History

In 1935, Hall and O’Toole described the identification of a new, anaerobic, gram-positive, spore-forming bacillus in the stools from four of 10 healthy, breastfed neonates. They named it Bacillus difficilis, in recognition of the fact that the organism was difficult to isolate and study in the laboratory. B. difficilis was noted to be highly pathogenic in guinea pigs and rabbits; guinea pigs inoculated subcutaneously with the bacteria commonly developed convulsions, which the investigators attributed to a soluble exotoxin isolated from broth culture filtrates. A link between intestinal colonization and illness in human infants was much less clear, although the authors postulated a potential association with occult blood in the stools of neonates or convulsions of unexplained origin. Neither association was proved, and, for decades, B. difficilis was thought to be a harmless intestinal commensal in infants. The role of C. difficile as a human pathogen was finally elucidated in the 1970s, when toxin-producing Clostridia were isolated from the stools of adult patients with clindamycin-associated pseudomembranous colitis.

Epidemiology

The epidemiology of CDI is evolving. Since 2000, cases of CDI have increased in both frequency and severity. From 2000 to 2009, the number of hospitalized patients with CDI doubled. Deaths attributable to CDI increased from 3000 deaths per year during 1999 to 2000 to 14,000 during 2006 to 2007. In 2011, C. difficile was responsible for nearly 500,000 infections and approximately 29,000 deaths. As in prior years, the highest burden of disease and the majority of deaths occurred in adults 65 years of age and older.

Nevertheless the burden of CDI in children is substantial. While CDI has historically been less common and less severe in children, since 1997 rates of pediatric CDI have increased in both hospitals and the community. In hospitalized children, CDI is associated with increased mortality, longer lengths of stays, and higher costs. CDI has been reported in children previously thought to be at low risk for the disease, including those without prior antibiotic or hospital exposure, as well as infants.

Forty to 70% of asymptomatic, healthy newborns may be colonized with C. difficile in the first 10 to 28 days of life. Strain acquisition appears to be highest in the first week of life, likely as a result of exposure to spores in the hospital. Colonization rates decrease to 3% to 10% by the second year of life. Rates of C. difficile colonization in children older than 2 years of age approximate those in healthy adults and may be as low as 2% to 3%.

Colonization occurs with both toxin-producing and nontoxigenic strains. Factors associated with increased colonization rates in infants include longer duration of hospitalization, hospitalization in an intensive care setting, low birth weight, and formula feeding. Even with a large number of toxigenic C. difficile bacteria and high levels of toxin A and B, infants usually remain asymptomatic, and most studies have failed to show an epidemiologic association between colonization and disease in infants younger than 1 year. For example, in a study of outpatient children, C. difficile was isolated from 7% of patients with diarrhea and 14.8% of healthy controls. Children with C. difficile were younger than children without the organism (mean age, 8.2 to 9.8 months). C. difficile was isolated with equal frequency in healthy Swedish children 1 week to 1 year of age (17%) and children younger than 6 years with diarrhea (18%). Most controlled studies in neonatal intensive care units corroborate high rates of asymptomatic colonization in young infants. C. difficile toxin was recovered from the stools of 55% of patients in one neonatal intensive care unit, but signs of enteric disease, including necrotizing enterocolitis (NEC), occurred with equal frequency in both toxin-positive and toxin-negative infants. Although severe CDI has occasionally been reported in young infants, especially those with underlying gastrointestinal disease, it has historically been thought to be rare. Guidelines published in 2013 from the American Academy of Pediatrics recommended avoiding routine testing for C. difficile in children younger than 1 year and testing for other causes of diarrhea first in children 1 to 3 years of age.

Recent reports challenge traditional paradigms about CDI; namely, that (1) CDI is uncommon in children, (2) that young children are spared, and (3) that most cases occur in hospitalized children. C. difficile is a common health care–acquired pathogen and the most common cause of health care–associated diarrhea in pediatric patients. Outbreaks of CDI among hospitalized pediatric patients are well documented, and cases of CDI in hospitalized children are increasing. In a study that included 21 free-standing children’s hospitals, CDI increased 53% in hospitalized children between 2001 to 2006, from 2.6 cases to 4.0 cases per 1000 admissions. Twenty-six percent of the children treated for CDI were younger than 1 year, and 5% were newborns. An analysis of two large administrative databases demonstrated a nearly twofold increase in CDI-associated hospitalization between 1997 and 2006. Children younger than 1 year had the second highest rate of CDI hospitalizations. Investigators acknowledged that their study methodology precluded an assessment of whether these children had true CDI or C. difficile colonization in the setting of diarrhea attributable to another cause. Nevertheless significant increases in CDI in hospitalized children have been documented even when children younger than 1 year are excluded.

Population-based studies also suggest significant increases in the incidence of CDI in children. In Olmsted County, Minnesota, the age- and sex-adjusted incidence of CDI in children 0 to 18 years of age increased from 2.6 per 100,000 person-years (1991–97) to 32.6 per 100,000 (2004–09). The median age of C. difficile cases was 2.3 years. Similar findings were reported in Monroe County, New York, where the incidence of CDI in pediatric patients was 33.8 per 100,000 in 2010, rose to 45.8 in 2011, and remained stable in 2012. The highest incidence rate was in 1-year-old children. In both Olmsted and Monroe Counties, more than 70% of cases were classified as community associated.

Some experts suggest that the apparent increase in CDI incidence be interpreted with caution because widespread use of sensitive polymerase chain reaction (PCR) tests may be leading to overdiagnosis and artificially inflated incidence rates, especially in community-associated disease. In a single-center study, 23% of children with community-associated CDI diagnosed by tcdB PCR had discordant results when stool samples were tested by other methods, including multiplex PCR. Investigators reported that these children lacked traditional risk factors for CDI, had symptoms more consistent with viral illness, and may have been misdiagnosed.

The largest active population-based study of CDI in children to date analyzed data collected from 10 geographic areas in the United States during 2010 to 2011. The highest burden of disease was observed in children 12 to 23 months of age. Clinical presentation, illness severity, and outcomes were similar across age groups, suggesting that isolation of C. difficile from young children in this study represented infection rather than colonization. Antibiotic use and outpatient health care exposures (i.e., visit to doctor’s office, dental office, or emergency department) within the 12 weeks preceding the onset of diarrhea were common among the 71% of cases deemed to be community associated.

Changes in the epidemiology of CDI in children may be related to the emergence of the epidemic BI/NAP1/027 strain. Although this strain is commonly isolated from children it is not yet clear if it is associated with more severe disease, as it is in adults. Ten percent of isolates from hospitalized children with CDI identified through the Canadian Nosocomial Infections Surveillance Program between November 1, 2004, and April 30, 2005, were the B1/NAP1/027 strain. Children with the NAP1 strain were more likely to suffer complications from CDI than were children with other strains (29% vs. 6%; relative risk, 4.6; 95% confidence interval [CI], 1.1–17.2; P = .04), although none died. Similar rates of NAP1 isolation were identified in a prospective study of children at two US hospitals with CDI. NAP1 was not associated with increased disease severity.

Antibiotic use remains a principal risk factor for the development of CDI in individuals of all ages. The risk of CDI increases 7- to 10-fold during a course of antibiotics and for 1 month after antibiotic discontinuation. Although the earliest reports of pseudomembranous colitis were associated with administration of clindamycin, nearly every antibiotic class has been implicated in the development of CDI, particularly cephalosporins, other β-lactam antibiotics, and fluoroquinolones. Diarrhea typically begins during antibiotic use or up to 30 days after antibiotic therapy, but symptoms may occur after a single dose of an antibiotic.

In one study of hospitalized children, receipt of three or more antibiotic classes in the 30 days before diagnosis of CDI was a risk factor for severe disease. Nevertheless antibiotic use is not a prerequisite for the development of CDI in children. Severe disease has been reported in the absence of antibiotic exposure, and in one study of community-acquired CDI, 43% of patients lacked a history of antibiotic use.

The use of gastric acid suppression medication is another potentially modifiable risk factor of CDI. Protein pump inhibitors and histamine-2 receptor antagonists increase the risk of CDI and recurrent CDI. Several studies have demonstrated that CDI is more common in boys. Underlying gastrointestinal disease, particularly Hirschsprung disease, is a risk factor for severe, complicated CDI. Other risk factors for CDI in children include solid organ transplant, inflammatory bowel disease, immune suppression, and cancer. In a prospective study of 141 children undergoing chemotherapy for a solid tumor or lymphoma, nine (6%) tested positive for C. difficile toxin A and were symptomatic. Another study suggests that rates of C. difficile diarrhea in children after chemotherapy are as high as approximately 15%. In an analysis of 3 years of data from the Kids’ Inpatient Database (2000, 2003, 2006), children with cancer accounted for 21% of all CDI cases, and rates of CDI were 15 times higher in children with cancer than in those without cancer.

Pathogenesis

The pathophysiology of CDI is complex and multifactorial. Ingestion of C. difficile spores and subsequent intestinal colonization precedes the development of CDI, but not all exposures result in symptomatic infection. In older children and adults, colonization is inhibited by normal intestinal flora, which is thought to compete for intestinal nutrients and space on the mucosal surface. Antibiotics, by altering normal flora, facilitate C. difficile colonization. Disease manifestations are related to toxin production; most toxigenic strains produce both toxin A and toxin B. Based on early animal studies, toxin A (TcdA) was thought to be the major virulence factor, but recent studies have documented severe CDI caused by toxin A–negative, toxin B–positive (TcdB) strains suggesting that TcdB plays the dominant role in human infections. Approximately 25% of C. difficile isolated by culture do not produce toxins and are not clinically significant.

C. difficile toxins act at the level of the epithelial cells and produce effects by two main pathophysiologic pathways. Toxin A and B cause the disorganization of actin microfilaments of the enterocyte cytoskeleton. The change in enterocyte structure leads to epithelial cell destruction and opening of tight junctions between enterocytes. Toxin A also induces activation of neutrophils and results in local inflammation. Release of proinflammatory cytokines, including interleukin-6 (IL-6) and IL-8, occurs from enterocytes, causing further damage to the intestinal mucosa. Both pathways lead to increased permeability of and damage to enterocytes, resulting in the clinical symptoms of C. difficile.

Why young infants are commonly colonized with toxin-producing C. difficile strains yet rarely have symptoms is not yet completely understood. Colonization density does not appear to be important because asymptomatic infants may have C. difficile colony counts as high as 10 ⁸ bacteria per gram of feces; these colony counts are similar to those in adults with pseudomembranous colitis. Experiments in newborn rabbits suggested that protection against disease may result from lack of receptors for toxin A. The appearance of receptors with age may be a species-specific phenomenon. Toxin receptors have been identified in the intestines of neonatal pigs, and these animals are susceptible to disease. The immaturity of the toxin receptor sites may also play a role in the absence of disease in neonates.

Transmission of C. difficile is fecal-oral. The organism is generally spread by person-to-person contact or environment-to-person contact. Both modes of spread are important in health care environments, where C. difficile is shed in the environment by infected and colonized patients and occasionally from asymptomatic hospital personnel. Patients may acquire C. difficile via the unwashed hands of health care workers, but direct or indirect transmission through contaminated objects and surfaces also occurs. C. difficile spores can survive for up to 5 months in the environment and are difficult to eradicate through routine cleaning and disinfection. Objects likely to harbor organisms are those contaminated with feces and include toilet seats, sinks, and scales, but C. difficile can also be isolated from the hands and feces of asymptomatic hospital personnel. Stethoscopes may be vectors for C. difficile transmission when not effectively cleaned after use on a patient with CDI. Electronic rectal thermometers have also been implicated in hospital spread of C. difficile.

Foodborne transmission has been postulated as a source of C. difficile spread in the community. Although C. difficile spores have been recovered from retail food products, there is limited evidence to link contamination of food to human illness. Likewise the role of zoonotic transmission has been explored because C. difficile is known to be both a pathogen and a commensal organism in domestic and food animals. To date, no study has convincingly demonstrated human C. difficile infection as a result of animal contact, although further research is needed.

Clinical Manifestations

C. difficile can cause a broad spectrum of disease from asymptomatic colonization or mild diarrhea to severe disease, pseudomembranous colitis (PMC), and even death. Watery diarrhea occurs in 90% to 95% of cases of CDI and bloody diarrhea in 5% to 10%. Most pediatric patients develop a self-resolving illness that is associated with low-grade fever, mild abdominal pain, and diarrhea, some of which may contain mucus or blood. However, some patients progress to PMC or severe disease.

While significant adverse outcomes are reported in children with CDI, especially those with hospital-onset disease, severe disease and severe complicated disease are less common in children than in adults. The Infectious Diseases Society of America guidelines define severe disease in adults as a white blood cell count of 15,000 cells/mm ³ or more or a serum creatinine of greater than 50% above baseline. These criteria have poor predictive value in children, likely reflecting underlying medical complexity of the child rather than severity of CDI. In general, severe disease is more likely to occur in neutropenic children with leukemia, infants with Hirschsprung disease, and children with inflammatory bowel disease. Severe complicated disease, defined as sepsis, hypotension or shock, ileus, toxic megacolon, perforation, need for intensive care, surgery for a CDI-related complication, or death, occurs in 2% to 5% of children with CDI. Other less common complications include pneumatosis intestinalis and rectal prolapse. If severe disease develops, anasarca can also occur because of protein-losing enteropathy.

Children with pseudomembranous colitis may not meet criteria for severe or severe complicated disease. PMC is generally associated with diarrhea and fever as well as abdominal distension, cramps, and systemic toxicity. Mucoid stools are the hallmark of PMC. PMC lesions almost always affect the colon, but involvement of the small intestine has been reported in adults. Focal necrosis and inflammation are initially found in PMC lesions, but these can progress to extensive involvement of the colon covered by confluent pseudomembranes. Although uncommon, fulminant colitis occurs and can lead to bowel perforation, peritonitis, and a high mortality rate.

Coinfection with other enteric viruses has also been described in children. Rotavirus and calicivirus have been reported in severe cases of CDI requiring intensive care. In one cohort, 24% of patients with CDI had either norovirus or Sapovirus coinfection. There were no differences in clinical severity or outcomes between those children with norovirus or Sapovirus and C. difficile coinfection compared to children with CDI alone.

Recurrence of disease occurs in up to 25% of pediatric patients after treatment. Risk factors for recurrent disease in children include receiving concomitant antibiotics during treatment of CDI or having received multiple classes of antibiotics prior to onset of CDI, presence of a tracheostomy tube, community-associated CDI, and malignancy. Chronic diarrhea secondary to C. difficile can occur without evidence of colitis and has been associated with failure to thrive.

The role of C. difficile in the pathogenesis of NEC among infants has been debated. Early studies reported no difference in colonization with C. difficile among infants with NEC and those without NEC. A prospective study of 50 preterm infants also demonstrated that C. difficile colonization was not associated with a higher incidence of NEC. In a retrospective cohort study of children with C. difficile –toxin-positive stool specimens, 50% of 22 patients in a neonatal intensive care unit had suspected or confirmed NEC, although causality was not proved.

In adults, C. difficile colitis has been reported to mimic acute peritonitis with fever, leukocytosis, and signs and symptoms of peritonitis on physical examination. Extraintestinal manifestations account for less than 0.2% of all C. difficile infections and can include bacteremia, sepsis, visceral abscesses, wound infections, pleural involvement, and reactive arthritis. Most cases of C. difficile bacteremia are polymicrobial with the isolation of other bowel flora. The mortality rate of bacteremia in adults is significant. In one recent study, mortality rate was 27% in those with bacteremia and in up to 39% of those reported in a review of the literature. Bacteremia is rare in children; one case has been reported in which the child survived.

Laboratory Studies

The diagnosis of CDI requires both the presence of diarrhea (or radiographic evidence of toxic megacolon) and the detection of C. difficile toxin or the toxin gene in a diarrheal stool specimen (or evidence of pseudomembranous colitis on colonoscopy or histopathology). Diarrhea is defined as stools that take the shape of their container, with three episodes that occur in a 24-hour period. Only diarrheal stool should be tested in most instances because testing of formed stool may detect colonization rather than infection. Occasionally a patient with CDI may have an ileus or evidence of toxic megacolon on imaging; a rectal swab may be the only available sample in this circumstance.

Historically culture using a selective medium for C. difficile has been the gold standard for diagnosing CDI. Because toxigenic and nontoxigenic strains can grow on culture, isolates must then be tested for the presence of toxin genes or gene products before the diagnosis of CDI can be confirmed. Additionally culture is time intensive, requires special equipment and trained personnel, and is not standardized, which may introduce bias. Stool culture is now generally reserved for epidemiologic investigations.

Cytotoxicity assays detect the presence of toxin B directly in stool by visualizing its cytopathic effects on cells in cell or tissue culture. These assays were reviewed and, as a reference method, best define true cases of CDI. However, while this test is highly specific, its sensitivity may be as low as 67%. Other limitations of cytotoxicity assays include the need for skilled personnel to interpret the findings in tissue culture, the subjective nature of test interpretation, and an approximate 72-hour turnaround time.

Enzyme immunoassays (EIAs) may detect toxin A or B or glutamate dehydrogenase (GDH), a product of C. difficile . EIAs offer the advantages of rapid turnaround time, ease of use, and lower cost. Those that detect toxin were introduced in the mid-1980s and became routinely used for diagnosis. However, their reported low sensitivity resulted in the historical practice that multiple specimens on a single patient be tested before the patient could be considered truly negative. This delayed diagnosis and increased the cost of testing. In children, both lower sensitivity and specificity of the EIA have been observed; positive tests alone should be interpreted with caution, and testing for toxins A/B should not be the first diagnostic assay for CDI because of this poor sensitivity.

EIAs for GDH offer improved sensitivity over EIAs for toxin A/B but lower specificity because GDH is present in both toxigenic and nontoxigenic strains of C. difficile , as well as in other Clostridium species. Its high negative predictive value makes it a useful screening tool. However, positive results should be confirmed in a two-step algorithm that detects toxin.

Nucleic acid amplification tests (NAATs) that detect toxin A ( tcdA ) or B ( tcdB ) genes or tcdC gene (a negative regulator of toxin A/B production), often by PCR, are FDA approved and now preferred by many laboratories. While more costly, these tests are rapid, specific, and more sensitive than EIAs and identify toxigenic C. difficile in a single step. NAATs detect bacteria that carry the toxin gene, regardless of toxin production, and therefore may detect toxigenic C . difficile in asymptomatic patients as well as in patients with diarrhea ultimately found to have another etiology. In children, NAATs may be positive in both hospitalized children with and without diarrhea, underscoring the ability of NAATs to detect C. difficile colonization. Additionally, NAATs may misdiagnosis community-acquired CDI in children with few risk factors for CDI and who may have other causes of diarrhea. These factors have led many to question the use of NAATs as a single diagnostic test for CDI.

Because no single diagnostic test is presently optimal for the diagnosis of CDI ( Table 45.1 ), many laboratories have adopted a multistep diagnostic algorithm for stool testing. In one approach, stools are initially screened with a rapid immunoassay that detects GDH and toxins A and B. Specimens that test both positive or both negative for GDH and toxin are immediately reported as positive or negative for toxigenic C. difficile . GDH-positive but toxin-negative specimens are subject to further testing by an NAAT. The sensitivity and specificity of this approach in adult populations have been reported to be from 91% to 100% and 92% to 98%, respectively. In a tertiary pediatric population, the sensitivity of the GDH-based algorithms was less but still superior to screening by toxin immunoassay (81% vs. 56%).

Approximately 90% of patients can be diagnosed with CDI on a single stool sample, and repeat stool testing does not increase the diagnostic yield significantly nor is it cost effective. Test of cure is also not recommended. Because asymptomatic colonization is common in children younger than 1 year, testing for C. difficile should not be performed routinely in this age group if other risk factors such as Hirschsprung disease are not present; other causes for diarrhea should be sought.

Leukocytosis, thrombocytosis, and hypoalbuminemia are commonly present in CDI. A leukocyte count greater than 15,000 cells/µL and a platelet count exceeding 400,000 cells/µL are more common in patients with CDI compared to those with diarrhea without CDI. Leukocytosis may be an early finding of CDI among hospitalized patients, even in the setting of mild or absent symptoms of colitis. The presence of fecal leukocytes is an insensitive method for CDI screening.

In seriously ill children in whom CDI is suspected but cannot be proved by standard laboratory testing, colonoscopy or sigmoidoscopy may be helpful. If pseudomembranes are found in a patient with consistent clinical symptoms, the diagnosis of CDI can be made because nearly all cases of PMC are caused by C. difficile. On examination, PMC has a characteristic appearance of yellowish-white raised plaques, usually 2 to 10 mm in diameter. Among infants with CDI, nonspecific colitis is more commonly found by endoscopy than are the characteristic pseudomembranes.

Other radiographic modalities may provide supplemental data to aid in the diagnosis of CDI. Plain abdominal radiographs in patients with PMC may demonstrate colonic ileus, small bowel ileus, ascites, and/or nodular haustral thickening. Abdominal computed tomography (CT) can be performed, but severity of colitis present on CT does not always correlate to clinical disease severity, and CT findings are less specific when compared with laboratory and clinical findings. Abnormalities on CT include nodular haustral thickening, colonic wall thickening, ascites, and pericolonic edema. Abdominal ultrasonography has also been used to aid in the diagnosis of CDI because bowel wall thickening can be visualized.