Chapter 26 ICU Infections

26

CHAPTER

ICU Infections

commonly is considered on the basis of fever or leu- kocytosis. Fever afflicts at least half of all patients during their stay in the intensive care unit (ICU) and often is an important clue to the presence of infec- tion. The magnitude and pattern of fever, typically defined as a temperature exceeding 101°F (38°C) to 101.4°F (38.5°C), are often accorded undue sig- nificance; fever characteristics actually have little diagnostic value. It is essential to recognize that not all fever is due to infection. Several diseases as deadly as disseminated infection can induce fever. Prominent among them are heat stroke, neuroleptic malignant syndrome, and the endocrine disorders of hyperthyroidism, adrenal insufficiency, and pheo- chromocytoma. Noninfectious causes of febrile syn- dromes are discussed in detail in Chapter 28. For some patients, diagnosis of infection is dif- ficult because both localizing and generalized signs are unimpressive. Patients infected with human immunodeficiency virus (HIV) often have minimal tissue inflammation when infected. Elderly patients and patients with hypothyroidism and renal failure often have reduced fever responses compared to younger patients. Both neutropenia and immuno- suppressive drug therapy tend to reduce the local response to infection, making erythema, pain, swell- ing, and pus formation less likely. CATEGORIES AND CAUSES Three categories account for most infections seen in the ICU: primary bacterial infections that prompt admission (e.g., pneumonia, urinary tract infection [UTI], meningitis); nosocomial infections (e.g., catheter-related sepsis, nosocomial pneumonia); and infections of the immune-compromised host. The broad topic of infection cannot be addressed comprehensively in a single chapter of reason- able length; therefore, the discussion that follows focuses on the most common and serious infec- tions occurring in the ICU. Patient characteristics

• Key Points 1. Antibiotic choices should be reassessed on a daily basis, keeping in mind that effective regimens rarely reverse the effects of any infection in less than 48 to 72 hours. Antibiotic choices should be trimmed to the simplest effective combination as clinical response becomes evident and culture data become available. 2. The speed with which an infectious diagnosis is pur- sued and the invasiveness of the techniques used should parallel the severity of illness. Stable patients with functioning immune systems and good physi- ologic reserves require less aggressive diagnostic approaches, whereas critically ill, highly vulnerable, and fragile patients require rapid, definitive diagnosis. 3. When several equally effective alternatives exist to treat the same infection, choose the combination with the best side effect and cost profile. Oral ther- apy and parenteral dosing of a long-acting antibiotic given on an infrequent schedule are usually the best methods of reducing antibiotic costs. 4. Antibiotics should be selected based on culture results, if available, and on microscopic examina- tion of body fluid specimens if not. In the absence of diagnostic material, the clinical history and pre- sumed site of infection should be the primary deter- minants of antibiotic selection. 5. For the unstable, infected, or septic patient, broad- spectrum empiric antibiotic therapy should be insti- tuted after obtaining appropriate cultures. As a rule, for such patients, it is best initially to give too many rather than too few antibiotics—second chances to choose the appropriate therapy may not arise. OVERVIEW Infection may be suspected on the basis of localizing signs (e.g., swelling, erythema, wound discharge) or localizing symptoms (e.g., pain, dyspnea, cough) but

547

548

SECTION II • Medical and Surgical Crises

profoundly influence the likely site of infection and the organisms most commonly responsible. The selection of antimicrobials must take allergies and organ dysfunctions into account. Furthermore, individual hospitals have different spectra of bacte- ria causing a particular clinical syndrome, and even within a single hospital antibiotic susceptibility can vary widely among units. Therefore, practitioners must have a thorough knowledge of the patient being treated, the likely pathogens, and the anti- microbial susceptibility pattern of the hospital in which they practice. Antimicrobial options are constantly evolving. In recent years, for example, entirely new anti- biotic categories have been exploited and better tolerated formulations of long-established drugs have been commercially released. In the former category are drugs directed at organisms resistant to most standard agents, such as linezolid, dapto- mycin, and quinupristin–dalfopristin for meth- icillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus . Others include nontraditional antifungal agents, such as caspofungin for invasive Aspergillus and Candida infections. Nephrotoxicity of traditional amphoteri- cin desoxycholate has been attenuated by its lipid complex, cholesteryl sulfate complex, and lipo- somal variants. Voriconazole, a modified triazole, now offers a well-tolerated alternative to ampho- tericin that can be given orally as well as parenter- ally in the treatment of aspergillosis. Modifications within well-established antimicrobial categories have extended their spectra and/or limited their side effects. For example, fluoroquinolones (e.g., levofloxacin, gatifloxacin, and moxifloxacin) offer potent competition to traditional drugs in the treat- ment of typical and atypical pneumonia. Allergic sensitivity, renal insufficiency, hepatic dysfunction, bleeding tendency, or other vital organ dysfunctions may restrict options, but almost always, there exist more than one potentially effective antimicrobial combination. Suggestions that follow for antibiotic therapy are based on the most common pathogens and their usual susceptibility patterns while consid- ering the frequency and severity of side effects and ease and cost of administration. “Broad-spectrum” effectiveness is both a luxury and a liability, as the desire to cover a large number of potential patho- gens comes at a high price. Indiscriminate use of broad-spectrum antibiotics is rapidly producing multidrug-resistant bacteria. MRSA and penicillin- resistant Pneumococcus are now pandemic. New

and menacing pathogens such as multiresistant Enterococcus and Acinetobacter are recognized with increasing frequency. Unless patterns of antibiotic use change, it is likely such infections will grow in importance. Antibiotics are the only class of drugs that, when misused, can injure not only the patient being treated but nearby patients and those admit- ted to the ICU in the future. For example, the rou- tine use of vancomycin to treat diarrhea or to cover for Gram-positive pathogens may inadvertently “select out” organisms resistant to this useful drug. These highly resistant organisms then lurk in the ICU, ready to infect subsequent patients. The urinary tract, the most common site of ICU infection, accounts for almost 40% of all infec- tions. Although UTIs usually are inconsequential, the mortality rate for a bacteremic UTI approaches 30%. Risk factors for UTI include presence of a urinary catheter, female gender, diabetes, and advanced age. Colonization of urinary catheters occurs at a rate of about 5% to 10% per day, and most ICU-acquired UTIs occur in such colonized patients. Presumably, the colonized catheter per- mits retrograde passage of pathogenic bacteria into the bladder where they proliferate. Urinary catheter composition (Teflon rather than rubber) may reduce the infective hazard; however, there is no evidence that routine changing of the catheter or external application of antibiotic ointment decreases risk. Keys to preventing nosocomial UTI are sterile cath- eter insertion, early catheter removal, and mainte- nance of a closed drainage system. Diagnosis The diagnosis of UTI is all but certain when greater than 10 5 bacteria/mL are isolated from culture of freshly collected urine. This level of bacteriuria correlates well with the presence of more than one organism per high-power field of unspun urine. Unfortunately, fewer bacteria do not exclude the presence of infection. True infections have been documented with colony counts as low as 10 2 /mL. Escherichia coli, the most common bacterial isolate, occurs in about 30% of UTIs. Enterococcus and Pseudomonas are each recovered about 15% of the time in the ICU population. Klebsiella and Proteus species represent less common isolates. Contrary to previous teaching, in many cases, pure cultures Urinary Tract Infections Pathogenesis

549

CHAPTER 26 • ICU Infections

of Staphylococcus epidermidis represent infection, not contamination. In the absence of frank pyuria or quantitative culture data, it is difficult to differ- entiate colonization from infection in critically ill patients with indwelling catheters. In the tenuous patient with bacteriuria, it is probably best to err on the side of brief, organism-directed antibiotic ther- apy. For more resilient patients in the ICU, treat- ment of asymptomatic bacteriuria may be deferred safely. Candida species are commonly recovered from urine. The choice of therapy for isolated candiduria should be based on a clinical judgment regarding whether the patient is “colonized” or “infected.” Unfortunately, few reliable signs distinguish these conditions. A clinical picture of sepsis, with recovery of Candida from blood cultures as well as urine, sug- gests disseminated infection that should be treated with intravenous antifungals such as amphotericin B, fluconazole, or caspofungin. Conversely, finding small numbers of yeast in an asymptomatic patient with an indwelling urinary catheter rarely requires systemic treatment (except expedited removal of the catheter). The most difficult situation occurs when large numbers of yeast or clumps of hyphal forms are found in the urine of an asymptomatic patient or a patient with only modest fever. Although sug- gestive of invasive infection, such patients usually respond promptly to fluconazole (oral or intrave- nous), especially if the urinary catheter can be removed. Without evidence of infection elsewhere, parenteral amphotericin B probably should be reserved for immunocompromised patients or those with limited physiologic reserves. Bladder irrigation with amphotericin B is time consuming, expensive, of uncertain benefit, and confounding to accurate assessment of urine output. Fluconazole has all but eliminated bladder irrigation. Pyocystis, an invasive infection of the bladder wall, may complicate oliguria or anuria, especially in patients requiring hemodialysis. In this setting, reduced urine flow allows bacteria to proliferate to massive numbers within the bladder. For oliguric patients with obscure fever, the bladder should be catheterized and the urine sediment examined. In the appropriate setting, murky, turbid, culture-posi- tive urine establishes the diagnosis. Treatment The aggressiveness of therapy should parallel the clinical severity of the acute syndrome and the underlying illness. As a rule, presumed UTIs should

be treated aggressively because patients in the ICU often have impaired immunity (diabetes, HIV infection, immunosuppressive therapy), numer- ous indwelling devices (e.g., vascular catheters, prosthetic heart valves, pacemakers), and marginal physiologic reserves. The treatment of UTI includes the promotion of urine flow and drainage, removal of urinary catheters (when feasible), and antibiotic therapy. Not all patients with bacteriuria require prolonged courses of expensive, broad-spectrum, intravenous antibiotics. Otherwise, stable immu- nocompetent patients can be treated successfully using enteral antibiotics (e.g., ampicillin, trime- thoprim–sulfamethoxazole, quinolones). Oral ther- apy is not appropriate for septic patients or patients with obstructive uropathy or a focal complication (e.g., renal abscess). The need for two drug cov- erage of pseudomonal infections in nonimmuno- compromised patients is uncertain, but two drugs effective against Pseudomonas should be given to patients with abnormal immunity. (These include intravenous aminoglycosides and antipseudomonal penicillins, fluoroquinolones, or third-generation cephalosporins.) If Enterococcus or Staphylococcus is deemed likely (based on the urine Gram stain or culture), vancomycin probably should be first-line therapy. Rarely, when the infection is life threaten- ing and the possibility of vancomycin resistance is high, linezolid is an appropriate choice. Urine con- centrations of renally excreted antibiotics often are dramatically higher than those used in sensitivity testing; therefore, UTIs often can be cured using an antibiotic to which the bacteria are found to be “resistant” in vitro . Because drainage bags provide important pathogen reservoirs, manipulations of the closed drainage system should be undertaken only when necessary and conducted with sterile tech- nique. Furthermore, drainage bags should not be raised above the level of the bladder, as often occurs during patient transport. Doing so, even briefly, pro- duces urinary stasis and promotes retrograde flow of potentially highly contaminated urine.

Pneumonia Pathogenesis

Pneumonia-producing organisms usually enter the lower respiratory tract in aspirated upper air- way secretions. Hematogenous seeding is a much less common mechanism. Unless the inoculum is very large, glottic closure, cough, and mucociliary clearance normally provide an effective mechanical

550

SECTION II • Medical and Surgical Crises

defense (Table 26-1). Even when mechanical bar- riers fail, infection usually is averted by effective cellular (neutrophil and macrophage) and humoral immunity (antibody secretion). Unfortunately, both mechanical and immune defenses are jeopardized commonly in critically ill patients, even in those without a recognizable immune deficiency. Common conditions that allow proliferation of organisms leading to pneumonia are listed in Table 26-2. The organism causing pneumonia is highly dependent on where the infection was acquired and on indi- vidual patient characteristics. Diagnosis In the community, a patient with acute onset of fever, dyspnea, chest discomfort, and cough produc- tive of purulent sputum is likely to be suffering from bacterial pneumonia. Leukocytosis with a predomi- nance of neutrophils and distinct (new) infiltrate(s) on chest radiograph are strong supporting data. Lobar pneumonias often have detectable air bron- chograms in the zone of infiltration (Fig. 26-1). Sputum that demonstrates an overwhelming pro- Table 26-1.  Conditions Promoting Lung Inoculation Aspiration Depressed consciousness Swallowing disorders Nasogastric and tracheal tubes Hematogenous seeding (e.g., endocarditis)  Bacteremia  Fungemia Infected aerosols Contaminated ventilator tubing and humidifiers

portion of neutrophils, intracellular organisms, and the predominance of a single morphologic bacterial form further strengthens the case. Finally, the diag- nosis is established unequivocally by recovering the same organism from blood and sputum or pleural fluid cultures. The presentation is not always clas- sic, even with community-acquired pneumonia: fever may be mild, infiltrates may be subtle, and self-medication with antibiotics often obscures a bacteriologic diagnosis. In the ICU, making a correct clinical diagnosis of pneumonia can be difficult for several reasons. Fever and leukocytosis are nonspecific, and patients often have several potential nonpulmonary sites to explain these findings. In addition, the radiographic infiltrates that suggest pneumonia are mimicked by atelectasis, aspiration pneumonitis, pulmonary embolism and infarction, pleural effusion, and pul- monary edema (Fig. 26-2). Computed tomography Table 26-2.  Conditions Favorable to Proliferation of Microorganisms in Lung Impaired immunity Parenchymal necrosis Malnutrition Steroids/cytotoxic drugs Chronic alcohol abuse Diabetes Secretion retention Atelectasis Smoking Obstructive lung disease Neuromuscular weakness Acute respiratory distress syndrome Viral infections

FIGURE 26-1. Air bronchograms in setting of pneumonia. Air bronchograms are usually best detected by CT scanning.

551

CHAPTER 26 • ICU Infections

FIGURE 26-2. Lobar atelectasis without air bronchogram ( left ) suggests occlusion of a lobar bronchus (e.g., by mucus plug). Note the closer spacing of ipsilateral ribs ( blue arrow ) and mediastinal shift along with compenstory contralateral lung expansion ( red arrow ). The presence of an air bronchogram ( yellow arrow, right ) in conjunction with lobar volume loss indicates collapse with an open lobar passage and argues against the value of therapeutic bronchoscopy.

(CT) sharpens discrimination but may not settle the issue. An elevated level of procalcitonin, a readily measured biomarker, strongly suggests bacterial (as opposed to viral) pneumonia. Finally, widespread use of antibiotics inhibits the ability to recover a single pathogenic organism, and even when sputum cultures are positive, small numbers of colonizing bacteria are usually recovered. Causative Organisms The organisms causing pneumonia differ dramati- cally, depending on site of acquisition—community versus hospital. Common causes of community- acquired lobar pneumonia and their clinical associations are shown in Table 26-3. In the com- munity, streptococci, especially Pneumococcus , and Haemophilus influenzae, Mycoplasma, and viruses are the most common pathogens in otherwise “healthy” adults. Many underlying conditions vary this spectrum, however. In addition to the organ- isms already mentioned, patients with alcoholism, diabetes, or heart failure are predisposed to infec- tions with Klebsiella, Legionella, enteric Gram- negative rods, and Staphylococcus . When aspiration is likely (e.g., alcoholism, drug abuse, esophageal disorders), Bacteroides and other anaerobes are potential culprits. S. aureus frequently is recovered

from patients with “postinfluenza” pneumonia, and Pseudomonas species and Staphylococcus are com- mon etiologic organisms among patients with cystic fibrosis. In fact, staphylococcal disease including MRSA is now frequently encountered in patients with severe community-acquired pneumonia. Pneumonia acquired in chronic nursing care facili- ties or within 3 weeks of hospital discharge is likely to be caused by organisms usually recovered in hos- pital-acquired infections. For pneumonias that develop after the first few days in the ICU, a different, hospital-specific spec- trum predominates. Such infections are frequently polymicrobial. Gram-negative rods ( Pseudomonas aeruginosa, Klebsiella species, Enterobacter species, Acinetobacter species, E. coli, Proteus, and Serratia ) cause approximately 50% of all ICU pneumonias. Which Gram-negative organism predominates at a given hospital has a great deal to do with antibiotic pressure placed on its environment. Acinetobacter , for example, represents a significant threat in some hospitals, but by no means all. S. aureus causes another 10% to 20% of infections and its incidence appears to be higher or rising in many ICUs. The predominance of Gram-negative rods and Staphylococcus seen in the hospitalized patient is explained partially by the rapid rate at which the

552

SECTION II • Medical and Surgical Crises

oropharynx of the critically ill patient becomes colo- nized. Almost all critically ill patients are colonized with nonnative Gram-negative bacteria (many of which are antibiotic resistant) by the third hospital day. All too often, a specific pathogen cannot be iden- tified, despite good samplingmethods and symptoms compatible with acute pneumonia. Mixed polymi- crobial aerobic/anaerobic infections, Mycoplasma, Chlamydia, Legionella, and viral agents become more likely candidates under these conditions. Although fungal pneumonia ( Candida/Torulopsis species, Aspergillus, or Mucor ) must be considered in the neutropenic (<500 neutrophils/mm 3 ), dia- betic, or severely debilitated patient, it occurs only rarely in immunocompetent ones. When fungal lung involvement occurs in the immunocompetent patient, it usually is the result of hematogenous seeding with Candida in a predisposed host. For some patients with chronic destructive lung dis- eases (e.g., chronic obstructive pulmonary disease [COPD] or healed cavitary tuberculosis), Aspergillus can produce a primary invasive pneumonia. Patients infected with HIV present a unique set of problems. When the CD4 T-cell counts are normal, patients infected with HIV are suscep- tible to the same organisms as any other adult in the risk categories outlined in Table 26-3. As the

CD4 count declines (especially as it falls below 200 cells/mm 3 ), the spectrum of infecting organ- isms broadens. Although routine bacterial pathogens still predominate, Pneumocystis carinii (jiroveci), Mycobacterium tuberculosis, atypical mycobacteria, and fungal infections become more likely. There is a rough correlation between the CD4 count and the infecting organism, but the linkage is not suf- ficiently strong to forgo detailed evaluation and broad coverage. Potential pathogens also are influ- enced by the prior use of prophylactic therapy. Oral trimethoprim–sulfamethoxazole prophylaxis, for example, has dramatically reduced the incidence of Pneumocystis and Toxoplasmosis infections. Regardless of the patient substrate, the choice of initial therapy for a bacterial pneumonia is always accompanied by some uncertainty, even in the pres- ence of a Gram stain “typical” of a specific organism. Historical features can help immensely in sorting through the diagnostic possibilities. For example, the sudden onset of chills, pleurisy, rigors, and high tem- perature are characteristic features for community- acquired Pneumococcus in a young adult. On the other hand, such findings may be inconspicuous in an older person, in whom confusion or stupor often predominate. A history of seizures, drug abuse, alco- holism, or swallowing disorder focuses attention on

Table 26-3.  Clinical Associations in Community-Acquired Pneumonia Patient Characteristics Likely Organisms Healthy adult

S. pneumoniae, Mycoplasma, viruses (e.g., influenza), Chlamydia, H. influenzae

S. pneumoniae, Bacteroides, oral anaerobes

Predisposed to aspiration  Stroke  Esophageal disease  Seizures  Alcohol and drug abuse  Recent dental work

All organisms listed for healthy adult plus Klebsiella species, enteric Gram negatives, Legionella , S. aureus, Branhamella species.

Chronically ill  Diabetes  COPD  Heart failure  Low-dose steroids

Postinfluenza Cystic fibrosis

S. pneumoniae, S. aureus, H. influenzae S. aureus, Pseudomonas species.

AIDS or HIV with CD4 < 200

P. carinii, S. pneumoniae, H. influenzae, M. tuberculosis , fungal infection (geographic predilection) All organisms listed for chronically ill plus Aspergillus, Mucor , and Candida

Neutropenia

553

CHAPTER 26 • ICU Infections

Diagnostic Techniques Although the history provides clues to the etiologic organism, laboratory studies are the cornerstone of the workup. Leukopenia often results from over- whelming infections, particularly those that are due to Staphylococcus, Pneumococcus, or Gram-negative organisms. A differential count that is not signifi- cantly left shifted suggests the possibility of virus, Mycoplasma, or Legionella . Cultures of blood and pleural fluid (when present) must be obtained, and if positive, they are the most convincing evidence of a causative organism. Unfortunately, such speci- mens are usually nondiagnostic, even in seriously ill patients, and sampling of pulmonary secretions becomes the primary diagnostic modality. The aggressiveness of the diagnostic evaluation should parallel the severity of the illness. In an oth- erwise healthy young person with a lobar pneumonia and good oxygenation, empiric therapy or treatment based on Gram stain alone is acceptable. For the septic, profoundly hypoxemic, or immunocompro- mised patient, however, a more systematic evalu- ation is often prudent. When performed correctly, stain and culture of pulmonary secretions or alveo- lar lavage fluid remain the most likely techniques to yield a diagnosis. For patients with severe com- munity-acquired pneumonia, high-quality sputum often is obtained immediately after endotracheal intubation because forceful coughing and suction- ing at this time often yield copious lung secre- tions not yet contaminated by ICU colonization. Expectorated sputum is appropriate for analysis and culture only if there is a high ratio of inflammatory to epithelial cells. Apart from the Gram stain, direct immunofluorescent antibody staining for Legionella and tuberculosis is another useful method for pro- cessing the expectorated sample that can yield an immediate, but presumptive, diagnosis. Inhalation of a hypertonic aerosol, particularly if given via an ultrasonic nebulizer, can stimulate a productive cough in patients otherwise unable to expectorate. When adequate sputum is not expectorated, naso- tracheal suctioning can be helpful. Transtracheal aspiration has been all but abandoned with wide availability of fiberoptic bronchoscopy. Urinary anti- gen testing is highly significant for both pneumo- coccus and legionella species. Properly performed on well-selected patients, fiberoptic bronchoscopy is a valuable technique for evaluation of pneumonic infection. In general,

aspiration. Recent travel history, occupational or rec- reational exposure, and concurrent family illnesses can help diagnose an unusual organism (Table 26-4). In the absence of intrinsic cardiac conduction abnormality or intense β -blockade, a pulse rate that fails to rise in proportion to fever (pulse–tempera- ture dissociation) suggests an “intracellular patho- gen” such as Legionella, Rickettsia , Mycoplasma, Q fever, psittacosis, virus, or tularemia. Mycoplasma often has accompanying pharyngitis, myringitis, or conjunctivitis. Contrary to popular teaching, extra- pulmonary symptoms (diarrhea, central nervous sys- tem disease) are no more common in Legionnaires disease than in other bacterial pneumonia. Unlike community-acquired pneumonia, noso- comial pneumonia offers few historical clues to diagnosis. Occasionally, however, a skin rash, gin- gival disease, or purulent sinus drainage helps nar- row the possibilities. Numerous classic radiographic features have been described, including lobar consolidation without air bronchograms (central obstruction), bulging fissures ( Klebsiella ), infiltrate with ipsilateral hilar adenopathy (histoplasmosis, tularemia, tuberculosis), widespread cavitation ( Staphylococcus, Aspergillus ), and sequential pro- gression to multilobar involvement ( Legionella ). These findings are not sufficiently consistent, how- ever, to be of real value in confirming the diagnosis. Table 26-4.  Clues to Uncommon Causes of Pneumonia Diagnosis Historical Clue Histoplasmosis Excavation, bird exposure, Ohio Valley and midwestern USA Coccidiomycosis Travel to southwestern United States, Central California Tularemia Tick bite or exposure to skinned animals Brucellosis Slaughterhouse work Psittacosis Exposure to pet birds Q fever Sheep contact Varicella Family exposure Measles Family exposure Respiratory syncytial virus Family exposure Blastomycosis Hunting, deep woods exposure Ehrlichiosis Tick bite

554

SECTION II • Medical and Surgical Crises

bronchoscopic procedures should be reserved for those who are seriously ill, immunocompromised, or unresponsive to conventional therapy. The safety and ease of bronchoscopy are facilitated by the presence of an endotracheal tube. When a decision is made to perform bronchoscopy, bronchoalveo- lar lavage and protected brush sampling from the involved region are both reasonable alternatives. Of these, lavage methodology is perhaps more popular, as it is easier, and the protected brush seldom con- flicts with or adds to its accuracy. If the lavage spec- imen yields a predominance of polymorphonuclear leukocytes and more than 10 3 to 10 4 organisms/mL are isolated, infection with the recovered organism is likely. Fewer bacteria suggest an active infec- tion but also may be seen with a partially treated bacterial pneumonia. Blind suctioning, conducted with or without lavage through a wedged catheter (“mini-BAL”), can be performed proficiently by respiratory therapists or trained nurses. In the set- ting of diffuse pneumonia, this is a low-cost and generally effective sampling option. Only bron- choscopy, however, offers the directed sampling so often necessary. Performing transbronchial biopsies to obtain tissue for histologic examination and/or culture is more hazardous in mechanically ventilated patients. Moreover, empirically chosen antibiotics are usu- ally effective in addressing potential pathogens in patients who are immunocompetent. Although not often performed because of the risk of pneumotho- rax, transbronchial biopsy may be attempted when tissue recovery is essential, oxygenation can be well maintained, and coagulopathy is not present. In this setting, the only diagnostic alternatives are open or thoracoscopic lung biopsy. (The latter may not be feasible because of altered anatomy, high ventilation requirements, or refractory hypoxemia.) The risk of developing a pneumothorax while supported by the mechanical ventilator must be balanced against the potential yield and the clinician’s ability to promptly recognize and evacuate the air leak. (Note that the incidence of pneumothorax approaches 100% after open lung biopsy.) In addition, there are situations in which a diagnosis can be made only by tissue biopsy. Open lung biopsy is rarely necessary for patients with intact host defenses, and the value in even compromised hosts is arguable. Transthoracic needle aspiration often yields an adequate speci- men but exposes the patient to attendant risks of pneumothorax and bleeding.

The optimal diagnostic evaluation of a pneu- monic process for patients infected with HIV continues to evolve. For patients with normal or minimally reduced CD4 counts, mild to moderate illness, and a history and examination compatible with acute bacterial pneumonia, empiric antibacte- rial therapy after obtaining cultures is reasonable. For patients with reduced CD4 counts, progres- sive dyspnea, nonproductive cough, elevated lactic dehydrogenase (LDH), and a radiograph with an interstitial or ground-glass pattern, empiric therapy for Pneumocystis and close observation may be rea- sonable. (This is especially true in the absence of prior Pneumocystis prophylaxis.) For patients with low CD4 counts, severe hypoxemia, uncharacteris- tic chest X-ray infiltrates, or an unusual exposure history, early bronchoscopy is the most prudent option. When bronchoscopy is performed, bron- choalveolar lavage alone often is not sufficient; fun- gal infections, tuberculosis, and even Pneumocystis are missed at an unacceptable rate without trans- bronchial biopsy. Because of the wide variety of potential radiographic presentations of tuberculosis, it probably is wise for all patients with HIV symp- toms and acutely abnormal chest radiographs to be placed in respiratory isolation until a diagnosis of tuberculosis can be excluded reasonably. Treatment Nutritional, fluid, electrolyte, and oxygen support of the patient with bacterial pneumonia are not controversial and are applied universally. The initial choice of antibiotic(s) must be guided not only by the nature of the suspected organism but also by the severity of the illness and underlying patient factors. Thus, although treatment should be directed as spe- cifically as possible for patients who are only mod- erately ill, the initial therapy of a fragile patient with serious illness should include broad-spectrum cover- age. Pruning of the antibiotic regimen occurs when the specific causative organism and its likely sensitiv- ity are confirmed. It makes sense to give the chosen antibiotics as quickly as is feasible. There is little mar- gin for error in critically ill patients with pneumonia; however, one can never treat all potential pathogens. Holes in coverage always exist, and there is almost always more than one acceptable antibiotic combina- tion. The selection of antibiotic therapy represents a calculated bet against the most likely organisms. Recognizing the imprecision of the following descriptions, otherwise healthy patients with com-

555

CHAPTER 26 • ICU Infections

munity-acquired pneumonia caused by an unknown organism who exhibit little systemic toxicity can be treated initially with either ampicillin or a macro- lide antibiotic, such as azithromycin or clarithromy- cin. Macrolides, fluoroquinolones, and doxycycline are good options when atypical organisms are sus- pected. If the same patient appears toxic, reason- able initial treatments include ceftriaxone with or without azithromycin, levofloxacin, or moxifloxacin with ceftriaxone or an extended-spectrum penicillin. The following caveats apply: if postinfluenza pneu- monia ( Staphylococcus ) is suspected or if the patient is from a geographic region with a high prevalence of penicillin-resistant pneumococci, the addition or substitution of vancomycin should be considered. For patients with a high likelihood of aspiration, clindamycin alone and amoxicillin–clavulanate with metronidazole represent good initial choices. Community-acquired pneumonia in a patient with HIV was discussed earlier. Because a second chance to institute the correct therapy cannot be guaranteed, broad empiric cover- age is necessary for the toxic patient with nosocomial pneumonia. Recognizing that many toxic-appearing patients will not have pneumonia documented, nonetheless, coverage in this situation must include enteric Gram-negative rods (including multiresis- tant organisms), Streptococcus (including penicillin- resistant organisms), and Staphylococcus (including MRSA). Important clues to etiology can be gleaned from knowledge of the patient’s recent antibi- otic treatment, the resident flora of the ICU, the patient’s underlying illnesses, environmental expo- sures, and available culture data. Yet, in the majority of instances, therapy must be initiated empirically. Regardless of the appearance of the Gram stain, ini- tial therapy for critically ill patients should include a coverage for multiresistant Gram-negative bacilli, such as an extended-spectrum penicillin plus an ami- noglycoside or appropriate fluoroquinolone (e.g., cip- rofloxacin) or a third-generation cephalosporin (e.g., ceftazidime) plus an aminoglycoside or fluoroqui- nolone. For patients predisposed to staphylococcal infection (e.g., recent influenza, neutropenia, insti- tutional prevalence, or a suggestive sputum Gram stain), vancomycin represents first-line coverage. A fluoroquinolone, macrolide, or doxycycline should be added if there is an “atypical” clinical or radiographic presentation or if fever persists despite usual therapy. Highly resistant bacteria can be transferred between patients in the ICU, necessitating measures

to decrease cross-contamination. Careful handwash- ing or use of a bactericidal lotion between patient contacts dramatically decreases the risk of nosoco- mial infection. Use of gloves does not diminish the need for handwashing, and it is essential that gloves be changed between patient contacts. Whenever suctioning intubated patients, gloves should be worn on both hands to prevent staff acquisition and trans- fer of pathogens, including herpes viruses. One pneumonic infection that deserves spe- cial discussion is pulmonary tuberculosis. Although patients may be admitted to the ICU with signs and symptoms typical of pulmonary tuberculosis (cavi- tary apical infiltrates, cachexia, fever), the presenta- tion is often subtle. Tuberculosis in the ICU can take on almost any clinical or radiographic presenta- tion. Cavitary lung disease is only marginally more common than other frequently encountered vari- ants: punctate interstitial infiltrates (“miliary pat- tern”), lobar pneumonia, “empyema,” lung nodule, or diffuse bilateral infiltrates compatible with acute respiratory distress syndrome (ARDS). When the suspicion of tuberculosis is high, respiratory isola- tion should be instituted as quickly as possible and maintained until firm evidence suggests that the likelihood of contagion is low. (This is accomplished simply by examining two or more good-quality spu- tum smears for acid-fast organisms.) The implica- tions of missing a case of tuberculosis are enormous: potential death or disability of the infected patient and transmission of infection to the staff and other nearby immunocompromised patients. Viral Pneumonia Certain forms of viral pneumonia occur with distressing frequency in severely immunocom- promised patients (e.g., CMV in transplant recipi- ents). Although rhinitis, sinusitis, laryngitis, and other familiar manifestations of the “common cold” afflict most persons one or more times per year, viral disease rarely extends to the alveolar level in immunocompetent adults. Yet, certain classes of organism—notably adenovirus, influenza, varicella zoster, and in the recent past the coronaviruses responsible for severe acute respiratory syndrome (SARS) and middle east respiratory syndrome (MERS)—can cause devastating illness in exposed individuals who are vulnerable. These diseases gen- erally present with a diffuse bronchopneumonia or ARDS. In addition to general supportive measures applied to patients in respiratory failure, isolation

556

SECTION II • Medical and Surgical Crises

is required and viral antimicrobials should be con- sidered. For example, acyclovir (varicella), riman- tadine (influenza), and ribavirin (adenovirus) each offer modest benefit when used in timely fashion in well-selected cases. Perhaps the most important aspect in the management of these pneumonias is to take appropriate measures to prevent the spread of these contagious diseases to health care workers and to the patients they treat. The importance of such precautions was dramatically emphasized in the high incidence of illness among physicians dur- ing the SARS outbreak of 2003. When an appro- priate vaccine is available for a contagious disease (e.g., influenza), immunization of exposed individu- als is prudent. Empyema and Parapneumonic Effusions Definition Small amounts of pleural fluid routinely accu- mulate adjacent to pneumonias, and such collec- tions are termed “parapneumonic effusions.” Most parapneumonic effusions are intermediate or even transudative in nature (protein <3.5 g/dL or 50% of the serum level; LDH < 200 U/dL or 60% of the serum level), freely flowing, and self-limited. The term “complicated parapneumonic effusion” has been applied to effusion with loculations, the characteristics of which fall somewhere between

an uncomplicated, self-resolving parapneumonic effusion and an empyema. Usually exudative by protein and LDH criteria, leukocyte counts usually are less than 20,000/mm 3 , and glucose levels fall between the serum value and 20 mg/dL. The pH of such effusions is commonly regarded as a discrimi- nator of the need for drainage, but its discriminat- ing value is often limited. Although it is true that the lower the pH, the more likely a pleural effusion is to have characteristics of an empyema (see fol- lowing), the pH alone neither makes the diagnosis of an empyema nor dictates a particular course of action. Effusions with a pH less than 7.0 (with a normal arterial pH) are likely to be empyemas and are likely to require drainage by tube thoracostomy, but such associations are not always valid. An acidic, thin, clear, or slightly cloudy sterile fluid does not necessarily require tube thoracostomy, whereas a thick, viscous, protein- and leukocyte-rich effusion would require thoracostomy, regardless of fluid pH. As a general rule, freely flowing effusions that separate the lung from the chest wall by more than 1 cm on a lateral decubitus film, those that are loculated, and those that do not flow freely, should be sampled and/or drained (Fig. 26-3). The exis- tence of parietal pleural thickening documented on a contrast-enhanced CT scan suggests an intense inflammatory response and probable empyema. The ease and safety of thoracentesis in the ICU

A B FIGURE 26-3. Ultrasonic (A) and CT (B) images of complicated pleural effusion and empyema. Though less precise than CT, bedside ultrasound is convenient and may be definitive. Both are helpful prior to interventional procedures such as chest tube placement.

557

CHAPTER 26 • ICU Infections

Therapy Three basic principles apply to treating empyema: early diagnosis, appropriate antibiotic therapy, and thorough drainage. Of these, drainage is most important. Because there are no radiographic or physical examination features to distinguish an empyema from a routine pleural effusion, thoracen- tesis is required. Prompt diagnosis minimizes both early (sepsis and respiratory failure) and late compli- cations (fibrothorax and debilitation). When turbid, viscous, pleural fluid (especially if foul smelling) is obtained at thoracentesis, cultures for aerobic and anaerobic bacteria, tuberculosis, and fungi should be sent. In addition to routine cell counts and chemistry analysis, it is prudent to obtain triglycer- ide and cholesterol levels to exclude a diagnosis of chylothorax, which can have an empyema-like (tur- bid) appearance. (Effusions caused by rheumatoid disease also can have a similar appearance.) The pleural fluid should be Gram stained and sputum and blood cultures obtained. Antibiotic coverage should be chosen initially on the basis of the Gram stain and then fine-tuned by culture results. The usual etiologic suspects for pneumonia also cause empyema ( Streptococcus pneumoniae, H. influen- zae, anaerobic mouth flora); however, staphylococci also should be covered. Antibiotics alone are insuf- ficient; prompt insertion of a thoracostomy tube(s) of sufficient caliber to completely drain the pleu- ral space is almost always indicated. Several tubes may be required to fully drain the collection when the fluid is multiloculated. Chest CT guidance can be invaluable in guiding placement. Effusions that do not resolve with antibiotics, tube thoracostomy, and intrapleural thrombolytics may require explora- tion and drainage by thoracotomy or video-assisted thoracoscopy (VATS). Failure to resolve the acute process satisfactorily can require later pleural strip- ping or decortication. Relatively large collections of fluid that form after appropriate antibiotic ther- apy has been initiated can usually be managed by serial thoracentesis, rather than by indwelling chest tube. The latter becomes necessary, however, if the patient unexpectedly remains toxic appearing or the fluid loculates. Intravascular Catheter-Related Infections Intravascular catheter-related infections remain one of the top three causes of nosocomial sepsis; however, in ICUs with organized prevention plans,

may be enhanced by using ultrasound localization (see Chapter 11). Empyema is defined as an effusion with organ- isms detected by Gram stain or as “pus” in the pleu- ral space. Unfortunately, observers vary widely in their definition of pus. The diagnosis of empyema is not made by laboratory testing, and there are no specific laboratory cutoff values for what consti- tutes an empyema. Many empyemas do not have microorganisms visible on Gram stain examination, and not all empyemas grow bacteria in culture, per- haps because antibiotic therapy has already been administered or because the responsible organism is inherently difficult to isolate (e.g., anaerobes). If infected with bacteria, especially anaerobic bacteria, the odor of an empyema is memorable. Generally, accepted characteristics of an empyema are grossly cloudy or opaque appearance and thick, viscous character because of high levels of protein and leukocytes. Certainly, not all infected pleural fluids are thick. Yet, it is the physical characteristics of the fluid that make empyema important to diag- nose and treat appropriately. Intrapleural streptoki- nase can reduce the need for pleural decortication if used early in the clinical course and is often helpful later when tube drainage slows and pockets remain. Intrapleural streptokinase is associated with a low risk of either allergic reaction or systemic coagulop- athy. Several types of pleural effusions can mimic an empyema: chylothorax, rheumatoid effusion, tuber- culous effusion, and resolving hemothorax all can have the thick, turbid appearance characteristic of bacterial empyema. The clinical presentation of empyema can be subtle. It is not uncommon for elderly or debilitated patients to have empyema as the primary cause of or cocontributor to chronic wasting illness. The diag- nosis should be suspected in patients with unresolv- ing or hectic fever and pleural effusions that do not improve with antibiotic therapy. Empyema becomes more likely if the suspect fluid collection is adjacent to pneumonia. Because ICU chest films are often taken supine or semiupright, the classic “layering” of an effusion can be missed. Although decubitus views and pulmonary ultrasound enhance the like- lihood of finding an effusion, CT currently is the most definitive way to confirm a free or loculated fluid collection, especially if small or loculated. For febrile or frankly septic patients, especially those with an underlying pneumonia, the search for an empyema is reasonable.

558

SECTION II • Medical and Surgical Crises

the incidence can be reduced to a very low level. Despite better antibiotics, earlier recognition and improved understanding of the mechanisms of catheter-related infections, the case fatality rate for catheter-associated bacteremia remains significant. Mechanisms Three basic mechanisms can produce catheter- related infections (Fig. 26-4). (1) Most commonly, catheters are colonized at the skin–air interface, as bacteria migrate along their outer surfaces. Subcutaneous and eventual intravascular migration results in local infection or bacteremia if bacterial growth is uncontrolled by host defense or antibi- otic therapy. (2) Catheters also can become colo- nized by exposure to circulating microorganisms introduced into the circulation at a distant site. As foreign objects, standard catheters routinely form a “fibrin sheath” or biofilm. This microenviron- ment is a stagnant, fertile environment for patho- gen growth, helping sources of bacteria or fungi far distant from the catheter to seed these indwelling lines. Although antiadhesive treatments that mimic the cellular glycocalyx and impregnated biocidal coatings such as silver salts and chlorhexidine are helpful, they encumber added cost and a universally effective prophylactic approach is not yet at hand. (3) Only rarely, catheter-related infections are due to the infusion of a contaminated intravenous fluid or drug. Although, in theory, such infusate contami- nation can occur with any drug, the problem has been reported most often with parenteral nutrition

solutions and first-generation formulations of pro- pofol, an intravenous sedative/anesthetic with a

lipid vehicle. Risk Factors

Characteristics of patients at particular risk for catheter-related infection include diabetes mellitus, immunosuppressive therapy (especially neutrope- nia), immune deficiency diseases, skin diseases at the insertion site, and presence of sepsis at a dis- tinct source. Physician and environmental factors increasing the risk of intravascular catheter infec- tions include (1) catheter placement under emer- gency or nonsterile conditions, (2) insertion of multilumen catheters, (3) catheterization of a cen- tral vein, (4) prolonged catheterization at a single site, (5) placement by surgical cutdown, and (6) inexperience of the operator. Most catheter infec- tions can be prevented by using sterile technique when inserting, dressing, changing, and reconnect- ing catheters and by minimizing the frequency of catheter access. Rigorous sterility during insertion is essential; apart from sterile gloves, wide sterile barri- ers, surgical gowns, caps, and masks should be used for elective insertions. Chlorhexidine is superior to povidone-iodine solutions for skin preparation. Inexperience with catheter insertion increases infec- tion risk. (It is not clear whether catheters become contaminated during insertion or if less experienced operators are prone to produce more tissue trauma during the insertion process.) Multilumen catheters or catheters entered repeatedly (even for antibiotic

Intraluminal spread

Contaminated entry port

Hospital staff

Contaminated infusate

Extraluminal spread

Skin microflora

Dressing

Subcutaneous tissue

Hematogenous spread

Catheter tip contaminated on insertion through skin

FIGURE 26-4. Portals and pathways through which catheter-related infections can develop.

559

CHAPTER 26 • ICU Infections

administration) seem to have a higher infection rate. Neither antibiotic ointment applied at the cath- eter entry site nor systemic antibiotics convincingly decrease the risk of bacteremia. There is no clear evidence that determines the relative infective risk of internal jugular and sub- clavian sites when the duration of catheterization is controlled. Although the risk of pneumothorax is averted, the femoral approach limits leg movement, predisposes to deep venous thrombosis, and places the catheter in a region at risk for contamination by urine and stool, probably explaining the higher infec- tion risk compared to sites above the waist. Central venous catheters are more likely to become infected than peripheral catheters (in part because of dura- tion of catheterization). Peripherally inserted central catheters (PICC), which are usually inserted via a brachial vein, provide intermediate to long-term cen- tral access with a somewhat lower risk of infection. Pulmonary artery monitoring catheters and multilu- men catheters (risk, 10% to 20%) are more likely to become infected than are single-lumen catheters (risk, approx. 5%). Interestingly, venous catheters are more likely to become infected than arterial catheters. Whether this differential risk relates to the shorter duration of arterial catheterization, the greater flow of blood in the artery, the shorter length of the arterial cannula, or the site of placement (usu- ally in the radial artery) is unclear. Hypertonic flu- ids (peripheral total parenteral nutrition [TPN]) or highly caustic drugs (e.g., amphotericin, diazepam, phenytoin, erythromycin) may induce a chemical phlebitis, facilitating bacterial overgrowth. To minimize the risk of infection, intravenous sites should be closely monitored and connecting

tubing should be changed every 24 to 48 hours. Dressings that allow continual (sterile) observation of the wound puncture site are helpful for surveil- lance. Placing impregnated disc barriers around the catheter at its point of skin entry during insertion may help reduce the infective risk (Fig. 26-5). Blood withdrawal increases the risk of infection, as does the filling of tubing systems in advance of their use. Even minute quantities of blood or fat provide nutri- ents adequate to support the growth of most bacte- ria; therefore, changing tubing after infusing blood or lipids reduces infection risk. Continuous flush solutions and pressure-monitoring devices attached to arterial catheters pose special hazards. Reducing the number of catheter entries for blood sampling reduces infection risk. It is especially important to avoid contamination of pressure measuring cath- eters during calibration. Contamination of Swan– Ganz catheters may be reduced by minimizing the number of cardiac output determinations and by using sterile precautions during preparation and introduction of the injectate. Because of the esca- lating risk of infection, central venous and arterial catheters should be removed within 3 to 5 days of placement whenever possible. Obviously, there are situations in which all potential access sites have been exhausted or the risk of catheter reinsertion outweighs the risk of infection posed by leaving an existing catheter in place. Therefore, the need for and timing of catheter replacement must be indi- vidualized. There are no credible data to support a practice of routinely changing catheters over a flex- ible guidewire, and doing so in patients with estab- lished severe sepsis makes little sense unless all other sites and options for catheter insertion have

FIGURE 26-5. Insertion site dressings that promote prevention of infection and monitoring of puncture wound status.