were absent, no patient died. In particular, the presence of cardiopulmonary failure was more strongly associated with mortality (OR, 5.5; 95% CI, 2.6-11.3; P < .001) than old age (OR, 2.3; 95% CI, 1.2-4.6; P = .01) or either leukocytosis or leukopenia (OR, 2.9; 95% CI, 1.3-6.6; P = .01).
Among the 136 patients admitted to the hospital with a primary diagnosis of fulminant C difficile colitis, the mortality rate was 3.1 times higher (95% CI, 1.4-6.8; P = .001) in those cared for by the nonsurgical department compared with the surgical department (Table 5). Colectomy was performed more frequently in the patients in the surgical department, and the interval from admission to operation was shorter. For example, patients admitted directly to the surgical department underwent colectomy within a mean (SD) of 1.0 (1.3) days of hospital admission and all immediately underwent colectomy on admission to the ICU. This is in distinction to patients admitted to the nonsurgical department, who underwent colectomy a mean (SD) of 3.4 (4.2) days after admission to the hospital (P = .003), and after 1.7 (2.8) days of admission to the ICU (P = .001). Patients treated in the surgical department had significantly more abdominal distention, tenderness, and peritonitis and had CT findings of colonic wall thickening, ascites, or perforation, although they were less likely to require vasopressor support compared with those treated by a nonsurgical department. Even after controlling for these variables and other predictors of mortality, the department that cared for the patient (nonsurgical vs surgical) remained an independent predictor of mortality (OR, 3.4; 95% CI, 1.1-10.6; P = .03).
Of the 75 patients who underwent colectomy, 24 (32.0%) died. The mortality rate after colectomy was higher in patients with hypoalbuminemia (42.9% vs 18.2%; P = .03), those who required vasopressor support (51.4% vs 15.0%; P = .001) or intubation (53.7% vs 5.9%; P < .001), or those who were not treated with vancomycin (47.2% vs 17.9%; P = .01). Preoperative intubation (OR, 16.4; 95% CI, 3.4-79.7; P <.001) and failure to treat with oral vancomycin (OR, 3.3; 95% CI, 1.0-10.8; P = .05) independently increased the odds of mortality after colectomy. Again, those who were admitted with fulminant C difficile colitis and treated in the surgical department after their operation were significantly more likely to survive than those treated primarily by a nonsurgical department (OR, 8.8; 95% CI, 1.6-49.0; P = .01), after controlling for other predictors of mortality. In all patients with fulminant C difficile colitis, there was a trend toward longer delay in operative intervention from the time of ICU admission in those patients who died compared with those who survived (mean [SD], 3.9 [6.8] vs 1.9 [3.5] days; P = .18).
The incidence of C difficile colitis in the United States increased significantly from 31 per 100 000 population in 1996 to 61 per 100 000 population in 2003.2 Approximately 300 000 cases of C difficile colitis occur annually among hospitalized patients and are responsible for 3.6 hospital-days per patient and a total annual cost of more than $1.1 billion.10 The severity of C difficile colitis also appears to be increasing, with an overall incidence of 1% to 5% of C difficile colitis patients progressing to ICU admission, colectomy, or death.11 There was a 2.7-fold increase in age-adjusted deaths due to fulminant C difficile colitis from 1999 to 2002 according to a review of US death certificates.4 In our study, we have found a 2-fold increase in the incidence of C difficile colitis and an almost 3-fold increase in the incidence of fulminant C difficile colitis during a 12-year study period.
Mortality rates attributable to fulminant C difficile colitis remain high and are reported to be 57% at the University of Pittsburgh,6 38% at Johns Hopkins University,12 37.5% in the United Kingdom,13 48% at the Veterans Affairs hospitals,5 34% in Quebec,14 and 34% in New York,15 even after colectomy. We report the largest series of patients with fulminant C difficile colitis to our knowledge. In our series, 75 patients (37.6%) underwent colectomy, and 69 patients (34.7%) died. The high mortality rates may be related to increased recognition of fulminant C difficile colitis, which could have remained undetected in the past, higher virulence and antibiotic resistance of the bacterium as a result of the indiscriminate use of antibiotics, or failure to improve our therapeutic strategies. Three independent predictors of mortality were identified: (1) older age (≥70 years), (2) severe infection (WBC count, ≥35 000/μL or <4000/μL; or neutrophil bands, ≥10%), and (3) need for cardiorespiratory support (vasopressors and/or intubation). No patient died in our study in the absence of all 3 risk factors, whereas 57.1% died when all 3 were present. Although we cannot retrospectively conclude that an earlier operation could have improved the outcome for these patients, it is reasonable to target subpopulations defined by these 3 risk factors for aggressive surgical intervention, particularly before the onset of cardiorespiratory failure.
In our fulminant C difficile colitis population, urgent colectomy seemed to improve survival. The medically treated patients were different at baseline from those who underwent colectomy, with the latter group being clearly sicker. Despite this fact, after controlling for the differences, patients who underwent colectomy were 2.1-fold more likely to survive (P = .08) compared with those who were treated medically. A study14,16 from Canada demonstrated a survival benefit from colectomy during an epidemic outbreak of a hypervirulent strain of C difficile, and another study17 from Connecticut suggested that earlier surgical intervention was associated with improved outcome. Our study is the first to demonstrate a survival benefit for colectomy in nonepidemic fulminant C difficile colitis. However, in all of these analyses, including ours, the lack of randomization limits the conclusions.
To further understand the effect of different treatment policies on outcome, we compared patients cared for in surgical and nonsurgical departments. As expected, there is a tendency for more aggressive operative intervention in patients in the surgical department. To exclude the confounding issues related to patients who developed fulminant C difficile colitis while being treated in the hospital for other reasons, we analyzed only those who were admitted primarily for fulminant C difficile colitis. Patients admitted to a nonsurgical department were taken to the operating room on average 1.7 days later than those admitted directly to a surgical department. Furthermore, patients treated in a nonsurgical department had a 3.4-fold greater mortality rate compared with patients treated in the surgical department. Obviously, multiple factors may have influenced the outcome of these 2 groups, but even when controlling for confounding variables, the type of care remained an important and significant predictor of outcome. Clearly, causality between mortality and the primary department that cared cannot be established by this retrospective study. Nevertheless, the possibility of a more aggressive treatment philosophy resulting in improved survival cannot be ignored and must be explored further.
Another point of interest confirmed by our study is the use of oral vancomycin. Overall, only 44.2% of our patients with fulminant C difficile colitis were treated with oral vancomycin, either as a single agent or as combination therapy with oral metronidazole. Vancomycin is reserved as a second-line agent in the treatment of mild to moderate C difficile colitis because metronidazole is considered to be equally effective and less costly and avoids vancomycin resistance. However, for patients who progress to severe disease, it is important to convert to or add vancomycin.18 Oral vancomycin independently increased the odds of survival by 4-fold in our population. It seems that adequate medical therapy combined with prompt surgical therapy are crucial in improving survival, although the statistical weight of the 2 variables was different in our multivariate analysis.
The identification of C difficile colitis patients through an administrative database by ICD-9 codes and the arbitrary definition of fulminant C difficile colitis by the need for ICU admission may affect the accuracy of our analysis. However, all cases were confirmed by toxin assay, endoscopy, or surgical and pathologic findings with the appropriate clinical context. The exact decision making could not be extracted through the review of medical records. The addition of 2 patients who required colectomy but no ICU admission may have introduced a small bias in our analysis of the need for operation among patients with fulminant C difficile colitis. Despite these limitations, our study is the largest series of fulminant C difficile colitis to date and identifies a disturbingly high mortality rate. Predictors of mortality exist and should be used to identify patients who require early and aggressive surgical intervention. Oral vancomycin should be given routinely. Admission of patients with fulminant C difficile colitis to surgical departments may offer a survival advantage. Our treatment strategies need to be reexamined to improve the outcome of these patients.
Correspondence: Elizabeth A. Sailhamer, MD, MMSc, 165 Cambridge St, Suite 810, Boston, MA 02114-2783 (firstname.lastname@example.org).
Accepted for Publication: January 5, 2009.
Author Contributions:Study concept and design: Sailhamer, DeMoya, and Velmahos. Acquisition of data: Sailhamer, Carson, Zacharias, Spaniolas, and Tabbara. Analysis and interpretation of data: Sailhamer, Chang, Alam, and Velmahos. Drafting of the manuscript: Sailhamer, Chang, Tabbara, and Velmahos. Critical revision of the manuscript for important intellectual content: Sailhamer, Carson, Zacharias, Spaniolas, Alam, DeMoya, and Velmahos. Statistical analysis: Sailhamer, Chang, Spaniolas, and Tabbara. Administrative, technical, and material support: Carson, Zacharias, Spaniolas, Tabbara, and Velmahos. Study supervision: Alam, DeMoya, and Velmahos.
Financial Disclosure: None reported.
Funding/Support: Dr Sailhamer was supported by the Ruth L. Kirschstein National Research Service postdoctoral fellowship F32 GM79880 at the National Institutes of Health/National Institute of General Medical Sciences and the Scholars in Clinical Science Program at the Harvard Medical School.
Previous Presentations: This paper was presented at the 89th Annual Meeting of the New England Surgical Society; September 26, 2008; Boston, Massachusetts; and is published after peer review and revision.