Chapter 57. Gastrointestinal Infections

• In addition to immunologic mechanisms, physical (motility), chemical (gastric acidity), and microbiologic (normal colonizing flora) factors normally protect the gastrointestinal tract against infection.
• Esophagitis, most commonly caused by Candida albicans or herpes simplex virus, may be underrecognized among patients in the intensive care unit.
• Infection with Helicobacter pylori may play a role in the pathogenesis of gastric stress ulceration among critically ill patients.
• The epidemiology and microbiology of diarrheal illness is significantly different among patients in the critical care unit than is observed in the community setting. Most infectious diarrhea is hospital-acquired and is usually attributable to Clostridium difficile.
• A systematic approach to the critically ill patient with diarrhea includes consideration of pathogens that cause noninflammatory, inflammatory, and hemorrhagic diarrhea. Thorough history taking supplements laboratory data in the diagnosis of these patients.
• C. difficile infection is the single most common cause of gastrointestinal infection among patients in the intensive care unit. The spectrum of disease induced by C. difficile infection is broad. Timely diagnosis and treatment is critical both for the management of the infected patient and to prevent the spread of infection through the unit.

While rarely severe enough to warrant admission to the ICU, gastrointestinal infections account for substantial morbidity and mortality among critically ill patients. Because of severe comorbid disease, impaired immune defenses, and the invasive interventions to which they are subjected, patients in the ICU are especially susceptible to hospital-acquired GI infection. Nevertheless, despite the frequency with which these infections occur, the morbidity and mortality that they cause, and the costs they incur, GI infections can go undetected and untreated in the ICU. While trying to manage patients with deteriorating cardiac function, marginal ventilatory performance, and life-threatening metabolic abnormalities, clinicians in the ICU may fail to recognize the important early signs of GI infection.

Any discussion of GI infections among critically ill patients must begin with a consideration of the host defenses that normally protect the alimentary tract. As such, the first section of this chapter is devoted to a description of the unique nonimmunologic mechanisms normally active in the GI tract. Particular consideration is given to the means by which these defenses may be compromised in patients in the ICU. Following this introduction, the clinical manifestations of infection affecting each segment of the GI tract are discussed (Table 57-1). In addition to describing the microbiology associated with each syndrome, a rational diagnostic and therapeutic approach is offered, based on the most up-to-date experience reported in the medical literature. The chapter concludes with an expanded discussion of the unique clinical challenges presented by the patient in the ICU with Clostridium difficile infection.

Motility

GI motility, in addition to its central role in normal digestion, is one of the principal host defenses against infection. By continuously flushing the lumen of the GI tract, normal motility prevents the accumulation of infectious organisms and the virulent toxins associated with disease. When bacteria are permitted to collect and reproduce unchecked, such as in blind bowel loops rendered devoid of normal motility by surgical interventions, infection can ensue. Patients in the ICU are routinely exposed to interventions that can have the effect of impairing gut motility, rendering the patient vulnerable to infection. For example, while narcotics can provide appropriate and necessary analgesia for the critically ill patient, these agents have the side effect of inhibiting gut motility.

Gastric Acidity

Gastric acidity provides a unique chemical barrier to the establishment of upper GI colonization and infection. In the highly acidic environment of the stomach, few pathogens are able to survive, much less thrive. However, the gastric pH of patients in the ICU is often much higher, providing an environment that is more hospitable to bacteria. More importantly, ingested microbes can pass into the lower GI tract. Once again, pharmacologic interventions are primarily responsible for this disruption of normal protective physiology. The attenuation of gastric acidity is deliberate—an effort to lessen the likelihood of stress-induced gastritis and resultant GI hemorrhage. Medications such as histamine (H2)-receptor blockers and proton pump inhibitors are commonly employed for this practice in both medical and surgical ICUs.

Normal Colonizing Flora

While not intuitively obvious as a component of host defense against infection, the normal colonizing flora of the GI tract provides as much protection as any physical or chemical barrier. Together, the host and normal GI flora comprise a delicate and varied ecology into which the introduction of new and potentially virulent flora is not favored. The bacteria that populate the GI tract are varied, depending on the anatomic segment under consideration. The mouth normally contains a mixed population of gram-positive, gram-negative, and anaerobic bacteria. In the esophagus, the population is less diverse. As already noted, the acidic environment of the stomach is distinctly inhospitable to the establishment of bacterial colonization. However, one organism, discussed in detail later, has been found to be of profound clinical relevance. Because of its ability to survive in the stomach, Helicobacter pylori plays a critical role in the pathogenesis of peptic ulcer disease. In contrast to the case of the stomach, the lower GI tract plays host to substantial microbiologic diversity. An enormous range of gram-negative, gram-positive, and anaerobic flora populates the intestines, especially the colon. Specific constituents include enterococci and Bacteroides species, as well as members of the family Enterobacteriaceae.

Disturbance of the dynamic between host and bacterial colonizers, such as occurs after exposure to broad-spectrum antimicrobials, predisposes patients to GI infection, most notably colitis caused by Clostridium difficile. While this association is well recognized, the factors that govern this phenomenon are still not well understood. It is not known if the normal flora compete with infecting pathogens for nutrients or substrates, occupy limited mucosal binding sites, or somehow otherwise alter the microenvironment in a way that reduces the likelihood of colonization. Regardless of the actual mechanism, an interesting therapeutic corollary can be inferred from the relationship between the normal host and GI colonizers. Deliberate intestinal colonization with so-called probiotics such as Saccharomyces cerevisiae may offer a means by which to preclude the onset of nosocomial infection or to attenuate the effects of these infections once established.1

The esophagus may be easily overlooked as a site of infection in patients hospitalized in the ICU. These patients may be unable to verbalize or otherwise express to caregivers the subjective complaints that indicate the presence of infection. To make matters worse, mechanical instrumentation commonly employed in the ICU, including endotracheal, nasogastric, and orogastric tubes, may limit the clinician's ability to thoroughly examine the patient for signs of upper GI infection. Moreover, even when characteristic physical findings of infection are visualized, they may be incorrectly ascribed to mechanical irritation or inflammation associated with such tools. When the opportunity to diagnose upper GI infection is missed, directed therapy may be withheld and infection allowed to proceed unchecked.

Clinical Presentation

Nearly 20% of ICU patients who underwent upper endoscopy in one study were incidentally noted to have esophagitis.2 These patients typically experience dysphagia with or without odynophagia. The pain of esophagitis is described as retrosternal and is typically exacerbated by the recumbent position. In the alert, awake, and communicative patient, these hallmark complaints are easily called to the attention of caregivers. However, as was already noted, the intubated and sedated patient in the ICU may not be able to express these complaints. Fever is an unreliable clinical finding in the patient with esophagitis. Regardless of the causative organism, fewer than one third of all patients with esophagitis will experience an elevation in temperature.3

Microbiology

Among hospitalized patients, esophagitis is most often caused by Candida albicans. While C. albicans remains the yeast species most frequently associated with esophagitis, an increasing proportion of cases have been linked to non-albicansCandida species, including C. tropicalis, C. parapsilosis, C. krusei and C. glabrata.4 This changing epidemiology has been attributed to the increasingly common use of empiric and prophylactic therapy with triazole antifungal agents such as fluconazole, to which many non-albicansCandida species are resistant. This epidemiologic phenomenon needs to be incorporated into the approach to therapy for such patients.

Herpes simplex virus (HSV) is another frequent cause of esophagitis and is the most common serious viral infection of the upper GI tract among patients in the ICU. For the most part, HSV type 1 is more likely to cause esophagitis than is HSV 2, which is more typically associated with genital infections. Less frequently, other viruses, including cytomegalovirus (CMV) can cause esophageal ulceration. For patients with CMV disease, lesions may extend throughout the length of the GI tract.

Diagnosis

Thorough physical examination is not only essential to the diagnosis of esophagitis, but may offer preliminary clues as to the causative pathogen. Both yeast and viral pathogens infecting the esophagus can produce telltale lesions in the oral cavity, where they will be easily detected on routine physical examination. Although present in fewer than one third of all patients with HSV esophagitis, oral or labial herpetic ulcers should not be missed in the physical examination of the critically ill patient with unexplained fever.5 Similarly, an adherent white coating to the lateral aspects of the tongue, which when scraped away reveals patches of inflammation, should point to Candida albicans as the cause of a suspected case of esophagitis. Despite the utility of such findings, it is equally important to recognize that esophagitis most often occurs in the absence of such clues. Nevertheless, to miss these clinical findings, when present, is to miss a critical opportunity for early diagnosis and intervention.

Upper GI endoscopy can be a useful tool to confirm the pathologic and microbiologic diagnosis of esophagitis. Unfortunately, even when visualized through the endoscope, the lesions of Candida and HSV esophagitis may appear quite similar. Even the large shallow ulcers typical of CMV esophagitis may be mimicked by Candida or HSV. Because of this lack of discriminatory power, it is advisable to proceed to confirmatory biopsy. Brush specimens alone can be inadequate, especially as Candida species can be isolated as colonizers of the upper GI tract in up to 20% of normal individuals.6 Once obtained, tissue should be sent for viral and fungal culture as well as for histopathologic examination to confirm tissue invasion.

Therapy

Under most circumstances, directed therapy for esophagitis should be withheld until the causative organism has been identified. However, for critically ill patients with suspected esophagitis in whom endoscopy is not practical and microbiologic diagnosis is impossible, it is reasonable to treat empirically for C. albicans, based on the prevalence of this entity in this population. For esophagitis when C. albicans is known or suspected to be the cause, the most effective treatment is fluconazole given intravenously at a dose of 100 to 200 mg per day for 14 to 21 days. For those infected with Candida species resistant to fluconazole, or for patients with persistent infection despite first-line therapy, amphotericin B can be used as salvage therapy. The role of newer antifungal agents, including voriconazole and the echinocandins has not yet been fully clarified.

For HSV esophagitis, the antiviral agent with which there is the most clinical and published experience is acyclovir. Most patients in the ICU will require parenteral therapy—5 mg/kg intravenously every 8 hours for 7 to 14 days. If the virus is resistant to acyclovir, intravenous foscarnet can be substituted.

While the stomach is not typically considered an important site of infection among hospitalized patients, the potential association between Helicobacter pylori infection and gastric stress ulceration suggests another means by which GI pathogens may take a toll among critically ill patients. The etiologic relationship between the presence of Helicobacter pylori and ulcerative disease of the upper GI tract, and particularly the duodenum and stomach, has been established. Importantly, treatment of H. pylori infection with combinations of antimicrobial agents and inhibitors of gastric acidity will eradicate H. pylori infection, and in so doing promote the resolution of peptic ulcer disease.7 Antibiotics employed for this purpose are active against H. pylori and include macrolides and β-lactam agents. Acid suppressive agents given concurrently include sucralfate, H2-receptor blockers, and proton pump inhibitors.

Given these findings and a growing clinical experience with this strategy, it is not surprising that a link between H. pylori infection and the stress-induced gastritis that affects patients in the ICU has been proposed. Thus far, the results have been generally inconclusive. In a prospective, single-institution study of patients admitted to a medical/surgical ICU, half of all patients admitted were positive for H. pylori by urea breath test. After adjusting for other risk factors, H. pylori infection was the only clinical factor significantly associated with major mucosal injury. The organism was detected in 80% of such patients.8 However, the same investigators observed that the prevalence of H. pylori infection among ICU patients declined to 8% by the third day of admission, and to 0% by 1 week, owing to intercurrent antibiotic exposure.9

Diarrhea, the principal manifestation of intestinal infection among the critically ill, affects approximately one third of all patients admitted to the ICU.10 Patients in the ICU with diarrhea are especially vulnerable to the clinical sequelae of infection. For the critically ill patient, the dehydration that frequently accompanies severe diarrhea strains a circulatory capacity already limited by impaired cardiac contractility and septic hemodynamics. Such individuals are at high risk for further systemic deterioration, often culminating in multisystem organ failure. In addition to life-threatening volume loss, diarrhea in the critically ill patient can precipitate metabolic derangements including electrolyte imbalances and acidosis, further exacerbating the potential for cardiac rhythm irritability. Finally, uncontrolled diarrhea in a severely ill immobile patient can predispose to compromise of the protective barrier of the skin. As such, the patient is rendered vulnerable to further infectious complications. Considering these dire clinical consequences, the prompt detection, microbiologic diagnosis, and therapy of diarrhea must be a high priority for clinicians in the ICU.

The epidemiology of diarrheal illness among patients in the ICU is substantially different from that seen among less severely ill patients in the community. Such differences render most of the schemes used to classify diarrhea in other settings less applicable to the evaluation of the critically ill patient. Infectious diarrhea acquired in the outpatient setting is rarely sufficiently severe to warrant admission to the ICU. Therefore, infectious diarrhea among patients in the ICU is most often acquired in the hospital. As a result, the spectrum of clinical disease and associated pathogens for the patient in the ICU tends to be less diverse than that encountered in the community. In fact, the majority of all cases of infectious diarrhea diagnosed in the ICU can be attributed to a single pathogen, Clostridium difficile. For many of these patients, the differential diagnosis consists largely of noninfectious entities, such as diarrhea induced by hyperosmolar enteral feeding solutions.11 Norwalk-like viruses, one of the most common causes of endemic and epidemic diarrhea in the outpatient setting, are rarely identified as the cause of diarrhea in critically ill patients. Similarly, while outbreaks of food-borne gastroenteritis have been reported among hospitalized patients,12 in the absence of an identified cluster, the work-up of the ICU patient with diarrhea usually need not include consideration of these pathogens.

It is in recognition of the rarity with which the clinician in the ICU will encounter diarrhea that is not hospital-acquired that a review of these less familiar presentations is warranted. While the distinctions between these syndromes are somewhat arbitrary, and there is considerable overlap between them, it is imperative that the clinician caring for critically ill patients at least be able to recognize these syndromes. In the sections that follow, inflammatory, noninflammatory, and hemorrhagic diarrheas are considered separately. Each is discussed with respect to the most common clinical presentations and pathogens that could be expected in the ICU (Table 57-2). A general approach to the diagnosis and treatment of diarrhea among patients in the ICU follows. The chapter concludes with an in-depth discussion of diarrhea caused by C. difficile—an organism whose central role as a cause of diarrhea among patients in the ICU has already been discussed.

Noninflammatory Diarrhea

In general, the noninflammatory diarrheal syndromes are characterized by the production of large volumes of watery stool devoid of gross blood or inflammatory cells. By definition, stool examination for fecal leukocytes in such patients will be negative. The typical presentation and pathophysiology of noninflammatory diarrhea is best exemplified by infection with Vibrio cholerae. While this gram-negative bacillus is the most prevalent cause of dehydrating diarrhea throughout the world, it is rarely encountered as a pathogen causing serious disease in the developed world. That said, the metabolic sequelae of cholera are capable of generating systemic illness sufficiently severe as to require ICU admission in a returning traveler.

While the diarrhea experienced by the patient infected with V. cholerae is characteristic of the other noninflammatory diarrheal infections, the severity of disease is unique to cholera. Diarrhea is voluminous, and patients can lose more than 1 liter of fluid everyhour. Affected patients are at high risk for life-threatening dehydration. Vital signs will reveal tachycardia and hypotension. The metabolic abnormalities can precipitate severe acidosis. To compensate, the patient may become tachypneic. Skin evaluation in these individuals reveals decreased turgor. The mucous membranes, including conjunctivae, appear dry. In extreme cases, the patient's eyes will appear sunken, producing a characteristic facies. If fluids are not replaced promptly and in sufficient quantity, the infection will be fatal.

The diarrhea of cholera is secretory in nature. Having established itself in the lumen of the bowel, V. cholerae releases an extracellular protein that binds to the membrane of intestinal epithelial cells. The enterotoxin induces an increase in intracellular cyclic adenosine monophosphate (cAMP). The high concentration of cAMP induces an increase in chloride secretion and a decrease in sodium absorption, producing the massive fluid and electrolyte loss characteristic of cholera.13

While V. cholerae is the prototypical pathogen associated with noninflammatory diarrhea, an array of other organisms can produce the same syndrome. While the diarrhea induced by these other pathogens tends to be less severe than that of cholera, the greater frequency with which these organisms cause infection in the developed world makes them more likely to be encountered as a cause of diarrhea in this setting. Most important among these are the so-called enterotoxigenic strains of Escherichiacoli. These isolates produce an extracellular toxin, a component of which is similar to that produced by V. cholerae. The end result is comparable—profuse watery diarrhea that challenges patient and clinician to maintain adequate hydration.

Inflammatory Diarrhea

Spanning a broad spectrum of clinical severity, inflammatory diarrhea is strictly defined by the presence of fecal leukocytes when the stool from affected patients is examined microscopically. In terms of pathophysiology, these infections are characterized by compromise of the integrity of the intestinal epithelium. Depending on the causative organism, there may be varying degrees of bacterial invasion. As a result of this process, inflammatory cells, including both neutrophils and lymphocytes, are recruited to the affected area, where some are shed into the intestinal lumen.

In the most extreme cases, inflammatory diarrhea causes the clinical syndrome commonly referred to as dysentery. The patient with dysentery presents with semisolid or liquid bowel movements that are not as voluminous as those seen with noninflammatory diarrhea. In fact, some patients report very scant production of fecal matter. For them, bowel movements are characterized by small quantities of gross blood and mucus. Fever is often present, but is usually not exceedingly high. Patients with dysentery may experience severe cramping abdominal pain or tenesmus—pain with the passage of bowel movements. Because of the limited ability of critically ill patients to report such complaints, clinicians should be alert for the presence of the unique stool characteristics that identify the patient with dysentery and inflammatory diarrhea.

A number of pathogens have been described in association with the clinical manifestations of inflammatory diarrhea, but the classical description of the syndrome is associated with infection with Shigella species, particularly S. dysenteriae. As was true for cholera and other noninflammatory diarrheas, this association once again points to a shared pathophysiology. Pathogenic Shigella species elaborate an exotoxin (Shiga toxin) that acts by inhibiting protein synthesis, damaging the intestinal mucosa. Analogous shiga-like toxins have been detected in association with other bacterial species linked to inflammatory diarrhea, including both enteroinvasive and enterohemorrhagic strains of E. coli.

Like Shigella, the other bacterial species commonly identified in cases of inflammatory diarrhea are generally acquired through fecal oral transmission, often in the setting of food-borne outbreaks. In the United States, the most common such pathogens are members of the Salmonella species. Outbreaks of food-borne salmonellosis, while most commonly linked to undercooked poultry and dairy products, have even been reported in the context of a deliberate release associated with an episode of domestic bioterrorism.14 Other important pathogens identified in association with inflammatory diarrhea include E. coli and Campylobacter jejuni.

Hemorrhagic Diarrhea

Patients with hemorrhagic diarrhea, characterized by the presence of frank blood, are increasingly being seen in the setting of the ICU. In addition to the hemodynamic and metabolic complications that characterize other inflammatory diarrheas, patients with hemorrhagic diarrhea have been found to be predisposed to systemic illness that can warrant admission to critical care. Strains of E. coli belonging to serogroup O157:H7 have been epidemiologically linked to the development of the hemolytic uremic syndrome.15 While the mechanism linking infection and this syndrome is not yet well understood, the resultant deposition of fibrin thrombi in the renal glomeruli can induce sufficient anemia, thrombocytopenia, and azotemia as to be life threatening. The hemolytic uremic syndrome typically follows the onset of diarrhea by about 1 week. While many of these patients will require intensive supportive therapy, including blood transfusion and hemodialysis, for most the condition is reversible. Of particular concern to clinicians caring for these patients is the observation that antimicrobial treatment may contribute to the emergence of this syndrome.16

Evaluation of the Critically Ill Patient with Diarrhea

The foremost consideration in the evaluation of the critically ill patient with diarrhea is the prompt recognition of infection with C. difficile. In the setting of prior exposure to antimicrobials or antineoplastic therapy, diarrhea can be presumptively attributed to C. difficile infection until proven otherwise. Identification of these patients is critical not only for the initiation of directed therapy, but to ensure that adequate infection control procedures are followed to limit the spread of infection to other vulnerable patients in the ICU. A comprehensive approach to the diagnosis of C. difficile is provided at the end of this chapter.

Whether or not C. difficile infection can be excluded, the evaluation of diarrhea in the ICU must progress in a systematic fashion with respect to the microbiology of the most likely infecting organism (Table 57-3). The initial assessment of these patients should include an accurate history of both the course of diarrhea and the presence of any precipitating factors that might suggest a causative organism. The patient, or for the uncommunicative patient a family member or friend, should be queried about the timing of onset of diarrhea, the progression of symptoms, associated systemic complaints such as fever or chills, and the nature and quantity of bowel movements (with particular emphasis on the presence or absence of bloody stools). Additional essential data includes information about recent travel, unusual dietary intake, and the presence of similar symptoms among companions with whom the patient has shared a meal. Of course, a history of prior antibiotic therapy or cancer chemotherapy will be needed to distinguish individuals at risk for C. difficile colitis. By the end of this process, the clinician should be able to characterize the diarrhea as acute or chronic, community- or hospital-acquired, and severe or mild. The last finding is of particular importance in that supportive therapy to alleviate severe dehydration should not be withheld pending further laboratory and microbiologic evaluation.

Given the emphasis placed on distinguishing inflammatory from noninflammatory diarrhea, it will come as no surprise that the initial laboratory work-up of the critically ill patient with diarrhea should include an objective measure of inflammation. Testing for fecal leukocytes offers a reliable means by which to do so. The test is performed by mixing a drop of stool with methylene blue on a slide, followed by examination under a microscope. Testing for stool occult blood has been suggested as another useful tool to identify patients in the ICU with inflammatory diarrhea. Unfortunately, testing for occult blood, even in the presence of a new fever in a critically ill patient, may be of little use in discriminating inflammatory infectious diarrhea from other common, noninfectious causes of bloody bowel movements among such patients, including stress-induced gastric ulceration and ischemic colitis.

In the setting of noninflammatory diarrhea, epidemiologic data must be interpreted to assess the likelihood of infection with V. cholerae. For a traveler returning from an endemic area presenting with signs and symptoms consistent with cholera, direct stool examination under darkfield or phase contrast microscopy can reveal the linear motility characteristic of V. cholerae. The organism will also grow on nonselective bacteriologic media, but the preferred method is culture on thiosulfate-citrate-bile salts-sucrose agar. If infection with enterotoxigenic E. coli is suspected, an assay to detect toxin is available from reference laboratories.

Especially in a patient with community-acquired diarrhea, the identification of fecal leukocytes on direct observation should warrant a search for bacterial pathogens that cause inflammatory diarrhea, including E. coli, C. jejuni, Salmonella, and Shigella species.17 Stool culture can be particularly useful in distinguishing the bacterial pathogens that are commonly associated with these food-borne gastroenteritides. Selective media, such as MacConkey, desoxycholate, and salmonella-shigella agars, are employed in the microbiology lab to enhance the ability to detect these pathogens. It is useful to identify particular epidemiologic concerns to the microbiology lab so that the appropriate media can be employed.

The examination of stool for the presence of ova and parasites is of limited utility among patients in the ICU. First and foremost, the clinical syndromes caused by infection with these organisms tend not to be so serious as to require admission to the ICU. Moreover, the incidence of parasitic infections in ICU patients is so low as to make the positive predictive value of the stool ova and parasite examination vanishingly small. In this context, even a positive finding on stool ova and parasite exam is more likely to represent a false-positive than it is to represent actual infection. Many clinical laboratories have gone so far as to not accept stool ova and parasite specimens from patients who have been hospitalized for 48 to 72 hours.

The role of endoscopy in the evaluation of infectious diarrhea is limited. While biopsy may detect the presence of specific pathogens such as Entamoeba histolytica, the relatively low incidence of these infections in this population makes this the diagnostic procedure of last resort. Lower GI endoscopy will not help to discriminate or diagnose infection with common bacterial pathogens such as E. coli or C. jejuni. As discussed later, sigmoidoscopy and colonoscopy can be helpful in the detection and diagnosis of colitis caused by C. difficile infection.

Treatment

The foremost objective in the care of the patient with infectious diarrhea is to restore the patient to a normal fluid and electrolyte balance as rapidly as possible. While oral rehydration solutions, such as that recommended by the World Health Organization, have proven safe and effective in settings in which intravenous therapy is either impractical or unavailable, most patients with diarrhea in the ICU will require parenteral replenishment. Effective regimens include lactated Ringer solution and normal saline with electrolyte supplementation. The use of large volumes of 5% dextrose and water for these patients may precipitate dangerous hyponatremia. No matter the regimen selected, serum chemistry analyses should be performed frequently to ensure adequacy of electrolyte replacement.

In general, empirical antimicrobial therapy for infectious diarrhea not associated with C. difficile should be avoided. Indiscriminant antibiotic use for this indication exposes the patient to needless toxicity, may precipitate the emergence of resistant organisms as a cause of systemic infection, can worsen the course of some infection (as in the case of E. coli serotype O157:H7), and might predispose the patient to prolonged carriage with the offending pathogen.18 However, once a specific pathogen has been identified, therapy can be directed by documented susceptibility information—or at least trends among known or suspected pathogens. Pathogen-specific recommendations are listed in Table 57-4.

As was already noted, Clostridium difficile is the single most common cause of infectious diarrhea among all hospitalized patients, including those in the ICU. When patients receiving antibiotics during hospitalization are considered, the incidence of C. difficile colitis approaches 10%.19 It is estimated that fully one quarter of diarrheal episodes among hospitalized patients may be attributable to C. difficile infection.20 Hospital costs for patients who acquire C. difficile diarrhea are more than 50% higher than for those without infection, and an episode of C. difficile colitis prolongs the average length of stay for infected patients by nearly 4 days.21 Of greater concern, the incidence of C. difficile colitis appears to be increasing.22

Clinical Presentation

The clinical presentation of patients with C. difficile can be quite diverse. In addition to asymptomatic carriage, some patients may report only a minimal increase in the frequency or liquidity of bowel movements. Such individuals, while at low risk of complication from infection, if not identified and appropriately isolated from other patients, represent an important reservoir for the spread of C. difficile within the ICU. At the other extreme, patients may experience severe clinical deterioration as a result infection with C. difficile. The bowel movements of the patient with C. difficile infection may be watery and voluminous, such as is seen in noninflammatory diarrheas. Other patients with C. difficile infection experience a clinical syndrome more consistent with that of the inflammatory gastroenteritides, occasionally even reporting gross blood in the stool. Fever is present in about half of patients. Leukocytosis, sometimes profound, is often a reliable indicator of the onset of C. difficile colitis.

In extreme cases, patients infected with C. difficile present with the complication of toxic megacolon. Such patients may or may not experience diarrhea. Abdominal radiography in these cases reveals grossly dilated colonic loops without evidence of mechanical obstruction. The patients frequently demonstrate end-organ dysfunction as is typical for the sepsis syndrome. The diagnosis of toxic megacolon carries with it a high degree of mortality. Colonic perforation can result if the infection is not treated promptly.

Microbiology

C. difficile is an obligate anaerobic gram-positive bacillus whose virulence is attributed to the production of extracellular toxins. C. difficile can be identified as part of the normal flora in patients without overt diarrheal disease. In these individuals, toxin-negative strains of C. difficile are likely no more than harmless commensal organisms. In fact, there are data suggesting that progression to diarrhea occurs early after acquisition of C. difficile or not at all.23 To produce diarrhea and other manifestations of clinical disease, infecting strains of C. difficile must produce extracellular toxins. Together, toxin A (enterotoxin) and toxin B (cytotoxin) cause epithelial cell necrosis through the disruption of the actin cytoskeleton.24 Recently, strains of toxin A–negative, toxin B–positive C. difficile have been described in association with outbreaks of hospital-acquired disease.25

Diagnosis

The spectrum of clinical disease associated with C. difficile infection makes diagnosis solely on the basis of clinical observations impractical. That said, experienced clinicians often point to what they consider to be a typical picture of C. difficile diarrhea, including foul-smelling stool that is greasy and green in color. Unfortunately, such observations are of limited value. On the other hand, readily available historical data, when accurately obtained, can be far more informative in leading the clinician to an accurate diagnosis.

While once closely linked to exposure to clindamycin or other agents from the lincosamide class, C. difficile colitis has since been observed to occur after therapy with nearly every antimicrobial agent available to clinicians, including metronidazole.26 Because of their widespread use and broad spectrum of activity, there appears to be an especially close association between C. difficile diarrhea and prior therapy with the cephalosporin class of β-lactam agents.26 While there may be an association between the anaerobic activity of an antibiotic and its likelihood of precipitating C. difficile diarrhea, the specific mechanisms and implications of this relationship remain to be explored. The observed association between some antineoplastic chemotherapy agents and C. difficile infection similarly challenges our understanding of the pathophysiology of these infections.

The onset of diarrhea due to C. difficile infection most commonly occurs in close relation to antibiotic administration. Twenty-five percent of cases present while dosing is ongoing.27 However, both the published literature and clinical experience reveal episodes of C. difficile colitis that occur within 48 hours of the initiation of antibiotic therapy, and others that develop months after exposure. In addition to the pharmacologic agents already mentioned, a number of factors are thought to predispose patients to C. difficile colitis, including advanced age, increased severity of illness, and impaired GI motility.28 In one study, the odds ratio for C. difficile infection among patients with the most severe underlying disease was 17.6.19 Each of these factors is commonly encountered among patients in the ICU, rendering them especially susceptible to infection.

While a patient's clinical history might suggest C. difficile infection, accurate diagnosis depends on the prudent use of objective laboratory assays. Attempts to culture C. difficile from the stool have proven clinically unreliable and have little utility in the evaluation of the hospitalized patient with diarrhea. The results of fecal leukocyte testing in the setting of C. difficile infection are similarly unreliable as an aid to diagnosis. The accurate identification of C. difficile colitis depends on the detection of the characteristic toxins produced by the organism. Previously, the gold standard for diagnosis was the cytotoxic assay. With this study, diluted stool is added to monolayers of cultured cells. Observation of a cytopathic effect that is neutralized by antibody against the toxins is more than 95% sensitive and specific for the identification of toxigenic C. difficile.29 Because of the labor and expertise required to perform this assay, most labs now rely on immunoassays for the direct detection of C. difficile toxins. Newer immunoassays capable of detecting both toxin A and B are more sensitive than earlier versions that detected only the enterotoxin.

Endoscopy can be helpful in the diagnosis of C. difficile colitis, but its widespread application for this purpose is limited. The characteristic finding of pseudomembranes comprised of necrotic epithelial tissue and inflammatory cells all but confirms the diagnosis of C. difficile colitis in the appropriate clinical setting. However, this finding is not always encountered, even in severe episodes of C. difficile colitis. Moreover, to perform colonoscopy or even flexible sigmoidoscopy in the setting of C. difficile infection is to expose the patient to the risk of unnecessary trauma that could result in the accidental perforation of an already inflamed and friable GI tract.

Treatment

Whenever feasible, the first and most important step in the treatment of the patient with C. difficile colitis is the discontinuation of the pharmacologic agent that precipitated the infection. In most cases, this means that ongoing antimicrobial therapy for other indications should be withdrawn as soon as it is safe to do so. This is a particular challenge when dealing with the critically ill patient. Broad-spectrum antimicrobial therapy, even when given empirically, is essential to the survival of the septic patients commonly encountered in this setting. For them, the discontinuation of therapy is not advisable, and an agent with activity against C. difficile (discussed below) must instead be added to the antimicrobial regimen.

The most appropriate antimicrobial strategy to treat C. difficile colitis can be the source of some controversy and confusion for even experienced caregivers in the ICU. Both metronidazole and vancomycin, when administered orally, have been shown to be effective in the treatment of C. difficile colitis. While many clinicians persist in the belief that vancomycin is superior to metronidazole for the treatment of C. difficile, the available literature does not yet support this position. Several clinical trials have pointed to the equivalence of vancomycin and metronidazole therapy.30 However, it is worth noting that the percentage of isolates of C. difficile that are resistant in vitro to metronidazole may be increasing, but remains less than 10%.31 The clinical implications of this observation have yet to be sorted out.

Reduction in the overall use of vancomycin, and specifically in the setting of initial treatment of C. difficile colitis, has been recommended by the Centers for Disease Control and Prevention as a means by which to stem the emergence of resistant bacteria.32 Vancomycin-resistant enterococci are familiar to ICU physicians as a frequent cause of serious hospital-acquired infections. Strains of Staphylococcus aureus resistant to vancomycin have also been reported recently.33

For patients with C. difficile colitis who can tolerate oral therapy, treatment should be initiated with metronidazole, 500 mg every 8 hours. While many clinicians elect to treat for longer periods, the duration of metronidazole therapy necessary to treat C. difficile need not extend beyond 10 to 14 days. For patients who cannot tolerate metronidazole, vancomycin is an effective alternative, but should not be used as first-line therapy for the reasons previously outlined.

When oral therapy is not feasible, when given intravenously, metronidazole achieves adequate intraluminal concentrations to eradicate C. difficile colitis.34 The recommended dosage is 500 mg every 8 hours. Intravenous vancomycin should never be used to treat C. difficile infection. After parenteral vancomycin administration, drug levels within the intestinal lumen are not sufficient to guarantee eradication. The role of vancomycin in treating C. difficile infection in the critically ill patient instead focuses on intraluminal therapy. When administered via a rectal tube or in the form of an enema, vancomycin can serve as a useful adjunct to intravenous metronidazole for the severely ill patient. Such methods are particularly useful in the setting of toxic megacolon caused by C. difficile, when GI motility has all but halted, and reliable delivery of drug administered orally or by a feeding tube cannot be assumed. However, caution is advised when employing these techniques. The GI mucosa is exceedingly friable in the setting of C. difficile infection, particularly when toxic megacolon has evolved. Such patients are at high risk for GI perforation. When the patient is critically ill as a consequence of C. difficile colitis (rather than critically ill and also having C. difficile colitis), combined therapy (500–750) mg IV every 6–8 h plus vancomycin 500 mg enterally every 6 h) has been advised.35

Retreatment of patients with either recurrent C. difficile colitis or those who fail to respond to initial metronidazole therapy is controversial and bears special attention. Clinicians should avoid the practice of routine laboratory testing to detect C. difficile toxin at the end of a course of therapy in an effort to confirm clearance. Infected patients may continue to shed detectable toxin after therapy is complete. Some individuals may do so intermittently for years. In these cases, additional therapy is not warranted. However, when diarrhea continues despite initial therapy or recurs soon thereafter, additional treatment is advised. In these circumstances, a switch to vancomycin or a more prolonged course of metronidazole has been advocated. However, in most cases, simple retreatment with metronidazole appears to be just as efficacious.