Chapter 50: Common Infections in the ICU

The evaluation of a new fever in an ICU patient should always begin with a thorough review of the patient's history and a focused physical examination, rather than automatic ordering of microbiologic studies.¹ After a differential diagnosis is formulated based on the history and physical examination, pertinent laboratory and radiologic studies should then be performed.

The differential diagnosis of a new fever in an ICU patient includes both infectious and noninfectious etiologies (Table 50-1).^{1, 2, 3} The most common infectious etiologies include hospital-acquired pneumonia (including ventilator-associated pneumonia), intravascular catheter-related bloodstream infection, catheter-associated urinary tract infection, surgical site infection, and Clostridium difficile infection. Infections unique to a neurologic/neurosurgical ICU include (1) superficial and deep surgical site infections that occur after neurosurgical procedures (eg, superficial wound infection after laminectomy or deep abscess after craniotomy) and (2) bacterial meningitis occurring after placement of internal and external ventricular catheters and lumbar catheters. Drug-related fever and intracranial hemorrhage are common noninfectious etiologies of fever in a neurologic/neurosurgical ICU.

Hospital-acquired pneumonia (HAP) is defined as the development of pneumonia 48 hours or longer after hospital admission. Ventilator-associated pneumonia (VAP) is a subtype of HAP that occurs more than 48 to 72 hours after endotracheal intubation.⁴ The diagnosis of HAP/VAP is initially suspected on the basis of clinical findings, including fever, peripheral blood leukocyte count, quality and quantity of tracheal secretions, changes in oxygenation, and chest radiography.⁴ The development of a new or progressive radiographic pulmonary infiltrate combined with two of three clinical criteria—(1) fever higher than 38°C (100.4°F), (2) purulent tracheal secretions, and (3) leukocytosis—has been shown to have a good balance of sensitivity and specificity for the diagnosis of HAP/VAP.⁵

Cultures of lower respiratory tract secretions help determine the microbial etiology of HAP/VAP and should be collected prior to starting or changing antibiotic therapy.⁶ Lower respiratory tract secretion samples may be cultured either (1) semiquantitatively (typically via an endotracheal aspirate sample), which is usually reported as heavy, moderate, light, or no growth; or (2) quantitatively (via bronchoalveolar lavage [BAL], protected specimen brush [PSB], or endotracheal aspirate), where a specific colony count is reported. For quantitative cultures, growth above a certain threshold may help distinguish infection from colonization. The following thresholds are commonly used to diagnose infection, assuming antibiotic therapy was not started or changed in the 24 to 72 hours before obtaining the culture: greater than or equal to 10⁴ colony-forming units (CFU)/mL for BAL samples, greater than or equal to 10³ CFU/mL for PSB samples, and greater than or equal to 10⁶ CFU/mL for endotracheal aspirate samples.^{4, 7, 8, 9} These thresholds may be lowered if recent antibiotic changes were made or if the pretest probability of HAP/VAP is very high.⁶

There is controversy regarding which culture method (semiquantitative or quantitative culture) is preferred. The benefits of obtaining a semiquantitative culture include rapid sampling (via an endotracheal aspirate) and lack of need for specialized microbiology techniques. Although semiquantitative cultures are more sensitive than quantitative cultures, they are less specific because tracheal secretions rather than deeper respiratory tract secretions are sampled and a specific colony count is not reported.⁴ This makes it more difficult to distinguish between colonization and infection. The benefit of obtaining quantitative cultures is that they are more specific, thus decreasing the rate of false-positive culture results.⁴ The disadvantages of quantitative cultures are that bronchoscopy is generally required, specialized microbiologic techniques are needed, and in some cases lower sensitivity may lead to a false-negative result.⁴

Clinical trial data comparing the two methods have not shown a difference in mortality or length of ICU or hospital stay.^{4, 10, 11, 12} The data are conflicting whether obtaining quantitative cultures influences the overall number of days of antibiotic use.^{10, 11, 12} We generally reserve obtaining quantitative cultures for patients with a low severity of illness who have a low to intermediate pretest probability of HAP/VAP. In these instances, assuming empiric antibiotic therapy is not given, the information gained from quantitative cultures is more likely to influence whether antibiotic therapy is ultimately administered or witheld.¹³ On the other hand, when the pretest probability of HAP/VAP is high or when a patient is clinically unstable, empiric antibiotic therapy is typically initiated, and a quantitative culture is less likely to influence discontinuation of antibiotic therapy.

Unless there is low suspicion for HAP/VAP, empiric antibiotic therapy should be initiated as soon as possible after a lower respiratory tract sample is obtained for culture.⁴ The rationale for prompt treatment is that a delay in appropriate antibiotic therapy has been associated with a higher attributable mortality rate from HAP/VAP.^{14, 15} In order to increase the probability of administering appropriate antibiotic therapy, the initial empiric antibiotic regimen should be based on the patient's risk for multidrug-resistant (MDR) pathogens (Figures 50-2 and 50-3).^{4, 16, 17, 18, 19, 20, 21, 22, 23} For patients at risk for MDR pathogens, initial combination therapy for gram-negative bacilli is recommended to maximize the chance of appropriate antibiotic therapy.

After initiating appropriate empiric antibiotic therapy for HAP/VAP, clinical improvement (based on peripheral blood leukocyte count, temperature, oxygenation, and quantity of tracheal secretions) typically begins to occur by 48 to 72 hours.^{24, 25} By this point in time, the results of lower respiratory tract culture data may be available, and the initial empiric antibiotic therapy regimen can usually be modified by either narrowing treatment or adding treatment for an organism not being covered. Narrowing antibiotic therapy (ie, de-escalation) is important to minimize the emergence of antibiotic resistance.²⁶

Modification of the empiric antibiotic regimen should be based on results of lower respiratory tract culture data. The interpretation of a positive lower respiratory tract culture is influenced by a number of factors, including whether semiquantitative or quantitative cultures are performed (eg, when quantitative cultures are obtained, growth above a predefined threshold may support infection over colonization) and the level of suspicion that that organism is causing HAP/VAP (eg, certain organisms are not considered pulmonary pathogens, such as coagulase-negative staphylococci and enterococci). A negative lower respiratory tract culture in an intubated patient may be clinically useful, as it has a strong negative predictive value for HAP/VAP in the absence of a recent change in antibiotic therapy in the preceding 72 hours.^{4, 6, 27} In addition, in an intubated patient, a lower respiratory tract culture that does not grow an MDR pathogen, in the absence of a recent change in antibiotic therapy in the preceding 72 hours, suggests that these pathogens are not causing HAP/VAP and antibiotic therapy can be narrowed accordingly.^{4, 25}

The rationale for using empiric combination antibiotic therapy for HAP/VAP while awaiting culture data is to maximize the chance of administering appropriate antibiotic therapy. Once antimicrobial susceptibility data return, therapy can usually be narrowed to a single agent. However, for cases where P aeruginosa is found to be the etiologic agent of HAP/VAP and antibiotic susceptibility data are known, a common question is whether combination antibiotic therapy with two antibiotics active against P aeruginosa should be given for the full treatment course (ie, definitive combination therapy).

Proposed theoretical reasons to support the use of definitive combination therapy for P aeruginosa pneumonia include the potential for (1) synergy, (2) prevention of emergence of resistance, and (3) change in clinical outcome. In vitro synergy has not been consistently established and improved outcome has not been demonstrated for patients receiving synergistic antibiotic combinations for P aeruginosa pneumonia as compared with combinations not synergistic in vitro.^{4, 28} Furthermore, prevention of emergence of resistance has also not been documented with the use of definitive combination therapy for P aeruginosa pneumonia.^{4, 29, 30} Lastly, definitive combination therapy for pseudomonal pneumonia has not been associated with a difference in mortality,^{30, 31, 32} clinical treatment failure,^{29, 31, 32} length of ICU and hospital stay,^{30, 31} or recurrence of VAP,³⁰ and it may lead to increased nephrotoxicity when aminoglycoside combination therapy is used.²⁹ An exception is that combination therapy may have a survival benefit in patients with pseudomonal pneumonia and bacteremia.²⁸

For select patients with severe pseudomonal HAP/VAP or pseudomonal pneumonia and bacteremia, we consider the use of combination antibiotic therapy for at least the first 5 to 7 days of treatment, although this decision should ultimately be individualized based on the specific clinical scenario.

There are insufficient data to suggest an advantage of linezolid over vancomycin for the treatment of HAP/VAP caused by MRSA. Two initial clinical trials did not show an advantage of linezolid over vancomycin for patients with HAP due to MRSA.^{33, 34} A retrospective analysis of data from these two clinical trials showed increased survival and clinical cure for patients treated with linezolid as compared with vancomycin³⁵; however, this study was criticized for methodologic issues (lack of randomization leading to unmeasured confounding, and lack of modification of a significant P value in the setting of multiple comparisons).³⁶ A subsequent clinical trial failed to show a difference in microbiologic cure, clinical cure, or survival between linezolid and vancomycin for the treatment of VAP due to MRSA; however, the sample size was small.³⁷ A larger clinical trial to further address this question is currently in progress.³⁷

An additional factor that may influence the treatment of MRSA HAP/VAP is the minimum inhibitory concentration (MIC) of MRSA to vancomycin. For patients with MRSA HAP/VAP where the vancomycin MIC is greater than or equal to 2 μg/mL, use of an alternate agent has been suggested (eg, linezolid) because of a concern for treatment failure if vancomycin is used; however, further data are needed to address this question.^{21, 38} In addition, for patients with HAP/VAP due to MRSA, suggested vancomycin trough levels are 15 to 20 μg/mL,^{4, 21} although clinical data documenting improved outcome with higher trough levels are lacking.^{38, 39}

In the past, the recommended duration of antibiotic therapy for HAP/VAP was as high as 14 to 21 days.⁴ However, prolonged antibiotic treatment (ie, more than 7 days of treatment) has been associated with newly acquired tracheal colonization with organisms such as P aeruginosa, Enterobacteriaceae, and S aureus.^{4, 25}

In a clinical trial comparing 8 days versus 15 days of antibiotic treatment for VAP, there was no difference in mortality rate or recurrent infections between the two groups.⁴⁰ Furthermore, among patients who developed recurrent VAP and received 15 days of antibiotic treatment, there was a higher rate of emergence of MDR pathogens as compared with patients who received 8 days of treatment. In a subgroup of patients who developed VAP due to nonfermenting gram-negative bacilli (ie, P aeruginosa and Acinetobacter species), there was a higher VAP recurrence rate in patients who received 8 days of antibiotic treatment as compared with 15 days of treatment.

Based on these data, a shorter duration (7-8 days) of antibiotic therapy for HAP/VAP is recommended for patients who clinically respond to treatment and who are not infected with nonfermenting gram-negative bacilli.⁴ For patients infected with nonfermenting gram-negative bacilli, 14 days of antibiotic treatment should be considered.

Infection control measures to prevent transmission of microorganisms from one patient to another via health care worker contact are critical; these include (1) strict compliance with hand hygiene, (2) strict compliance with isolation precautions for MDR pathogens, and (3) staff education.⁴

Mechanical ventilation is one of the most important risk factors for HAP, and the following measures are recommended to reduce the risk of HAP/VAP: (1) avoiding intubation if possible, (2) minimizing the duration of mechanical ventilation by using protocols that accelerate weaning and improve sedation use, (3) avoiding reintubation if possible, and (4) use of noninvasive positive pressure ventilation as an alternative to intubation when indicated.⁴

Preventing aspiration of gastric and oropharyngeal contents into the lower respiratory tract also reduces the risk of HAP/VAP. Maintaining intubated patients in the semirecumbent position (ie, 30- to 45-degree elevation of the head of the bed) and avoiding the supine position (ie, 0-degree elevation) reduces the risk of aspiration of gastric contents, especially during enteral feeding.⁴ The endotracheal tube cuff pressure should be maintained at greater than 20 cm H₂O to reduce the risk of aspiration of oropharyngeal bacteria around the cuff and into the lower respiratory tract.⁴ Continuous aspiration of subglottic secretions can be considered via a specially designed endotracheal tube, although these tubes are not readily available.⁴

Selective decontamination of the digestive tract with oral and/or intravenous antibiotics may reduce the risk of HAP/VAP; however, this strategy is not recommended because of the potential for selecting for antibiotic-resistant microorganisms.⁴

Bacteriuria (ie, significant growth of bacteria in urine) is a very common finding in patients with indwelling urinary catheters, and the majority of patients are asymptomatic.⁴¹ It is estimated that the incidence of bacteriuria is between 3% and 8% per day of urinary catheterization.^{41, 42, 43, 44, 45} The duration of catheterization is the most important risk factor for the development of catheter-associated bacteriuria, and almost all patients develop bacteriuria after 1 month of catheterization.^{41, 46}

The majority of patients with bacteriuria do not develop symptoms or signs of infection (Figure 50-4). The proportion of patients with catheter-associated bacteriuria that develop symptomatic infection is less than 25%.^{42, 43, 47, 48} The proportion of patients with catheter-associated bacteriuria that develop bacteremia is even lower, ranging between less than 1% and 4%.^{41, 47, 48, 49, 50} The reason for the low rates of symptomatic infection in patients with catheter-associated bacteriuria is unclear; however, it has been hypothesized that decompression of the urinary tract with a urinary catheter prevents obstruction, which is known to predispose to symptomatic urinary tract infection and bacteremia.⁴⁷

Guidelines for the diagnosis and treatment of CA-UTI in adults have recently been published, and the importance of distinguishing between CA-ASB and CA-UTI is emphasized, since they are managed differently (ie, CA-ASB does not require antibiotic treatment except in certain clinical scenarios, while CA-UTI should be treated).⁴¹

CA-ASB is defined by growth of greater than or equal to 100,000 cfu/mL of one or more bacterial species from a catheterized urine specimen (either from an indwelling urinary catheter, suprapubic catheter, or intermittent catheterization) in a patient who does not have symptoms or signs compatible with CA-UTI (Tables 50-2 and 50-3).^{41, 51} Obtaining a screening urine culture in an asymptomatic patient is only recommended in the setting of pregnancy and in patients undergoing urologic procedures where visible mucosal bleeding is anticipated.^{41, 51} Other than these clinical scenarios, obtaining a urine culture is not recommended unless there are signs or symptoms compatible with CA-UTI.

CA-UTI is defined as a patient with symptoms and signs compatible with CA-UTI, with growth of greater than or equal to 1000 cfu/mL of one or more bacterial species from a catheterized urine specimen (either from an indwelling urinary catheter, suprapubic catheter, or intermittent catheterization) or from a midstream voided urine specimen from a patient whose catheter was removed in the preceding 48 hours (see Tables 50-2 and 50-3).⁴¹ The diagnosis of CA-UTI can often be challenging, since patients with urinary catheters do not have typical symptoms of UTI (ie, dysuria, frequency, or urgency) that are seen in noncatheterized patients.^{41, 47} In addition, many ICU patients are intubated and/or sedated and soliciting symptoms is difficult or impossible.

As discussed above, the quantity of growth of an organism in urine culture does not distinguish symptomatic from asymptomatic infection in catheterized patients (see Tables 50-2 and 50-3). The quantitative count of greater than or equal to 100,000 CFU/mL has only been validated for use as a cutoff for the diagnosis of UTI in patients with symptoms of UTI who do not have a urinary catheter and provide a midstream voided sample.^{51, 52, 53} This quantitative count has been shown to be able to reliably distinguish true bacteriuria from contamination by periurethral flora when a midstream voided sample is obtained.^{51, 52, 53}

However, for patients with urinary catheters, there is no standard definition for significant bacteriuria.⁴¹ The specific quantitative count of bacterial growth required to differentiate contamination from true bladder bacteriuria is recommended to be lower than the traditional cutoff of greater than or equal to 100,000 CFU/mL for noncatheterized patients.⁴¹ There are several reasons why a lower quantitative culture cutoff (ie, ≥ 1000 CFU/mL) is likely to reflect true bladder bacteriuria in catheterized patients and is suggested for the diagnosis of CA-UTI.⁴¹ First, catheterized urine specimens are less likely to be contaminated by periurethral flora, as are voided urine specimens; therefore, lower colony counts are more likely to represent true bladder bacteriuria.⁴¹ Second, patients with symptomatic UTI have been shown to have colony counts considerably below 100,000 CFU/mL, and as low as 100 CFU/mL.^{52, 53} Last, almost all catheterized patients with colony counts as low as 1000 cfu/mL progress rapidly to concentrations greater than or equal to 100,000 CFU/mL, usually within 72 hours, if antimicrobial therapy is not given.⁵²

With respect to collection of urine specimens for patients with long-term catheters (ie, catheters in place for ≥ 30 days), it may be useful to obtain a urine culture after removing and placing a new catheter.⁴¹ The reason for this is that some organisms (eg, staphylococci, enterococci, P aeruginosa, and Proteus species) are more likely to adhere to the urinary catheter rather than to uroepithelial cells; this is in contrast to other organisms (eg, E coli and Klebsiella species) that are more likely to adhere to uroepithelial cells and persist after catheter exchange.⁵⁴ Therefore, urine specimens obtained from a freshly placed catheter in patients with long-term catheters may be a better reflection of organisms present in the bladder.

Pyuria (commonly defined as > 5 leukocytes per high-power field) is common in patients with CA-ASB and CA-UTI and cannot reliably distinguish between asymptomatic and symptomatic infection.⁵¹ This is in contrast to patients without a urinary catheter who provide a midstream voided specimen, where pyuria has been shown to have high sensitivity and specificity for the diagnosis of UTI.^{55, 56} For patients with a urinary catheter, pyuria has a low positive predictive value for CA-UTI (32% in one prospective study of 761 catheterized patients).⁵⁵ Thus, in a catheterized patient, presence of pyuria is not diagnostic of CA-UTI. The negative predictive value of pyuria for CA-UTI appears to be high (94% in that same study)⁵⁵; therefore, the absence of pyuria makes the diagnosis of CA-UTI less likely.⁴¹

Screening for or treating CA-ASB is not recommended, except in select clinical scenarios.^{41, 51} Studies have shown that catheter-associated bacteriuria may be transiently cleared with antimicrobial therapy; however, bacteriuria recurs in the majority of patients.^{45, 57, 58} Antibiotic therapy has not been shown to reduce the incidence of future episodes of bacteriuria or urinary tract infection, since bacteriuria rapidly recurs, especially if the catheter is still in place.^{41, 51, 57, 58, 59} Furthermore, in patients with CA-ASB who are treated with antibiotic therapy, recurrence of bacteriuria typically occurs with resistant microorganisms.^{41, 51, 57, 58}

The exceptions are that asymptomatic bacteriuria is recommended to be screened for and treated in (1) pregnant women, since treatment has been shown to decrease the risk of pyelonephritis, and (2) patients undergoing urologic procedures associated with mucosal bleeding, where the risk of postprocedure bacteremia is high.⁵¹ Other patient groups where treatment of ASB may be considered include renal transplant patients and patients with ureteral stents or nephrostomy tubes, although evidence is limited to support this recommendation. Another group where treatment of ASB may be considered is in catheterized women with asymptomatic bacteriuria that persists 48 hours after catheter removal. A clinical trial in this population showed that treatment with antibiotic therapy reduced the risk of subsequent development of a symptomatic UTI as compared with women who did not receive antibiotic therapy.⁶⁰

To help guide antibiotic therapy, obtaining a urine culture prior to initiating antibiotic therapy is recommended. Microorganisms that frequently cause CA-UTI include E coli, Klebsiella species, Serratia species, Citrobacter species, and Enterobacter species.⁶³ Of the nonfermenting gram-negative bacilli, P aeruginosa is also commonly isolated.⁶³ Enterococci and coagulase-negative staphylococci are also commonly isolated, but are less virulent and more commonly lead to asymptomatic bacteriuria.⁶³ For patients with long-term urinary catheters (ie, catheters in place for ≥ 30 days), CA-UTI is frequently polymicrobial, and in addition to the above listed organisms, the following organisms may be seen: Proteus species, Morganella species, and Providencia species.⁶³ If antibiotic coverage for CA-UTI is desired prior to availability of culture results, empiric coverage for the most likely pathogens is recommended. Prior culture results may affect empiric antibiotic choice (eg, recent isolation of fluoroquinolone-resistant E. coli should prompt empiric use of a different antibiotic other than a fluoroquinolone). The antibiotic treatment should subsequently be modified based on identification of the pathogen and antimicrobial susceptibility results.

Prior to initiating therapy for CA-UTI, it may be reasonable to change the catheter if it has been in place for 2 weeks or longer.⁴¹ A clinical trial in nursing home residents who had urinary catheters in place for at least 2 weeks showed that changing the urinary catheter prior to treatment led to a greater proportion of patients with clearance of bacteriuria at 72 hours and a lower recurrence rate of bacteriuria 28 days after treatment was completed.⁶⁴

The optimal duration of therapy for CA-UTI is not known. Recommended duration of therapy varies from 7 to 14 days.^{41, 63, 65} A shorter duration is desired to limit the emergence of antibiotic resistance. For patients with milder infection and those who respond to treatment promptly, a 7-day course of antibiotic therapy is recommended.^{41, 63} For patients with a delayed response to treatment, a 10- to 14-day course is recommended.⁴¹ For CA-UTI with bacteremia and/or pyelonephritis, a 14-day course of antibiotic therapy is suggested.

Catheter-associated (CA) candiduria can be divided into CA asymptomatic candiduria and CA symptomatic candiduria. CA candiduria is most commonly asymptomatic^{47, 65, 66},; however, it can rarely lead to complications such as bladder fungus ball or ascending infection.⁶⁵ The diagnostic approach is similar to CA-ASB and CA-UTI, in that a positive urine culture for Candida does not distinguish symptomatic from asymptomatic infection in catheterized patients. Thus, the diagnosis of CA symptomatic candiduria can be challenging, since typical symptoms of UTI (ie, dysuria, frequency, or urgency) are not seen and symptoms and signs compatible with CA-UTI (see Table 50-3) may be difficult to attribute to the candiduria.

Treatment of CA asymptomatic candiduria is not recommended, except in certain clinical scenarios.^{67, 68} In a clinical trial of 316 patients with candiduria (over 95% were asymptomatic, and about half had a urinary catheter), there was a higher eradication rate of candiduria in catheterized patients receiving 14 days of fluconazole (50% versus 29%); however, the rates of candiduria 2 weeks after therapy was completed were similar in both groups (61% versus 56%), pointing out the difficulty of eradicating candiduria in patients with an indwelling urinary catheter.⁶⁸ Treatment of CA asymptomatic candiduria is only recommended in certain high-risk patient groups including patients undergoing urologic procedures, neutropenic patients, and infants with low birth weight.⁶⁷ Other patient groups where treatment may be considered include renal transplant patients and patients with ureteral stents or nephrostomy tubes, although further data are needed to support this recommendation.

Treatment of CA symptomatic candiduria is recommended. The preferred agent for fluconazole-susceptible Candida species (eg, Candida albicans) is fluconazole (200 mg daily for 2 weeks for cystitis; 200-400 mg daily for 2 weeks for pyelonephritis; and 400 mg daily for 2 weeks for patients with disseminated candidiasis).⁶⁷ For fluconazole-resistant Candida species, infectious diseases consultation is recommended, where intravenous amphotericin B deoxycholate may be considered.⁶⁷ Echinocandins, such as caspofungin, have been used for cases of symptomatic UTI caused by fluconazole-resistant Candida species; however, clinical trial data are lacking.⁶⁹

The most important strategy for reducing the risk of CA-UTI is to minimize the inappropriate use of urinary catheters. Indwelling urinary catheters should only be placed for acceptable indications and they should be removed as soon as they are no longer necessary.⁴¹ Urinary catheters should be placed using aseptic technique and sterile equipment.⁴¹ A closed catheter drainage system should be used and the drainage bag and connecting tube should always be kept below the level of the bladder; disconnection of the catheter junction should be minimized.^{41, 45} Prophylaxis with systemic antimicrobials and catheter irrigation with antimicrobials are not recommended, and addition of antimicrobials or antiseptics to the drainage bag should be avoided.⁴¹ Daily meatal cleansing with povidone-iodine solution, silver sulfadiazine, or antibiotic creams is not recommended.⁴¹

There is currently no standard definition for severe C difficile–associated disease (CDAD). In one study of risk factors for complicated CDAD (defined as presence of toxic megacolon, perforation, requirement for colectomy, shock requiring vasopressor therapy, or death) in Quebec, age 65 years or older, an elevated creatinine level greater than 1.1 mg/dL, immunosuppression, and a peripheral blood leukocyte count greater than 20,000 cells/mm³ were predictive of complicated CDAD.⁷⁰ When the creatinine was above 2.3 mg/dL, the risk of complicated CDAD was even greater.⁷⁰

In a clinical trial comparing metronidazole versus vancomycin for CDAD, variables used to define severity of disease included age older than 60 years, temperature greater than 38.3°C (101°F), albumin less than 2.5 mg/dL, peripheral blood leukocyte count greater than 15,000 cells/mm³, endoscopic evidence of pseudomembranous colitis, and intensive care unit admission.⁷¹ In a subsequent clinical trial comparing metronidazole, vancomycin, and tolevamer for CDAD, severe CDAD was defined as one of the following: 10 or more bowel movements per day, peripheral blood leukocyte count greater than 20,000 cells/mm³, or severe abdominal pain due to CDAD.⁷²

Deciding whether a patient has severe CDAD should be based on clinical judgment; however, most patients with severe disease will have an elevated peripheral blood leukocyte count greater than 15,000 to 20,000 cells/mm³ (which is likely a marker of the severity of colonic inflammation) and/or severe diarrhea that may be complicated by acute renal failure (Table 50-4). Severe CDAD can progress to complicated CDAD, where shock, perforation, or toxic megacolon may be seen.

There are two recent clinical trials comparing oral metronidazole versus oral vancomycin for the treatment of CDAD.^{71, 72} Both trials showed no difference in clinical success or cure rate between metronidazole and vancomycin for mild or moderate CDAD; however, for patients with severe CDAD, vancomycin was superior to metronidazole. Based on these data, oral vancomycin has been recommended by many experts as the preferred treatment for patients with severe CDAD (see Table 50-4).^{62, 73} Treatment duration should be 10 to 14 days.⁶² Doses of oral vancomycin that have been used include 125 mg, 250 mg, or 500 mg orally every 6 hours. It appears that 125 mg every 6 hours may be as efficacious as 500 mg every 6 hours, although more data are needed.⁷⁴

For patients with severe infection associated with ileus or for patients with complicated CDAD, metronidazole 500 mg intravenously every 6 to 8 hours may be combined with oral vancomycin (if patient is able to take medications enterally).⁶² The rationale for this is that if there is delayed passage of oral vancomycin to the colon, intravenous metronidazole may have activity as it has been shown to have adequate fecal concentrations.^{62, 75} Use of intracolonic vancomycin via retention enema may be considered in select situations, although data are limited.⁷⁶ Although cases of intravenous immunoglobulin (IVIG) being used for severe CDAD have been reported,^{77, 78} a recent small retrospective cohort study with propensity matching did not show a benefit of IVIG for severe CDAD.⁷⁹ Other treatments, including probiotics, rifaximin, cholestyramine, and fecal bacteriotherapy, do not play a role in the treatment of severe CDAD, but may be beneficial in patients with recurrent CDAD.⁶² For patients with signs of complicated CDAD including perforation, shock, or toxic megacolon, antibiotic coverage for bowel flora, including enteric gram-negative bacilli, should be considered. Lastly, early surgical consultation is recommended for patients with severe CDAD, since early subtotal colectomy may be required if complicated CDAD develops.^{62, 80}

Oral vancomycin 250 mg orally every 6 hours and metronidazole 500 mg intravenously every 8 hours were initiated. Despite this, her peripheral blood leukocyte count increased to 44,800 cells/mm³, with 54% band forms, and she required vasopressor support. She was taken to the operating room for exploratory laparotomy, where ischemia of the colon was identified, necessitating subtotal colectomy. Important points to learn from this case include avoiding unnecessary antibiotic use, clinical features that suggest severe CDAD, and treatment considerations for severe CDAD.