Blood samples, which were obtained from patients who lived in a rural area with ∼500 acute-care hospital beds, were cultured from 1990 through 1997. We retrospectively reviewed the blood cultures that yielded Clostridium species (74 [0.12%] of 63,296 cultures). These were obtained from 46 different hospitalized patients (incidents per hospital, 0.03%). The source of the Clostridium species was a gastrointestinal site in 24 patients (52.2%). The most frequently identified Clostridium species was Clostridium perfringens (in 10 [21.7%] of patients), followed by Clostridium septicum (in 9 [19.6%]). Thirty-one patients (67.4%) were aged ⩾65 years, 13 patients (28.3%) had diabetes mellitus, and underlying malignancy was present in 22 patients (47.8%). The mortality rate of patients whose condition had been managed surgically was 33%; for those patients whose conditions required medical management, the mortality rate was 58%. Clostridium bacteremia in these patients usually had a gastrointestinal source, it often occurred in patients with serious underlying medical conditions, and it rarely was the result of traumatic farm accidents.
During the first half of the 20th century, clostridial infections, usually with Clostridium perfringens, were associated with extremely serious soft-tissue and muscle infections, such as gas gangrene, which often occurred in patients who had fecally contaminated farm wounds, war wounds, intra-abdominal sepsis, or septic abortion [1–3]. However, in the past 50 years, with refinements in anaerobic bacteriologic techniques and more routine anaerobic culturing , it has become apparent that the anaerobic, gram-positive, sporulating bacilli that constitute the genus Clostridium are unusual, but not rare, causes of tissue and bloodstream infections. Patients who are infected with this microorganism display a wide spectrum of clinical presentations [5, 6].
In other reviews in which blood samples were obtained from patients in tertiary-care facilities, only 0.5%–2.0% of all isolates yielded on blood culture revealed clostridial septicemia; of these, C. perfringens was the microorganism that was most frequently identified in blood, accounting for 20% to ⩾50% of isolates . Within the past 30 years, observational studies from large metropolitan hospitals emphasized that clostridial isolation in blood cultures was secondary to multiple possible sources, including the following: contamination; transient bacteremia, often from an unknown source; traumatic or surgically related bowel leakage or abscess; and, rarely, gas-forming tissue infection, such as emphysematous cholecystitis, crepitant cellulitis, fasciitis, or myonecrosis [8–11].
With changes in the human epidemiologic factors that are associated with infection, more recent articles have focused on spontaneous clostridial bacteremia in compromised hosts, such as patients with diabetes, adenocarcinoma of the colon, lymphoma, or leukemia [7, 12–16]. Because recent clinical reviews concerning anaerobic bacteremia, including clostridial bacteremia, have predominantly presented data that were obtained in large metropolitan hospitals [5–11], we undertook a retrospective review of cases of clostridial bacteremia in a rural setting to see whether the clinical features and epidemiologic factors varied from those presented in reports that have focused on urban areas.
Materials and Methods
Clinical evaluation. La Crosse, Wisconsin, is a small city with a population of ∼50,000 in a county with a population of <100,000. The surrounding rural community population contributes another 150,000 to the referral base. This overall population is served by 2 teaching hospitals of ∼300 and ∼200 beds, respectively. Blood samples, which were obtained from patients in both institutions, and which were cultured from 1 January 1990 through 31 December 1997, were retrospectively reviewed, and the clinical evaluations of all patients with blood cultures that were positive for Clostridium species were analyzed.
Cases that were not associated with a likely female genital or gastrointestinal source for Clostridium species and cases that occurred in patients who lacked symptoms or signs of sepsis were considered to have been caused by contaminants. Patients with localized infections, such as diabetes-related foot ulcers or diverticulitis, were considered to have transient bacteremia if they had only fevers of brief duration (duration, <1 h) and no persistent symptoms or signs of infection. All other cases were considered to be clostridial sepsis. These cases were classified according to probable original source of clostridial infection, and they were further subdivided into the categories of either “sepsis that required surgical procedures” or “sepsis that was not clinically judged to have been associated with the need for tissue resection or drainage.”
Charts for all patients in whom either bacteremia or septicemia was present were reviewed for age, sex, premorbid condition, presence of malignancies, American Society of Anesthesiology (ASA) classification, pertinent history, the results of physical and laboratory testing, and radiographic data. When available, autopsy reports were also reviewed.
Microbiologic evaluation. During the study period, blood samples that were obtained for culture from patients at both hospitals were processed by means of a similar technique. In general, all blood samples were cultured aerobically (8–10 mL) and anaerobically (5–7 mL) by use of the BACTEC nonradiometric method (Becton Dickinson). Cultures of blood samples that were positive for Clostridium species were subcultured on tryptic soy blood agar or chocolate agar, then on thioglycollate that had been supplemented with hemin and vitamin K. Isolates that did not grow aerobically were subcultured on brain-heart infusion blood agar and pea blood agar plates that had been supplemented with hemin and vitamin K. Anaerobe plates were incubated for 48 h by use of the GasPak jar system (Becton Dickinson).
All gram-positive anaerobic rods were further evaluated, and any rods of the genus Clostridium were further identified, when practical, to species. The phenotypic characteristics that were tested included the following: the presence of sporulation (when demonstrated); the presence of double-zone hemolysis; gelatinase by the use of gelatin agar; lecithinase activity with egg yolk agar; and, when indicated, indole, nitrate, esculin, catalase, and fermentation of sugars, by use of gas chromatography.
Of 63,296 blood samples that were obtained for culture during the study period, a total of 74 cultures (0.12%) were positive for Clostridium species; these samples were obtained from 46 patients out of a total of 164,304 hospitalizations, for an incidence of clostridial bacteremia in hospitalized patients of 0.03% (no cases occurred in nonhospitalized patients). The number of blood samples that were obtained for culture per year and the number of those cultures that were positive for Clostridium per year were relatively constant throughout the 8-year study period.
Probable sources for Clostridium species that were yielded on blood culture are listed in table 1. Isolates recovered from 24 patients (52%) had a female genital or a gastrointestinal source; 12 of these 24 patients had isolates with a source in the colon (5 patients with bowel cancer, 3 patients with pseudomembranous colitis, 3 patients with appendicitis or diverticulitis, and 1 patient with traumatic tear of the rectum). The second most common source was the lung, which was the case for 6 patients (13%), including 3 patients who underwent aspiration (1 of whom had a tracheoesophageal fistula due to adenocarcinoma), 2 patients with adenocarcinoma of the lungs, and 1 patient with empyema.
The species that were identified in the 46 patients who were considered to have clinically significant positive cultures for Clostridium bacteremia are included in table 2. The most common species was C. perfringens, followed by Clostridium septicum.
Thirty-one patients (67.4%) were aged ⩾65 years. The underlying conditions of the patients who had clostridial bacteremia are detailed in table 3. The most common underlying condition was malignancy, often associated with chemotherapy, radiation therapy, or both. Diabetes and neutropenia were also relatively common. Most patients had an ASA classification of ⩾3.
Twenty-one patients died, for an overall mortality rate of 46%. Of the 24 patients whose conditions were medically managed, 14 (58%) died; of the 21 patients whose conditions could have been managed with surgery, 7 (33%) died. Surgical procedures included open abscess drainage (in 2 patients), CT scan-directed abscess drainage (in 3); bowel resection or major operation (in 6); moderate operation (i.e., appendectomy and diverticulitis with localized abscess requiring drainage and minimal resection; in 2); elimination of biliary obstruction (in 2); and other surgical procedures (in 2). Thirty-four patients (74%) were identified as having had a condition for which a surgical procedure was standard therapy; in 13 of these patients, the septic source was not recognized in time, the patients were not candidates for surgery because of the presence of underlying conditions, or both. Therefore, the mortality rate of patients with a nonsurgical condition (10 patients [41%]) was similar to that of patients with a surgical condition (11 patients [50%]).
During the study period, 102 isolates of C. perfringens that were recovered from any culture site were 100% susceptible to penicillin, cefmetazole, clindamycin, and metronidazole, whereas 124 isolates of non-C. perfringens species (excluding Clostridium difficile) were 98% susceptible to penicillin, 99% susceptible to cefmetazole, 94% susceptible to clindamycin, and 100% susceptible to metronidazole. Because most of the patients with bacteremia received appropriate initial antibiotic therapy, and because the number of more antibiotic-resistant Clostridium species, such as Clostridium tertium (5 isolates) and Clostridium ramosum (4 isolates), was small, no analysis of the effectiveness of antibiotic therapy was possible.
We found that the intestinal tract was the major source of the clostridial bacteremia: 17 cultures of blood samples (23%) yielded another gut microorganism in addition to Clostridium species, the most common of which was Escherichia coli (5 patients [10%]). In addition, 6 patients who had undergone surgical drainage were found to have Clostridium species in samples of pus; usually, it was the same species that had been isolated from the blood cultures, and often it was mixed with other intestinal bacterial flora.
Although no cases of clostridial myonecrosis developed, 1 woman with C. septicum sepsis related to neutropenia developed localized calf myositis. This patient, who did not display clinical features of clostridial myonecrosis, responded to treatment with antimicrobial agents alone without surgical resection.
Infection was found in only 1 patient who had experienced a farm-related trauma—a farmer developed a rectal leak after he fractured his pelvis in a tractor rollover accident. No patients with animal manure-related infections were seen. It is noteworthy that there were no cases of gas gangrene related to trauma, gunshot wound infection, or subcutaneous infection associated with narcotic abuse.
One patient developed severe hemolysis from overwhelming C. perfringens septicemia that was secondary to emphysematous cholecystitis. In the emergency department, her initial hemoglobin level was 11 mg/dL and her centrifuged hematocrit level was 2%. Gram stain of a blood sample revealed gram-positive rods with ruptured RBCs (figure 1A), and her anticoagulated blood sample revealed marked spontaneous hemolysis of blood (figure 1B). Despite attempts to resuscitate the patient, she died of septic shock, disseminated intravascular coagulation, and renal and other multiorgan failure within hours of admission.
We found a lower incidence of clostridial bacteremia (0.12% of blood cultures and 0.03% per hospitalization) than have recent studies. However, epidemiologic factors, unlike those that occurred during the early part of the 20th century, were similar to those described in more recent reports [8–11]. Specifically, the predominant source of clostridial sepsis in these patients was the gastrointestinal tract, but this was not the result of penetrating traumatic wounds. The lack of traumatic wounds as a source of infection in this study is probably different from what a large city hospital experiences; the population served by the La Crosse medical centers has not had a significant problem with gunshot wounds and injection drug abuse, which are unfortunately common in large American and European cities [17, 18]. Therefore, our cases tended to be of a milder clinical nature than are those that have been reported elsewhere; the cases that we report were more often associated with underlying malignancies, diabetes, and other host-related conditions. In rural America, clostridial septicemia is analagous to Enterobacteriaceae septicemia—both usually occur in patients with underlying defects in the defense mechanisms that allow for the infection.
Despite our experience with significant, aerobic, gram-negative rod infections in patients with farm-related, traumatic extremity wounds , we did not find any such traumatic limb infections in this retrospective review. We believe this is because patients with potentially infected farm wounds who are seen in the La Crosse area receive rapid surgical debridement and antibacterial therapy.
Although one of the patients died of fulminant clostridial sepsis with massive hemolysis, presumably due to phospholipase C  (figure 1), most infections tended to be of a more benign nature. Therefore, clostridial bacteremia was frequently a marker for the clinician to evaluate for an underlying illness and source and to decide, in addition to whether to initiate antibiotic therapy directed against Enterobacteriaceae and anaerobes, whether surgery or an image-directed drainage procedure was required . The need for surgery was determined in most cases by the use of radiographic means, such as CT scans, MRI scans, and ultrasounds. Of the 35 patients who had antibacterial therapy that could be assessed by use of recently recommended regimens for mixed anaerobic-aerobic bowel-related infections , only 3 patients received antibacterial therapy that was inappropriate for the clinical situation. Therefore, on the basis of this study, no conclusion can be made regarding the efficacy of anticlostridial antimicrobial agents. However, inappropriate antibacterial therapy for anaerobes, in general, appears to have serious consequences for patients .
Finally, the overall mortality rate, which reflects the poor underlying state of the patients, was 48%. Because this high mortality rate was related to underlying conditions in the hosts (as was reflected by high ASA scores), clostridial bacteremia remains a potentially serious clinical marker, less frequently because of clostridial virulence, and more frequently because of the severity of associated medical illnesses in elderly and immunocompromised patients.