Clinical review: Bacteremia caused by anaerobic bacteria in children
© BioMed Central Ltd 2002
Published: 9 May 2002
Skip to main content
© BioMed Central Ltd 2002
Published: 9 May 2002
This review describes the microbiology, diagnosis and management of bacteremia caused by anaerobic bacteria in children. Bacteroides fragilis, Peptostreptococcus sp., Clostridium sp., and Fusobacterium sp. were the most common clinically significant anaerobic isolates. The strains of anaerobic organisms found depended, to a large extent, on the portal of entry and the underlying disease. Predisposing conditions include: malignant neoplasms, immunodeficiencies, chronic renal insufficiency, decubitus ulcers, perforation of viscus and appendicitis, and neonatal age. Organisms identical to those causing anaerobic bacteremia can often be recovered from other infected sites that may have served as a source of persistent bacteremia. When anaerobes resistant to penicillin are suspected or isolated, antimicrobial drugs such as clindamycin, chloramphenicol, metronidazole, cefoxitin, a carbapenem, or the combination of a beta-lactamase inhibitor and a penicillin should be administered. The early recognition of anaerobic bacteremia and administration of appropriate antimicrobial and surgical therapy play a significant role in preventing mortality and morbidity in pediatric patients.
Infections caused by anaerobic bacteria can occur in children, and may be serious and life-threatening. The recent increased recovery of these organisms from children has led to greater appreciation of the role anaerobes play in pediatric infections at all body sites, including the bacteremia.
Anaerobes are one of the predominant components of the normal human skin flora and the most predominant component of the bacterial flora of the mucous membranes , and are therefore a common cause of bacterial infections of endogenous origin. Because of their fastidious nature, these organisms are difficult to isolate from infectious sites, and are often overlooked. Their exact frequency is difficult to ascertain because of the inconsistent use of adequate methods for their isolation and identification. A lack of direct adequate therapy against these organisms may lead to clinical failures. Their isolation requires appropriate methods of collection, transportation and cultivation of specimens . Treatment of anaerobic infection is complicated by the slow growth of these organisms, by their polymicrobial nature and by the growing resistance of anaerobic bacteria to antimicrobial drugs.
Although anaerobes have been reported to account for 8% to 11% of episodes of bacteremia in adults , anaerobic organisms have rarely been isolated from blood cultures of pediatric patients. These microbes represent a small percentage of the total number of positive blood cultures recovered from children, which may be because of the difficulty in isolating and identifying these organisms. There is, however, a growing awareness of the role of anaerobes in bacteremia [2,3–7], especially in children with certain predisposing conditions and in newborns, who are at high risk, and in those with necrotizing enterocolitis. This review describes the microbiology and management of bacteremia due to anaerobic bacteria in children.
In a survey of anaerobic infections in children, blood cultures have been found to be the second most frequent source of anaerobic organisms [2–4]. In one of these reviews of the recovery of anaerobes from children in a university hospital over a period of one year , 13 blood cultures were positive and contained 14 anaerobes. In a large prospective study lasting a year, only 0.3% of blood cultures contained anaerobic bacteria that were involved in the pathogenesis of the patient's disease . In contrast, pathogenic aerobes were recovered from 9% of the cultures tested during that period. Anaerobes accounted for 5.8% of all bacteremic episodes (8.7% in the newborn period and 4.8% in children over 1 year of age). Notably, 10% of the newborns with clinical bacteremia had only anaerobes recovered from their blood cultures.
Zaidi et al. , reviewed the use of anaerobic blood cultures for children and noted that 15 (2.1%) of 723 cases of bacteremia were caused by strict anaerobes and they concluded that use of the entire volume of blood drawn should be reserved for aerobic cultures. Recent studies have suggested that there has been a decline in the incidence of anaerobic bacteremia. Some authors [9–13] have speculated that this might be as a result of the use of bowel preparations prior to abdominal surgery and the more routine use of antibiotics active against anaerobes.
Anaerobic bacteremia has rarely been described in pediatric patients [13,14]. Sanders and Stevenson  in a review of the literature in 1968 summarized 11 cases of Bacteroides bacteremias in children. In one study, anaerobic organisms were recovered from 6 of 34 children who required general anesthesia and nasotracheal intubation for dental repair . Another study documented bacteremia in 28 children who were undergoing dental manipulations . Among the 28 isolates recovered, 21 were anaerobes (Propionibacterium sp., nine; Veillonella alcalescens, five; Prevotella melaninogenica, three; Peptostreptococcus sp., two; and Eubacterium sp. and Fusobacterium sp., one each).
Brook et al.  reviewed their experience in recovery of anaerobes in the blood over a 12-month period. A total of 13 blood cultures were positive and contained 14 anaerobic agents: five were Bacteroides fragilis, three others were Bacteroides sp., two were Fusobacterium sp., three were Propi-onibacterium sp., and one was Peptostreptococcus sp. In one instance two organisms were isolated from a blood culture: Peptostreptococcus sp. and Fusobacterium sp.
Dunkle et al.  recovered 14 anaerobes from blood cultures over a 1-year study. The dominant anaerobes rec overed were Clostridium sp. (four), Fusobacterium nucleatus (three species), Gram-positive cocci (three species), and B. fragilis (two species). Although 27 isolates of Propionibacterium acnes were recovered, only three were associated with clinical infection.
Thirmuoothi et al.  reviewed their experience over a period of 18 months, and reported 35 anaerobic isolates from 34 blood cultures. The predominant isolates were four each of Gram-positive cocci and Bacteroides sp. and two isolates each of Fusobacterium sp., Bifidobacterium sp., and Clostridium sp. Although Propionibacterium sp. were recovered in 18 instances, there was no apparent relationship between their recovery from the blood and the 18 patients' clinical illness.
Brook and colleagues  summarized their experience in the diagnosis of anaerobic bacteremia noted in 28 children. Twenty-nine anaerobic isolates were recovered from 28 patients ranging in age from 1 week to 15 years. Of these isolates, 14 were Bacteroides sp. (11 of which belonged to the B. fragilis group); four were Clostridium sp.; four were anaerobic Gram-positive cocci; four were P. acnes; and three were Fusobacterium sp. Although the predominant isolate from blood cultures (56–65%) is P. acnes [2,3], a normal inhabitant of the skin, many of these isolates may reflect contamination of the blood cultures by the skin flora. Propionibacterium acnes can cause bacteremia, however, especially in association with shunt infections . All of the patients with P. acnes bacteremia included in the study by Brook et al.  had clinical infection, and all but one responded to antimicrobial therapy. Furthermore, two patients had meningitis caused by this organism after installation of cardiovascular shunts.
An important aspect of anaerobic bacteremia is that anaerobes frequently are present in cases of polymicrobial bacteremia , reflecting the fact that localized anaerobic infections are usually polymicrobial. Polymicrobial bacteremia involving anaerobic bacteria were reported by several authors. Frommell and Todd  reported 56 children with bacteremia with multiple bacterial isolates. Five anaerobes were isolated: two Bacteroides sp., two Peptostreptococci and one Clostridium perfringens. Rosenfeld and Jameson  reported a 15-year-old child with polymicrobial bacteremia involving seven isolates (including four Bacteroides sp. and an anaerobic cocci) associated with pharyngotonsilli-tis. Seidenfeld et al.  reported an adolescent with a fatal bacteremia caused by Fusobacterium necrophorum and Pep-tostreptococcus sp. associated with peritonsillar abscess. Givner et al.  recovered Bacteroides capillosus with Corynebacterium hemolyticum from the blood of a child with primary Epstein–Barr virus infection who developed sinusitis.
Caya and Truant summarized 65 cases of non-infant pediatric clostridial bacteremia . The predominant isolates were Clostridium septicum (25 isolates), Clostridium perfringens (21 isolates) and Clostridium tertium (six isolates). Of the 63 children analyzed, 29 (46%) survived their episode of clostridial bacteremia. Three clinical indices were shown to have a statistically significant negative impact on survival: hypotension, hemolysis and lack of antibiotic therapy. Of the 36 patients with known underlying neoplastic disease, 27 had acute leukemia, five had sarcoma, three had a malignant lymphoproliferative disorder and one had glioblastoma multiforme. Of the 23 patients with no underlying neoplasia, three of them had cyclic neutropenia, two were in sickle cell disease crisis, two had neutropenia associated with aplastic anemia, and one was mildly immunocompromised as a result of renal transplantation.
Brook reported the microbiology of 101 specimens obtained from 95 children with malignancy . A total of 17 patients had bacteremia. Four had Escherichia coli, in one instance mixed with B. fragilis. Bacteroides fragilis group isolates were recovered in three instances (two in patients with leukemia who had a perirectal abscess), Staphylococcus aureus in three patients, Clostridium spp. in two (one C. perfringens and one C. septicum) and two Proteus spp.
Brook summarized clinical and microbiological data of 296 adults with anaerobic bacteremia . Anaerobes were isolated with aerobic or facultative bacteremia in 23 instances. The B. fragilis group accounted for 148 (70%) of 212 isolates of anaerobic Gram-negative bacilli. Bacteroides fragilis accounted for 78% and B. thetaiotaomicron for 14%. Among other species, there were 20 (6%) Fusobacterium organisms, 63 (18%) Clostridium isolates, and 53 (15%) anaerobic cocci. Seventy-five patients died: 40 of these had B. fragilis group isolates (including B. fragilis, 28, and B. thetaiotaomicron, 8) and 21 had Clostridium organisms isolated.
Anaerobic bacteremia is almost invariably secondary to a focal primary infection. As reported for adults , the strain of anaerobic organisms recovered depended to a large extent on the portal of entry and the underlying disease. Bacteroides fragilis is the most frequent anaerobic isolate [13, 23–28] and, with other members of the B. fragilis group species, accounts for 36–64% of anaerobic blood isolates. Bacteroides thetaiotaomicron is the second most common member of the group to be isolated from blood. Clostridia, especially C. perfringens, and peptostreptococci are also frequently isolated from blood. The gastrointestinal tract accounted for half of the anaerobic bacteremias and the female genital tract was the source of 20% of these bacteremias [13, 27–30].
Brook  noted in adults that the gastrointestinal tract was the principal source of B. fragilis and clostridial bacteremias and that the female genital tract was the principal source of peptostreptococcal and fusobacterial bacteremias. Redondo et al.  reported that bacteremias caused by the B. fragilis group of organisms originated from: the gastrointestinal tract (69% of bacteremias); soft-tissue wound infections (16%); the female genitourinary tract (5%); and lung infections (4%). Fainstein et al.  found bacteremia caused by B. fragilis to be common in patients with genitourinary and gynecological tumors, acute leukemia, and gastrointestinal malignancies.
The probable portals of entry for the blood culture isolates in the 28 children studied by Brook and associates  were: the gastrointestinal (GI) tract (13 patients), the respiratory tract (ear, sinus, and oropharynx, seven), the lower respiratory tract (three), cardiovascular shunts and neurologic shunts (three), and skin and soft tissue (three). When the GI tract was the probable portal of entry, Bacteroides sp. (eight isolates, including five B. fragilis) and Clostridium sp. (four isolates) were the organisms most frequently recovered from blood. The predominant anaerobic organisms recovered in association with infections of the ear, sinus and oropharynx were Peptostreptococcus sp. (from four patients) and F. nucleatum (from two patients). Propionibacterium acnes was grown in cultures taken from four patients, three of whom had artificial cardiac valves or ventriculoatrial shunts. Two of these patients also were initially observed to have meningitis caused by a similar organism. All lower respiratory tract infections that served as a probable source of bacteremia were caused by isolates belonging to the B. fragilis group.
No obvious focus of infection was noted in six patients; interestingly, however, all of these patients had some GI problem that might have served as a source of the bacteremia. Furthermore, four of these patients had bacteremia caused by Clostridium species.
These findings therefore support studies of adults [13,32,33] and children [6,14] that report that Bacteroides species, including the B. fragilis group, were the predominant isolates from patients in whom the GI tract was the probable portal of entry. As summarized by Sanders and Stevenson , however, other anaerobic Gram-negative bacilli caused bacteremia in children with otitis media and abscesses.
The ear, sinus, and oropharynx were found to be possible portals of entry that predisposed patients to bacteremia with Peptostreptococcus sp. and Fusobacterium sp. This is not surprising because these organisms are part of the normal flora of these anatomic sites and can be involved in local infections .
Three newborns developed bacteremia in conjunction with pneumonia with organisms belonging to the B. fragilis group . This has also been noted before in newborns  and adults . Although Bacteroides accounted for the majority of the episodes of bacteremia in this study, other studies have shown relatively infrequent isolation of these organisms from children , except during the neonatal period .
An association between surgical procedures and anaerobic septicemia was recently reported. Pass and Waldo  observed anaerobic bacteremia in two infants following suprapubic bladder aspiration. Bacteroides fragilis was isolated in one instance and in another instance was mixed with Veillonella alcalescens. An accidental bowel perforation was the assumed etiology of these infections. Kasik et al.  observed sepsis and meningitis caused by E. coli and Bacteroides sp. after anal dilatation. Fusobacterium mortiferum was also recovered in the blood.
Fisher et al.  described bacteremia caused by B. fragilis in four of 75 children after elective appendectomy in renal transplant recipients. The bacteremia was associated with profound lymphopenia. Fusobacterial infection generally is associated with otolaryngological processes. Seidenfeld et al.  reported five patients, four of whom were children, who developed F. necrophorum septicemia following oropha-ryngeal infection. Septicemia caused by Streptococcus mor-billorum was reported by Rushton to have complicated herpetic pharyngitis .
Bacteroides fragilis, anaerobic Gram-positive cocci, and Fusobacterium sp. were the clinically significant anaerobic organisms most commonly isolated from blood cultures in three recent studies [2–4]. Most of the patients described in these studies were either newborns or were over 6 weeks of age and suffered from chronic debilitating disorders such as malignant neoplasms, immunodeficiencies, chronic renal insufficiency, or decubitus ulcers and carried a poor prognosis. Bacteroides sp. were also isolated frequently after perforation of viscus and appendicitis [38,39].
Clostridium sp. may complicate leukemias. Caya et al.  reported 11 children with leukemia who presented with sepsis caused by Clostridium septicum (seven children), C. perfringens (two children), and Clostridium sp. (two children). None of these children survived the sepsis, which was characterized by thrombocytopenia, gastrointestinal lesions, and neutropenia.
Infectious mononucleosis can also predispose to anaerobic bacteremia. Dagan and Powell  observed three patients who developed postanginal anaerobic sepsis following Epstein–Barr virus infection. All three had Fusobacterium species isolated (two were F. necrophorum) and in one case a Peptostreptococcus was also recovered.
Predisposing factors to anaerobic bacteremia in adults include malignant neoplasms [42,43], hematologic disorders , transplantation of organs , recent gastrointestinal or obstetric gynecologic surgery [43–46], intestinal obstruction , diabetes mellitus , post-splenectomy , use of cytotoxic agents or corticosteroids , and use of pro-phylactic antimicrobial agents for bowel preparation prior to surgery [43,46].
Predisposing conditions were noted also in one study of pediatric patients . Two patients had malignant neoplasms, two suffered from hematologic abnormalities, and one had an immune deficiency. Interestingly, 82% of the bacteremias in this series of patients  occurred in children who had no immunosuppression or malignant neoplasms. This is in contrast to another study  in which anaerobic bacteremia occurred more frequently in children with these predisposing factors. Dental or oral surgery can also predispose to anaerobic bacteremia in adults and children [13,15,16].
The clinical features of anaerobic bacteremia are not much different from those associated with other types of bacteremia in children; however, a relatively longer period is generally needed before an etiologic diagnosis can be made. This can be a result of the smaller volume of blood drawn from children for culture inoculation and the longer time needed for growth and identification of anaerobic organisms.
Diagnosis should include detection of the primary infection. The clinical presentation of anaerobic bacteremia relates, in part, to the nature of the primary infection, which will typically include fever, chills and leukocytosis. Anemia, shock and intravascular coagulation may also be present. Bacteroides bacteremia is generally characterized by thrombophlebitis, metastatic infection, hyperbilirubinemia and a high mortality rate (up to 50%). Clostridium perfringens bacteremia may have a most dramatic clinical picture, consisting of hemolytic anemia, hemoglobinemia, hemoglobinuria, disseminated intravascular coagulation, bleeding tendency, bronze-colored skin, hyperbilirubinemia, shock, oliguria and anemia. Clostridial bacteria may, however, be transient and inconsequential. However, C. septicum infection may be a marker for a silent colonic or rectal malignancy .
Blood culture supporting the growth of anaerobic bacteria should be used routinely in all patients. In addition to supporting the growth of strict anaerobes, blood cultures also facilitate the growth of many facultative anaerobes. Some cases of culture-negative endocarditis, fever and systemic toxicity with negative blood cultures are undoubtedly cases of anaerobic bacteremia that elude detection because of inadequate methodology.
Because of the high mortality rate (15–35%) associated with anaerobic bacteremia, it is imperative to establish early effective therapy. Prolonged therapy with antimicrobial agents apparently is adequate for most patients. However, any source of infection, such as an abscess, should be surgically drained. The average duration of therapy in the patients who recovered in one study  was 20 days (range, 7–72 days), and the duration of therapy was related to the presence and severity of other infectious sites and complications. Therapy was longest in the treatment of bacteraemia associated with meningitis, wound abscess, sinusitis and empyema. When anaerobes resistant to penicillin, such as the B. fragilis group, are suspected or isolated, antimicrobial drugs, such as clin-damycin, chloramphenicol, metronidazole, cefoxitin, a car-bapenem, or the combination of a beta-lactamase inhibitor and a penicillin (i.e. ticarcillin-clavulante, piperacillin-tazobactam), should be administered. Local surveillance of antimicrobial susceptibility patterns can provide guidelines as to the choice of the best antimicrobial agent. The development of resistance to all known agents by anaerobes, makes the selection of reliable empirical therapy difficult. Many anaerobic species besides the B. fragilis group have acquired the ability to produce beta-lactamase. Rarely, resistance to imipenem, induced by metalloenzymes, and to metronidazole has been reported [48–50]. Consequently, one is not able to predict the susceptibility of some anaerobic isolates. Performing susceptibility testing is of great importance in treating bacteremia caused by anaerobes.
Organisms identical to those causing anaerobic bacteremia can often be recovered from other infected sites (as in 16 patients, 57%, in the study by Brook et al. ). No doubt these extravascular sites may have served as a source of persistent bacteremia in some cases; however, the majority of patients will recover completely if prompt treatment with appropriate antimicrobial agents is instituted before any complications develop. The early recognition of anaerobic bacteremia and administration of appropriate antimicrobial and surgical therapy play a significant role in preventing mortality and morbidity in pediatric patients.
Preventing bacteremia associated with dental or oral surgery can be accomplished by prophylactic administration of penicillin . It was demonstrated that, although penicillin pro-phylaxis reduced the total number of facultative anaerobes and strict anaerobes recovered from the blood, metronidazole was more effective in decreasing the recovery of Gram-negative anaerobes . Therefore, a combination of the two may be more effective than either agent alone in eliminating bacteremias after dental procedures.
The source of anaerobic bacteremia is generally clinically suspected, so therapy with antimicrobial agents active against anaerobes is often instituted empirically. Empirical therapy may provide coverage for anaerobes in only half of the patients with anaerobic bacteremia, and failure to pay attention to the results of anaerobic blood cultures may have serious consequences .
Mortality as a result of anaerobic bacteremia remains high. Risk factors for a fatal outcome include compromised status of the host, advanced age, inadequate or no surgical therapy, and the presence of polymicrobial sepsis. Additionally, mortality varies between the infecting B. fragilis group species [53,54]. Bacteroides fragilis is the most common anaerobic isolate in these studies [53,54], with associated mortality between 24% and 31%, while the mortality associated with B. thetaiotaomicron bacteremia ranges between 38% and 100%, and that associated with B. distasonis bacteremia is about 50%. Whether these differences are the result of differences in virulence factors such as endotoxins, encapsulation, host defenses, or differences in antimicrobial susceptibility remains unknown.
The mortality following anaerobic bacteremia varies. In one study  it was 18% (five of 28 patients) and depended on such factors as age of the patient, underlying disease, nature of the organism, speed of diagnosis, and surgical or medical therapy instituted. This mortality rate is similar to that reported in adults . Of the three infants who died, two were newborns and one was 8 months old. Four patients were infected with organisms of the B. fragilis group that were resistant to penicillin; inappropriate antimicrobial therapy was administered to two of these patients, owing to the length of time needed for identification of the organisms, and the other two patients had underlying disorders that further aggravated their condition. The fifth child who died had a ventriculoatrial shunt that was infected with P. acnes, in addition to severe hydrocephalus and mental retardation.
Certain other serious concomitant sites of infection can be present in children with anaerobic bacteremia. Most of these sites serve as the source of the infection, however others may represent a site of secondary hematogenous spread of the organism(s). The most frequent conditions are meningitis, peritonitis, subdural empyema, and septic shock. Although some of the children with these infections may become seriously ill, most will respond well to surgical and medical therapy.
In five (18%) of the children included in the report by Brook and co-workers , meningitis occurred that was associated with B. fragilis (two children), P. acnes (two children), and Peptostreptococcus species (one child). A direct extension of the organism from an infection site to the meninges might have occurred in two of these children, both of whom had surgical drainage of local collection of pus. One of these children had pansinusitis and required a Caldwell–Luc procedure, where a direct extension of the inflammation to the sub-dural space through the cribriform plate was demonstrated. Ethmoid drainage and frontal craniotomy yielded pus from the sinus as well as from the subdural space.
Anaerobic organisms recovered from blood were isolated from other infected sites in 16 (57%) of the patients reported by Brook and coworkers . In eight of the 16 patients, anaerobic bacteria were mixed with other anaerobic and/or with aerobic organisms (two to five bacteria/specimen of pus). Extravascular sites from which anaerobic organisms were recovered included abscesses (four patients), cere-brospinal fluid (three patients), peritoneal fluid (four patients), tracheopulmonary aspiration (two patients), sinuses (two patients), and sinus and subdural empyema (one patient). Seven of the eight children who had soft-tissue abscesses or local collections of pus required surgical drainage. Some of these children had recurrent or persistent bacteremia until proper surgical drainage was performed. Four patients also had extravascular collections of pus, however anaerobic organisms were not recovered from these sites, either because anaerobic cultures were not obtained or because the specimens were inappropriately transported.
Shanks and Berman reported two children with multiple pulmonary abscesses who developed hematogenous spread from head and neck infections . Porphyromonas asac-charolytica was isolated from the blood of one child, and B. fragilis from the other child.
Bacteroides fragilis, Peptostreptococcus sp., Clostridium sp., and Fusobacterium sp. are the most common clinically significant anaerobic isolates causing anaerobic bacteria in children. The strains of anaerobic organisms recovered depended largely on the portal of entry and the underlying disease. Predisposing conditions to anaerobic bacteremia include: neoplasms, immunodeficiencies, chronic renal insufficiency, decubitus ulcers, perforation of viscus and appendicitis, and neonatal age. Organisms identical to those causing anaerobic bacteremia often can be recovered from other infected sites that may serve as a source of persistent bacteremia. When anaerobes resistant to penicillin are suspected or isolated, antimicrobial drugs such as clindamycin, chloramphenicol, metronidazole, cefoxitin, a carbapenem, or the combination of a penicillin and a beta-lactamase inhibitor should be administered. The early recognition of anaerobic bacteremia and administration of appropriate antimicrobial and surgical therapy play a major role in preventing mortality and morbidity in children.