Pediatric Clostridium difficile: a phantom menace or clinical reality?

Abstract

Clostridiumdifficile is the leading cause of nosocomial gastrointestinal illness in adult patients in hospitals. Even though C. difficile disease in adults has been well studied, research on pediatric C. difficile disease is still in its infancy. For many years it has been believed that C. difficile was a disease that affected only adults and was not a problem for children. This erroneous belief arose from the observation that neonates acquire C. difficile quickly (within 48 hours of birth) but show no intestinal symptoms (1–4). More recent evidence has been documented in case reports of pediatric C. difficile and outbreaks of C. difficile disease in pediatric populations (5–7). Pediatric C. difficile disease has also been associated with the occurrence of severe complications and high mortality rates (8–10). The cascade of events in the pathogenesis of C. difficile disease is similar in children and adults. Normal intestinal microflora are disrupted by antibiotic exposure, medications, or surgery. If the child is then exposed to C. difficile (or its spores), colonization may occur, with production of toxins A and B. These toxins act on enterocytes, causing an inflammatory response and morphologic changes that lead to diarrhea or colitis. Host factors (age, diet, immune response) play an important role in determining whether C. difficile develops into asymptomatic carriage or active disease. Treatment for pediatric C. difficile disease usually relies on metronidazole or vancomycin, but clinical guidelines have not been defined for the pediatric population (11). As in adults, recurrent C. difficile disease that does not respond to conventional therapy develops in a proportion of children treated with antibiotic therapy. EPIDEMIOLOGY Acquisition of Clostridium difficile During the first few weeks of life, as many as 67% of infants become colonized with C. difficile if they are delivered in the hospital (1,4,12,13). Older children and adults are protected from colonization by C. difficile by the ability of the normal intestinal microflora to inhibit the overgrowth of pathogenic organisms by a phenomenon called colonization resistance (14,15). C. difficile disease does not usually occur in adults unless antibiotics or other factors disrupt this colonization resistance (16,17). However, because the neonatal intestine is immature and does not have the protective milieu of normal microflora established in adults, the neonate is susceptible to bacterial colonization and overgrowth. The source of the C. difficile is usually not the mother, but rather exposure to C. difficile spores in the hospital nursery (1,18,19). Delmee et al. (2) found that 60% of neonates in a hospital nursery had C. difficile colonization, and 50% of the environmental surfaces were positive for C. difficile spores. Prevalence and Incidence The frequency of pediatric C. difficile is dependent on the age of the children (Table 1). Neonates (birth through 1 month of age) have high frequencies (up to 64%) of C. difficile colonization, but are usually asymptomatic carriers (3,4,13). In infants (<2 years of age), the prevalence ranges from 3% to 62%, but more infants with colonization have symptomatic disease (Fig. 1). In older children (3–18 years of age), the prevalence is similar to adult frequencies (5–8%), with similar distributions of symptomatic and asymptomatic carriers.TABLE 1: Frequency of pediatric Clostridium difficile disease from different patient populationsFIG. 1.: Pathophysiology of toxin A and B–mediated Clostridium difficile disease. Toxins cause fluid secretion, inflammation, increased permeability, and activation of neuroimmune cells through several pathways. Data compiled from references 94–97.Outbreaks Several outbreaks of pediatric C. difficile have been described in the literature (1,5,20,21). Delmee et al. (2) prospectively observed children admitted to one neonatal ward at a Belgian hospital for 6 months and found that 76 (67%) of 114 of the neonates acquired C. difficile(2). Once the isolates were identified, 83% of the strains were in one of two serogroup types and 85% of the environmental isolates were also of these two serotypes (2). No significant association was found between acquisition of C. difficile and the development of intestinal symptoms in this neonatal population, because of a high frequency of asymptomatic carriers. However, there were two cases of severe neonatal necrotizing enterocolitis in these neonates. Ferroni et al. (5) described an outbreak of nosocomial diarrhea that occurred in a pediatric orthopedic service in France. Of the 37 children with nosocomial diarrhea, 6 had documented nosocomial C. difficile. These children were aged from 8 months to 18 years, with a mean age of 9.1 ± 7.9 years and five of six (83%) were boys. All C. difficile isolates belonged to serogroup C. The only significant risk factor was lincomycin, and when this antibiotic was discontinued and increased infection control measures were begun, the outbreak ceased. Larson et al. (1) prospectively observed 451 newborn infants in five postnatal wards in the United Kingdom. The acquisition of C. difficile ranged from 6% to 52%, depending on the ward location. In total, 65 (14.4%) of 451 of the neonates were positive for C. difficile. A cluster of 32 infants on one ward was found, with 52% of the infants positive. The common source of C. difficile was identified to be the nursery's infant bath. In this study, the isolates were not identified. None of the vaginal cultures from the mothers were positive for C. difficile. Frequency of diarrhea was not reported in this study. Small outbreaks of C. difficile diarrhea have also been reported in pediatric oncology wards. Cartwright et al. (6) reported an outbreak involving six patients in which five of the six were infected with the same strain as identified by polymerase chain reaction (PCR) ribotyping. Pediatric outbreaks have occurred in day-care centers. Kim et al. (7) reported five outbreaks of diarrhea in three day-care centers that involved 65 children aged more than 2 years (7). Of the 65 children, 13 had C. difficile diarrhea and 4 had asymptomatic carriage. When children in one day-care center were prospectively observed for 13 weeks and cultures obtained, 50% acquired C. difficile disease. Thus, as in adults, outbreaks of pediatric C. difficile disease have been well documented. CLINICAL PRESENTATION The result of colonization by C. difficile may be separated into four categories: 1) asymptomatic carriage, 2) acute and protracted diarrhea, 3) colitis (pseudomembranous colitis [PMC], fulminant colitis, toxic megacolon, and non-PMC colitis), and 4) recurrent infections. Although there is a high carriage rate in neonates, symptomatic disease is uncommon (21,22). There are several theories to explain why neonates do not exhibit symptoms, despite high colonization rates. C. difficile produces two toxins, A and B, that cause cytokine release, changes in enterocyte morphology, and the activation of the enteric nervous system, which may result in diarrhea (Fig. 1). Receptor sites for the two toxins are located on the epithelial surfaces of the intestine. Several researchers have shown that it is not the absence of receptor sites per se that accounts for the failure of the disease to develop in neonates; receptor sites for C. difficile toxins have been detected (2,18,23–26). The absence of disease may be related to the immaturity in neonates of the toxin receptor sites, which are not able to bind the toxins (22). In addition, a neutralizing effect of maternal antibodies may play a role (22,26). The absence of an inflammatory response may also be caused by the immaturity of the neonatal immune system itself (22). Diet may also play an important role, although this area has not been well studied. Specific dietary substrates may be needed for full expression of the toxins and production of disease. Mahe et al. (27) found that mice fed a complex commercial diet died in response to a challenge with C. difficile, whereas those mice fed a simplified semisynthetic diet survived after C. difficile exposure. That there are no complex substrates in neonatal diets may be one explanation for the absence of symptoms in neonates, even though C. difficile is frequently present. Pediatric C. difficile diarrhea occurs at highest frequencies when the intestinal microflora is becoming established and changes in dietary sources are made. A change from a maternal milk diet to semisolid food was associated with a change in strain types of C. difficile carried by an 11-month-old infant, but the child remained asymptomatic (19). Acute and protracted diarrhea in children may be similar to the symptoms seen in adults, in that diarrhea is observed, and the duration usually ranges from 2 to 9 days (5). Most pediatric patients become symptomatic a few days to weeks after antibiotic exposure, if antibiotics have been administered. Mitchell et al. (28) studied 76 children who were receiving amoxicillin-clavulanate for otitis media, and diarrhea developed in 15 (68%) of the 22 children within 2 days of antibiotic initiation but all 22 developed diarrhea within 7 days. However, many occurrences of pediatric C. difficile disease are not associated with antibiotics (see Risk Factor section), and the incubation period is therefore unclear. C. difficile has also been reported to cause chronic diarrhea, repeated attacks of colic, and other intestinal symptoms in children. Buts et al. (29) reported, in a series of 19 infants with C. difficile disease, 15 had persistent diarrhea, and 4 had evidence of malnutrition, failure to grow and poor appetite (29). Other symptoms in these children included recurrent colic (n = 11), hypermeteorism (n = 13), abdominal distension (n = 8), and repeated emesis (n = 7). In another study in children infected with C. difficile between the ages of 7 weeks and 7 years, diarrheal stools were noted for periods ranging from 7 weeks to 19 months (18). Sutphen et al. (30) observed seven children with C. difficile and found that the diarrhea persisted for an average of 4.5 months before it was correctly diagnosed. Whether the persistent diarrhea in some children is due to undiagnosed C. difficile or to unresponsiveness to treatment is unclear. In some cases, diarrhea is not the chief symptom of pediatric C. difficile disease. Qualman et al. (8) reported four children with C. difficile and Hirschsprung's disease that manifested with obstruction instead of diarrhea. Symptoms of pediatric colitis may include pseudomembranous colitis (PMC), toxic megacolon, and non-PMC colitis. Symptoms of colitis may include profuse watery diarrhea, intense abdominal pain, hyperthermia, edema, and ascites from protein-losing enteropathy and neutrophil leukocytosis (9,22). The age range for children who have PMC is broad. The disease has been reported in children aged 5 days to 17 years (21). Infrequent cases of fulminant PMC have been reported in neonates, most associated with endotoxemia (8,10). The onset of PMC is usually during or at the end of the inciting antibiotic treatment. The most commonly implicated antibiotics are similar to those in adult patients (penicillins, cephalosporins, or lincosamides). In children with PMC, the consequences may be severe. Zwiener et al. (9) found of 43 pediatric patients with PMC, 9 (21%) died. Qualman et al. (8) also described seven cases involving fatal PMC in children aged 13 days to 17 years. Recurrent C. difficile disease in children has been reported in the literature, and recurrence rates have been reported from 15% to 57% after standard antibiotic treatment (30,31–35). Tvede et al. (36) reported two pediatric patients with C. difficile diarrhea in whom chronic diarrhea necessitated vancomycin treatment, but both patients had several relapses after the vancomycin was discontinued. Brunetto et al. (20) studied 29 cases of C. difficile in pediatric oncology patients and found 11 (37.9%) patients with a reported history of relapses. Of the 21 symptomatic patients who were then re-treated with vancomycin (125 mg three times daily), 9 (43%) had recurrences. Some studies do not report recurrences, which may be because there was no proper follow-up or because of a small sample size in the study. In a series of four children with C. difficile diarrhea (aged 7–24 months), all four responded initially to metronidazole treatment (40 mg/kg body weight for 10 days), and no recurrences were noted during a 2-month follow-up (34). Unfortunately, documentation of the rate of recurrences of pediatric C. difficile disease is infrequent in the literature and is usually only reported as a minor point. Whether recurrent C. difficile disease in children is as large a clinical problem as it is in adults is presently not known. Complications of pediatric C. difficile disease are periodically described in the literature but not in great detail. Protein-losing enteropathy with increased levels of fecal α1-antitrypsin have been found in asymptomatic infants with C. difficile(18). α1-Antitrypsin is a marker for subclinical enteric protein loss, and thus, even infants with no diarrhea may cause clinical concern. Other complications that have been reported include rectal prolapse (9,37), chronic osteomyelitis caused by C. difficile in a patient with cancer (38), ascites reported in a 2-year-old infant with PMC (9), intestinal perforation (8), and necrotizing enterocolitis (39). Patients with cystic fibrosis are at risk for C. difficile infection, most likely because of the regular use of antibiotics. In fact, it is known that approximately 50% of children with cystic fibrosis harbor C. difficile(40,41). This colonization only occasionally results in disease in these circumstances, although four cases of fulminant colitis have been reported recently (42). RISK FACTORS As in adults, risk factors for pediatric C. difficile disease may be broken down into several areas: factors that disturb the normal microflora, host factors (age, immune response, diet), and concurrent infections or medical conditions. Disruption of the Normal Colonic Microflora Antibiotic-Associated Disruption Pediatric incidences of C. difficile disease that have been associated with antibiotics typically occur from 4 to 18 days after the first antibiotic dose (5). Use of lincomycin was found to be associated with a pediatric incidence of C. difficile disease (5). Thompson et al. (43) studied 208 pediatric patients with suspected diarrhea for whom stool samples were submitted to the clinical microbiology laboratory for C. difficile assay and found 18 (8.6%) were positive. The patients receiving antibiotics had significantly greater recovery of toxin B than those who did not receive antibiotics (18% vs. 4%, P = 0.001). However, in this study no further investigation into the type, dose, or duration of antibiotic was documented. The types of antibiotics that should be suspected of inciting C. difficile disease are shown in Table 2. Most of these reports are from hospital populations of children and adults; prospective studies in children are rare.TABLE 2: Anbiotics associated with Clostridium difficile diseaseNon–Antibiotic-Associated Disruption Unlike in adults with C. difficile disease, prior antibiotic exposure does not seem to be a prerequisite for disease development in children. In a study of pediatric outpatients, 78% of those with C. difficile disease had not been exposed to antibiotics in the prior month (44). There have been several case reports of children with C. difficile disease who have had no exposure to antibiotics before the onset of symptoms (45,46). Delmee et al. (2) notes in a prospective study of one neonatal ward in Belgium, noted no association between antibiotic exposure and C. difficile acquisition. In 82 neonates who were in the ward for at least 1 week, only 40 (49%) had been exposed to antibiotics (2). Other exposures may have been responsible for disrupting normal colonic flora, as in one patient with an anticholinergic overdose that precipitated C. difficile disease (45). Host Factors Age The frequency of symptomatic C. difficile disease varies sharply depending on the age of the children and seems to peak at ages ranging from 6 months to 2 years (Fig. 2). Neonates acquire C. difficile easily in neonatal units. In one prospective study, Enad et al. (12) showed that 48% were colonized with toxin-negative C. difficile, and 52% were infected with toxin-positive strains. Neonates with toxin A-positive results had more days of diarrhea (8.2 ± 5.7 days) versus those with toxin negative results (2.2 ± 2.2 days, P < 0.001). Neonates with toxin A–positive results were smaller and less mature, with longer hospital stays. Thus, this study challenges the long-established concept that no pathogenicity from C. difficile exists in the neonatal population. Karsch et al. (47) studied 766 children in hospitals and reported 17.3% of the children less than 12 months of age had C. difficile diarrhea. The frequency was found to decrease if the children were older (5.6% for ages 12–24 months and 2.7% for ages >24 months). Dutta et al. (34) observed 111 pediatric patients in a hospital in Calcutta, India, and found no incidents of C. difficile disease in children 0 to 6 months of age, an occurrence in 4.5% of the children aged 7 to 24 months, and no occurrence the older children (25–60 months). Hyams et al. (48) observed 115 outpatients receiving antibiotic therapy for otitis media, and C. difficile diarrhea developed in 28% of the children less than 1 year of age, whereas it developed in only 3% of the infants 1 to 6 years of age. A close association with age seems plausible, because the development of the anaerobic intestinal flora does not reach the complexity seen in adults until the second year of life (18).FIG. 2.: Ages of occurrence of Clostridium difficile in children. Data are presented from literature listing the individual age of each case. Distribution is not representative of prevalence or incidence, because ages of asymptomatic carriers are usually not listed. Often ages were grouped from neonate to 2 years of age and thus the data were not presented here. Compiled from references 3, 8, 9, 29, 32, 33, 50, 54, 72.Because of the high frequency of asymptomatic carriage (similar to adults with C. difficile), age may not reach significance as a risk factor if data from children of vastly different ages are combined and frequencies of diarrhea and asymptomatic carriage are compared. For example, Boenning et al. (44) studied pediatric outpatients aged from 2 weeks to 16 years and reported that the mean age did not differ significantly in children with C. difficile diarrhea compared with the mean age of control subjects without diarrhea (9.8 months versus 8.2 months, respectively). Cerquetti et al. (3) tested 753 stool samples from young infants (median age, 18–21 months) and found C. difficile toxin B in 6.6% of the control subjects without diarrhea and in only 3.7% of the infants with diarrhea. Many studies report incidences of C. difficile as occurring in “ages less than 2 years” or just “in infants,” which may mask the differences in C. difficile disease rates in neonates, weaning infants, and older children. Mucin In addition to the barrier effect by the normal intestinal microflora, the thickness and composition of the mucin layer may be associated with more asymptomatic carriage by infants. The mucin layer has been shown to be protective against C. difficile(49). Fatalities in cases of pediatric C. difficile disease have been associated with altered mucosal mucin (8). Diet Young age may be associated with an increased risk of C. difficile disease caused by differences in diet. Breast-fed infants have C. difficile colonization less frequently than formula-fed infants. Tullus et al. (26) found that significantly more bottle-fed infants (39%) were C. difficile positive at 6 months of age than breast-fed infants (19%, P < 0.001) at 6 months, and that of those carrying C. difficile at 6 months, 27% had diarrhea. Immune Response Transient hypogammaglobulinemia normally occurs in neonates and young infants (<2 years) until such time as the immune system of the children reaches full maturity. Gryboski et al. (50) showed that children with recurrent C. difficile disease have lower levels of γ-globulin than do children with no disease. In this study of 43 children with C. difficile disease, 15/43 (35%) had low serum immunoglobulin IgA levels and 12 (26%) had low serum IgG levels. The children with hypogammaglobulinemia were significantly younger (mean of 18.8 months; age range, 2–70 months) than 28 children with normal γ-globulin levels (mean age, 4.4 years). Children with hypogammaglobulinemia had a higher frequency of C. difficile recurrences (7/15, 47%) compared with children with normal IgG levels (5/28, 18%). The ability of the immune system to mount protective levels of serum IgG against C. difficile toxin A has been shown to be important in adults, but has not been documented in children (51). The immaturity of the neonatal immune system may also result in lower levels of C. difficile-specific antibodies. Leung et al. (32) found that serum antibody to C. difficile toxin was present in only 19% of children less than 2 years of age (even though the prevalence of C. difficile in that age group is high). In contrast, 64% of children more than 2 years of age and adults had serum antibodies to C. difficile toxin, although the carriage rate is usually lower. Concurrent Medical Conditions or Diseases Cancer C. difficile disease has been reported in patients in pediatric oncology for the past two decades (8,20,52,53). Burgner et al. (54) studied 149 stool samples from 60 symptomatic pediatric patients, and 58 stool samples from 44 asymptomatic patients admitted as oncology inpatients to the Royal Alexandra Hospital for Children in Australia. Only 8.7% of the children with diarrhea had C. difficile isolated, and 19% of the asymptomatic patients were found to be carriers of C. difficile. No association was found with the use of antibiotics or chemotherapy and C. difficile. The only risk factors significantly associated with C. difficile disease compared with asymptomatic carriers were younger age (39.5 months vs. 68.2 months;P < 0.02) and a shorter hospital stay (4.3 days vs. 14.3 days, P < 0.05). Burgner et al. concluded that C. difficile disease was not a major disease in this population. However, several studies have also reported C. difficile disease in pediatric oncology wards (6,20,52,55). In one report of seven cases of fatal PMC, three were associated with lymphoma, leukemia, or Hodgkin's disease (8). Kavan et al. (56) described a case involving a 13-year old boy with Hodgkin's disease who had undergone bone marrow transplantation with subsequent development of C. difficile diarrhea. He had also been treated with intravenous antibiotics and oral amoxicillin. Whether the bone marrow transplantation, Hodgkin's disease, or his immunosuppressed state was associated with C. difficile disease is unknown. Transplantation C. difficile disease has been reported in pediatric patients who undergo renal transplantation. Chavers et al. (57) studied 164 pediatric patients who underwent renal transplantation at the University of Minnesota. The most common bacterial infection in these patients, who were less than 5 years of age, was C. difficile. In the first 2 weeks after surgery, C. difficile diarrhea developed in 13 (42%) of 31 of the children less than 2 years of age and in 3 (30%) of 10 children from 2 to 5 years of age. When patients were observed for 1 to 6 months after the transplantation, C. difficile remained the most common infection in children less than 2 years of age (n = 24) and in children aged 2 to 5 years (n = 13). Hypogammaglobulinemia has been reported in infants with uremia who undergo peritoneal dialysis. In that group, 46% of the infants less than 2 years of age and 54% of those aged 2 to 5 years were receiving peritoneal dialysis before renal transplantation. None of the children more than 5 years of age had received peritoneal dialysis, and the rate of C. difficile diarrhea in these children was quite low. Other possible risk factors in pediatric patients who undergo transplantation may include the use of immunosuppressive agents and environmental transmission by hospital personnel (57). Underlying Gastrointestinal Conditions Children with C. difficile disease also have been reported to have concurrent intestinal infections or conditions. Several incidents of C. difficile have been reported in children with Hirschsprung's disease (a partial obstruction of the distal colon) (8,18). Rare cases of pediatric C. difficile diseases have been reported with short bowel syndrome and small bowel obstruction (58). Coinfection With Other Enteric Pathogens Evidence is limited and case reports are few, but infections with other causes that occur concurrently with C. difficile have been reported. Tvede et al. found that 18 (56%) of 32 pediatric patients with C. difficile disease had no other enteric pathogens, but 44% had other bacterial pathogens, including Campylobacter, Salmonella, Yersinia enterocolitica, cytopathogenic Escherichia coli, and hookworm (36). In another study of 111 children aged 0 to 14 years, 36% of the children with C. difficile diarrhea carried another stool pathogen concurrently (59). Two other reports showed no association of C. difficile with Salmonella(60,61). DIAGNOSIS The diagnosis of C. difficile disease in pediatric patients is not as apparent as in adults with the disease. Adults with C. difficile disease are typically defined by a recent history of antibiotics, presence of diarrhea (more than three loose, watery stools per day for at least 2 days, presence of C. difficile (culture or toxin A or B), and exclusion of other causes of diarrhea (11). As reviewed earlier, not all children with C. difficile disease have been exposed to antibiotics, and many have concurrent causes of diarrhea. Pediatric Diarrhea The definition of diarrhea in pediatric patients with C. difficile disease varies in the literature from more than two watery stools per 24 hours (28) to more than four unformed stools per 24 hours (34,62). Pediatric diarrhea may also be defined as a fluid content of stools in excess of 10 mL/kg per day. Many studies do not provide a quantitative definition for diarrhea but simply refer to “children with diarrhea.” If the diarrhea continues for more than 14 days, then it may be considered protracted or chronic. Detection of Clostridium difficile Detection may be achieved by several types of assays: microbiologic cultures, toxin assays for toxins A and/or B, endoscopic examinations, or a combination of these. Comparison studies of the different assays for C. difficile have not been well documented in the pediatric population. In adults, most studies have shown that selective culture with cycloserine-cefoxitin-fructose agar (CCFA) or cell tissue cytopathic assays have the highest sensitivities and specificities (Table 3). Broth cultures that stimulate C. difficile spore germination can detect low numbers of C. difficile (>103/g) in the stool (63,64). The cytotoxin assay is considered the gold standard for C. difficile detection and can detect toxin levels as low as 1.6 × 10−2 mg/g stool (11,65). These two tests (culture and cytotoxin) have the advantage of providing high sensitivity, but the results may not be available for 24 to 72 hours, and the tests may not be available at all institutions or clinics.TABLE 3: Reported ranges of sensitivities and specificities for diagnostic assays for Clostridium difficileIn an effort to overcome these hurdles, rapid enzyme immunoassays (EIAs) have been developed. Although their sensitivities and specificities are not as high (Table 3), the results are available within 24 hours, and their predictive values are acceptable. However, in a recent study, Kader et al. (66) retrospectively analyzed all stool specimens submitted to the Children's Hospital in Philadelphia during a 20-month period and found that toxin detection alone (either A or B) may not be sufficiently sensitive in pediatric patients. Of 1061 stool specimens from 1031 children, 276 (26.8%) specimens were positive for at least one C. difficile toxin. Of the 276 positive stool samples, 51 (18.5%) were positive for only toxin A, 133 (48.2%) were positive for only toxin B, and 92 (33.3%) were positive for both toxins A and B. The frequency of toxin B detection was significantly higher in older children, but not in infants. Kader et al. concluded that assay for toxin A along would identify only 51.8% of the C. difficile infections. Tissue culture assay for toxin B alone would identify 81.5% of the C. difficile infections. The conclusion of this study is that it is necessary to test for both toxins, that either toxin A positive or toxin B positive (or both) results should be accepted, and that both toxins need not be positive before C. difficile is correctly detected. A note of caution regarding the diagnosis of C. difficile disease: simple detection of the organism and/or its toxin is not diagnostic by itself. An empirical decision should be made based on both the laboratory detection of C. difficile and the clinical presentation of the disease. If the clinical presentation suggests C. difficile, but the laboratory assay is negative, from our experience, at least three more samples should be assayed before C. difficile is ruled out. If laboratory assays detect C. difficile, but the clinical picture does not support disease, asymptomatic carriage is present. Exclusion of Other Causes of Diarrhea Unlike in adults with C. difficile disease, the exclusion of other causes of diarrhea is not as conclusive in pediatric patients, because children often have underlying gastrointestinal or immunologic diseases (67). Other Methods of Diagnosis Sigmoidoscopic or colonoscopic examinations are usually reserved for seriously ill children or for disease that defies noninvasive diagnosis. C. difficile is responsible for nearly all incidences of PMC, and therefore if pseudomembranes are found, the diagnosis is fairly certain. Pseudomembranous colitis has characteristic raised yellowish white plaques (Fig. 3) that are 2 to 10 mm in diameter (11). As shown in Figure 4, computed tomographic (CT) scans have been used to diagnose C. difficile in children, but the results were less specific than laboratory and clinical results (68), and severity on CT scan does not always correlate with clinical severity (69). The CT scan findings in nine children with C. difficile disease included nodular haustral thickening (44%), accordion patterns (22%), colonic wall thickening (33%), ascites (11%), and pericolonic edema (33%).FIG. 3.: Endoscopic view of pseudomembranous colitis caused by Clostridium difficile. Plaques 0.2 to 2.0 cm in diameter are elevated, adherent and yellow-white. (Printed with permission of Dr. Christina Surawicz, Harborview Medical Center, Seattle, WA, U.S.A.).FIG. 4.: Computed tomographic scan of a 10-year-old boy child with Clostridium difficile disease. The ascending and descending colonic mucosa are swollen (Printed with permission of Dr. Phillip Tarr, University of Washington, Seattle, WA, U.S.A.).STANDARD TREATMENTS Discontinuation of the Inciting Antibiotic The universal recommendation in adults is to discontinue, whenever possible, the antibiotic that is initiating the C. difficile disease. Although such a recommendation is reasonable in children, it should be stressed that most pediatric C. difficile disease is not associated with antibiotics (31). Metronidazole In adults, the first episode of C. difficile is usually treated with metronidazole or vancomycin (11,70). Metronidazole is rapidly absorbed from the upper small intestine and, although high levels are achieved in tissue, lower levels are found in the lower intestine. Despite this, metronidazole has been found to be as effective as vancomycin in adults with C. difficile disease and is 50% less expensive (70,71). Metronidazole is associated with a higher frequency of side effects, including nausea, metallic taste, emesis, extremity numbness, convulsive seizures, and neuropathy. Metronidazole (20–40 mg/kg per day) has also been prescribed for infants and children (34,72). Vancomycin Vancomycin is not absorbed to a great extent after oral administration, causes infrequent side effects, and is highly effective in reducing the C. difficile concentration in the colon. However, recurrences in adults may range from 20% to 65%, depending on the history of the patient (73). Concerns about the expense of vancomycin and the increasing frequency of vancomycin-resistant pathogens have limited its use (74). For adults with recurrent C. difficile disease, a tapering dose of vancomycin (prolonged course with a gradual reduction in dose) or intermittent short pulses (2–3 days of full doses, followed by one dose every 3 days for 2–4 weeks) have been recommended (11). In infants and children, there are no clear recommendations. Vancomycin is usually prescribed (40 mg/kg per day) in only severe occurrences of PMC, or recurrent colitis, or in immunocompromised children (21,72). It has been administered to children in doses ranging from 500 mg to 3 g per day for 3 to 10 days (5). Recurrences of C. difficile have been noted in several studies, with the frequency ranging from 43% to 67%, after vancomycin is discontinued (20,30,36,73). INVESTIGATIONAL TREATMENTS Because C. difficile disease arises from the disruption of the intestinal flora, replacement with beneficial microbes or biotherapeutic agents has been investigated (75,76). This approach is appealing, because these microbes replace disturbed intestinal flora and have less impact on the recovery of the normal flora than antibiotics. Several types of investigational microbes show promise for pediatric C. difficile disease. Saccharomyces boulardii Buts et al. (29) studied 19 pediatric patients (aged 2–32 months; median age, 8 months) who had chronic diarrhea (lasting longer than 15 days) and C. difficile toxin B identified as the sole causative agent for the diarrhea. S. boulardii was administered orally for 15 days, with dosage according to the age of the child (500 mg/day if <1 year of age, 750 mg/day if 1–4 years, and 1 g/day if >4 years). No antibiotics (vancomycin or metronidazole) were administered during the treatment part of the trial. Within 1 week, symptoms resolved in 18 (95%) of the children. Clearance of the toxin B was observed within 15 days in 16 (85%) of 19 of the patients. No side effects were noted during the study. Two patients (11%) had subsequent relapse of disease, which resolved rapidly after a second treatment with S. boulardii. No other studies of S. boulardii in pediatric patients with C. difficile have been published. Lactobacillus rhamnosus GG Four children (aged 5–70 months; mean, 34 months) with a history of recurrent of C. difficile were treated with Lactobacillus rhamnosus strain GG (72). Lactobacillus GG was administered orally for 2 weeks at doses of 125 mg twice daily. No antibiotics (vancomycin or metronidazole) were administered during the treatment part of the trial, but all patients were treated with antibiotics in the follow-up period (median of 9.5 months). In all four children the disease responded clinically within 5 to 7 days, and they were asymptomatic by the end of the 2-week treatment. No side effects were noted during the study. Two (50%) of the children had relapse within 2 months after Lactobacillus GG was discontinued. No other studies have been published of Lactobacillus GG in pediatric patients with C. difficile. Immunoglobulin G Treatment It has been shown that immunoglobulin levels (serum IgA or IgG) are lower in adults with C. difficile disease than in asymptomatic carriers. Adult patients with recurrent C. difficile have also been shown to have low levels of IgG to toxin A (77–79). Adults with disease have been successfully treated with intravenous therapy with immune globulin, but controlled double-blind trials have not been reported (78,80). Leung et al. (32) studied six children with relapsing C. difficile colitis (median duration, 7 months; range, 4–32 months) who were treated at Children's Hospital in Boston. These six children were found to have lower IgG anti-toxin A levels (median 0.13 units) compared with 24 healthy children, who had a median of 0.23 units (P = 0.03). Five of the six symptomatic children were treated with 400 mg/kg intravenous γ-globulin every 3 weeks. No antibiotics were used during the first 3 weeks of the study. The children showed significant increases (from 0.13 units before treatment to 0.94 units after treatment) in serum IgG levels, but not serum IgA anti-toxin A levels (from 0.13 units before treatment to 1.5 units after treatment). All five children showed clinical resolution of the diarrhea, but one child had relapse of disease after IgG treatment. Controlled double-blind trials using IgG treatments have not been reported. Colonic Irrigation Liacouras and Piccoli (81) reported two children with recurrent C. difficile who had been treated with whole-bowel irrigation, which was administered until profuse, clear liquid stools were produced, followed by a 3-week course of oral vancomycin and Lactobacillus GG. In both instances, the child became asymptomatic within 3 days of therapy and remained symptom-free until the end of follow-up (36 and 48 months). The authors concluded that whole-bowel irrigation may clear active C. difficile, toxins, or spores from the colon and may be effective as an adjunctive therapy. There have been no further reports of this treatment. Other Cholestyramine has been tried in a few children but the studies are limited and have not been placebo-controlled trials (82,83). Cholestyramine also binds vancomycin, and therefore these two treatments should not be prescribed concurrently. CONCLUSION There is increasing recognition of the role of anaerobic bacteria in pediatric infections. C. difficile disease in pediatric patients is now accepted as an major cause of disease, ranging from mild, self-limited diarrhea to life-threatening colitis. However, a number of issues remain unsolved. First, regarding the long-established notion of the frequent asymptomatic colonization of newborns and infants, the specific mechanism remains unknown. Consequently, it is not possible to make recommendations for management in an individual case of C. difficile-associated diarrhea in this very young age group. Secondly, the optimal treatment strategy has not been identified, and although it appears that metronidazole is the drug most commonly prescribed, solid data on its use (dose and duration), as well as on the frequency of recurrences, are not available. Furthermore, the promising role of biotherapeutic agents also appears to be a fertile, but largely unexplored, area. It is clear that this fascinating field needs well-designed clinical trials to answer the many questions that remain unresolved.