Summary

This document updates and replaces CDC's previously published "Guideline for Prevention of Nosocomial Pneumonia" (Infect Control 1982;3:327-33, Respir Care 1983;28:221-32, and Am J Infect Control 1983;11:230-44). This revised guideline is designed to reduce the incidence of nosocomial pneumonia and is intended for use by personnel who are responsible for surveillance and control of infections in acute-care hospitals; the information may not be applicable in long-term-care facilities because of the unique characteristics of such settings. This revised guideline addresses common problems encountered by infection- control practitioners regarding the prevention and control of nosocomial pneumonia in U.S. hospitals. Sections on the prevention of bacterial pneumonia in mechanically ventilated and/or critically ill patients, care of respiratory-therapy devices, prevention of cross-contamination, and prevention of viral lower respiratory tract infections (e.g., respiratory syncytial virus {RSV} and influenza infections) have been expanded and updated. New sections on Legionnaires disease and pneumonia caused by Aspergillus sp. have been included. Lower respiratory tract infection caused by Mycobacterium tuberculosis is not addressed in this document. Part I, "An Overview of the Prevention of Nosocomial Pneumonia, 1994," provides the background information for the consensus recommendations of the Hospital Infection Control Practices Advisory Committee (HICPAC) in Part II, "Recommendations for Prevention of Nosocomial Pneumonia."

Pneumonia is the second most common nosocomial infection in the United States and is associated with substantial morbidity and mortality. Most patients who have nosocomial pneumonia are infants, young children, and persons greater than 65 years of age; persons who have severe underlying disease, immunosuppression, depressed sensorium, and/or cardiopulmonary disease; and persons who have had thoracoabdominal surgery. Although patients receiving mechanically assisted ventilation do not represent a major proportion of patients who have nosocomial pneumonia, they are at highest risk for acquiring the infection. Most bacterial nosocomial pneumonias occur by aspiration of bacteria colonizing the oropharynx or upper gastrointestinal tract of the patient. Because intubation and mechanical ventilation alter first-line patient defenses, they greatly increase the risk for nosocomial bacterial pneumonia. Pneumonias caused by Legionella sp., Aspergillus sp., and influenza virus are often caused by inhalation of contaminated aerosols. RSV infection usually occurs after viral inoculation of the conjunctivae or nasal mucosa by contaminated hands. Traditional preventive measures for nosocomial pneumonia include decreasing aspiration by the patient, preventing cross-contamination or colonization via hands of personnel, appropriate disinfection or sterilization of respiratory-therapy devices, use of available vaccines to protect against particular infections, and education of hospital staff and patients. New measures being investigated involve reducing oropharyngeal and gastric colonization by pathogenic microorganisms.

Part 1. An Overview of the Prevention of Nosocomial Pneumonia, 1994

INTRODUCTION

This document updates and replaces CDC's previously published "Guideline for Prevention of Nosocomial Pneumonia" (Infect Control 1982;3:327-33, Respir Care 1983; 28:221-32, and Am J Infect Control 1983;11:230-44). This revised guideline is designed to reduce the incidence of nosocomial pneumonia and is intended for use by personnel who are responsible for surveillance and control of infections in acute-care hospitals; the information may not be applicable in long-term-care facilities because of the unique characteristics of such settings.

This revised guideline addresses common problems encountered by infection-control practitioners regarding the prevention and control of nosocomial pneumonia in U.S. hospitals. Sections concerning the prevention of bacterial pneumonia in mechanically ventilated and/or critically ill patients, care of respiratory-therapy devices, prevention of cross-contamination, and prevention of viral lower respiratory tract infections (e.g., respiratory syncytial virus {RSV} and influenza infections) have been expanded and updated. New sections on Legionnaires disease and pneumonia caused by Aspergillus sp. have been included. Lower respiratory tract infection caused by Mycobacterium tuberculosis is not addressed in this document; CDC published such recommendations previously (1).

Part I, "An Overview of the Prevention of Nosocomial Pneumonia, 1994," provides the background information for the consensus recommendations of the Hospital Infection Control Practices Advisory Committee (HICPAC) in Part II, "Recommendations for Prevention of Nosocomial Pneumonia." HICPAC was established in 1991 to provide advice and guidance to the Secretary and the Assistant Secretary for Health, U.S. Department of Health and Human Services; the Director, CDC; and the Director, National Center for Infectious Diseases (NCID), CDC, regarding the practice of hospital infection control and strategies for surveillance, prevention, and control of nosocomial infections in U.S. hospitals. HICPAC also advises CDC on periodic updating of guidelines and other policy statements regarding prevention of nosocomial infections. This guideline is the first of a series of CDC guidelines being revised by HICPAC and NCID.

This guideline can be an important resource for educating health-care workers (HCWs) regarding prevention and control of nosocomial respiratory tract infections. Because education of HCWs is the cornerstone of an effective infection-control program, hospitals should give high priority to continuing infection-control educational programs for these personnel.

BACKGROUND

Pneumonia is the second most common nosocomial infection in the United States and is associated with substantial morbidity and mortality. Most patients who have nosocomial pneumonia are infants, young children, and persons greater than 65 years of age; persons who have severe underlying disease, immunosuppression, depressed sensorium, and/or cardiopulmonary disease; and persons who have had thoracoabdominal surgery. Although patients receiving mechanically assisted ventilation do not represent a major proportion of patients who have nosocomial pneumonia, they are at highest risk for acquiring the infection.

Most bacterial nosocomial pneumonias occur by aspiration of bacteria colonizing the oropharynx or upper gastrointestinal tract of the patient. Because intubation and mechanical ventilation alter first-line patient defenses, they greatly increase the risk for nosocomial bacterial pneumonia. Pneumonias caused by Legionella sp., Aspergillus sp., and influenza virus are often caused by inhalation of contaminated aerosols. RSV infection usually occurs after viral inoculation of the conjunctivae or nasal mucosa by contaminated hands.

Traditional preventive measures for nosocomial pneumonia include decreasing aspiration by the patient, preventing cross-contamination or colonization via hands of HCWs, appropriate disinfection or sterilization of respiratory-therapy devices, use of available vaccines to protect against particular infections, and education of hospital staff and patients. New measures being investigated involve reducing oropharyngeal and gastric colonization by pathogenic microorganisms.

BACTERIAL PNEUMONIA

1. Etiologic Agents

   The reported distribution of etiologic agents that cause nosocomial pneumonia differs between hospitals because of different patient populations and diagnostic methods employed (2-10). In general, however, bacteria have been the most frequently isolated pathogens (2-6,9,11-13). During 1986-1989, aerobic bacteria comprised at least 73%, and fungi 4%, of isolates from sputum and tracheal aspirates obtained from patients who had pneumonia at the University of Michigan Hospitals and at hospitals participating in the National Nosocomial Infection Surveillance (NNIS) System; only a few anaerobic bacteria and no viruses were reported, probably because anaerobic and viral cultures were not performed routinely in the reporting hospitals (Table_1) (3). Similarly, cultures of bronchoscopic specimens obtained from mechanically ventilated patients who had pneumonia have rarely yielded anaerobes (5-7,9,11,14,15). Only one study, which was based primarily on cultures of transtracheal aspirates obtained from patients not receiving mechanically assisted ventilation, reported a predominance of anaerobes (4).

   Nosocomial bacterial pneumonias are frequently polymicrobial (4,7,9,11, 12,15-19), and gram-negative bacilli are usually the predominant organisms (Table_1) (2-6,9,11-13). However, Staphylococcus aureus (especially methicillin-resistant S. aureus) (5,7,10,15,20,21) and other gram-positive cocci, including Streptococcus pneumoniae (5,7), have emerged recently as important isolates (14). In addition, Haemophilus influenzae has been isolated from mechanically ventilated patients who had pneumonia that occurred within 48-96 hours after intubation (3-5,12,15,22). In hospitals participating in the NNIS, Pseudomonas aeruginosa, Enterobacter sp., Klebsiella pneumoniae, Escherichia coli, Serratia marcescens, and Proteus sp. comprised 50% of the isolates from cultures of respiratory tract specimens obtained from patients for whom nosocomial pneumonia was diagnosed by using clinical criteria; S. aureus accounted for 16%, and H. influenzae, for 6% (Table_1) (3). Another study reported that gram-negative bacilli were present in 75% of quantitative cultures of protected-specimen brushings (PSB) obtained from patients who had acquired nosocomial pneumonia after receiving mechanically assisted ventilation; 40% of these cultures were polymicrobial (5). In another published report, 20% of pathogens recovered from cultures of PSB, blood, pleural fluid, or percutaneous lung aspirate were gram-negative bacilli in pure culture, and 17% were polymicrobial; however, 54% of specimens did not yield any microorganism, probably because the patients from whom these cultures were obtained had been treated with antibiotics (6).

2. Diagnosis

   Nosocomial bacterial pneumonia has been difficult to diagnose (7,8,16,23-32). Frequently, the criteria for diagnosis have been fever, cough, and development of purulent sputum, in conjunction with radiologic evidence of a new or progressive pulmonary infiltrate, a suggestive Gram stain, and positive cultures of sputum, tracheal aspirate, pleural fluid, or blood (3,4,23,25,33-36). Although clinical findings in conjunction with cultures of sputum or tracheal specimens may be sensitive for bacterial pathogens, they are highly nonspecific, especially in patients receiving mechanically assisted ventilation (8,9,12-15,18,24-26,29,31,37-42); conversely, cultures of blood or pleural fluid have very low sensitivity (8,18,19,43).

   Because of these problems, a group of investigators recently formulated consensus recommendations for standardizing methods used to diagnose pneumonia in clinical research studies of ventilator-associated pneumonia (44-46). These methods involve bronchoscopic techniques such as quantitative culture of PSB (5,7-9,13,15,27,31,38,41,47,48), bronchoalveolar lavage (BAL) (7,12,41,47,49-54), and protected BAL (pBAL) (14). The reported sensitivities of such methods have ranged, depending on the tests or diagnostic criteria with which they were compared, from 70% to 100%, and the reported specificities of these methods have ranged from 60% to 100%. These methods are invasive and might cause complications such as hypoxemia, bleeding, or arrhythmia (8,13,42,44,52,55,56). In addition, the sensitivity of the PSB procedure may be decreased for patients receiving antibiotic therapy (9,13,27). Nonbronchoscopic (NB) procedures (e.g., NB-pBAL {12,27,57, 58} or NB-PSB {13}, which utilize blind catheterization of the distal airways) and quantitative culture of endotracheal aspirate (59,60) have been developed recently. Of these procedures, endotracheal aspirate culture might be the most practical. The use of these bronchoscopic and nonbronchoscopic diagnostic tests could help to better define the epidemiology of nosocomial pneumonia, especially in patients receiving mechanically assisted ventilation; however, additional studies are needed to determine each test's applicability in daily clinical practice.

Epidemiology

Results of the NNIS indicate that pneumonias (diagnosed on the basis of the CDC surveillance definition of nosocomial pneumonia) account for approximately 15% of all hospital-associated infections and are the second most common type of nosocomial infection after those of the urinary tract (2,61). In 1984, the overall incidence of lower respiratory tract infection was six cases per 1,000 discharged patients (2). The incidence per 1,000 discharged patients ranged from 4.2 cases in nonteaching hospitals to 7.7 in university-affiliated hospitals, probably reflecting institutional differences in the level of patients' risk for acquiring nosocomial pneumonia.

Nosocomial bacterial pneumonia often has been identified as a postoperative infection (62,63). In the Study of the Efficacy of Nosocomial Infection Control, which was conducted in the 1970s, 75% of reported cases of nosocomial bacterial pneumonia occurred in patients who had had a surgical operation; the risk was 38 times greater for patients who had thoracoabdominal procedures than for those who had procedures involving other body sites (63). More recent epidemiologic studies, including NNIS studies, have identified other subsets of patients at high risk for acquiring nosocomial bacterial pneumonia. Such patients include persons greater than 70 years of age; persons who have endotracheal intubation and/or mechanically assisted ventilation, a depressed level of consciousness (particularly those with closed-head injury), or underlying chronic lung disease; and persons who have previously had an episode of a large-volume aspiration. Other risk factors include 24-hour ventilator-circuit changes, hospitalization during the fall or winter, stress-bleeding prophylaxis with cimetidine (either with or without antacid), administration of antimicrobials, presence of a nasogastric tube, severe trauma, and recent bronchoscopy (6,34,35,64-74).

The NNIS has stratified the incidence density of nosocomial pneumonia by patients' use of mechanical ventilation and type of intensive-care unit (ICU). From 1986 through 1990, the median rate of ventilator-associated pneumonia cases per 1,000 ventilator-days ranged from 4.7 cases in pediatric ICUs to 34.4 cases in burn ICUs (66). In comparison, the median rate of nonventilator-associated pneumonia cases per 1,000 ICU-days ranged from zero cases in pediatric and respiratory ICUs to 3.2 cases in trauma ICUs.

Nosocomial pneumonia has been associated with high fatality rates. Crude mortality rates of 20%-50% and attributable mortality rates of 30%-33% have been reported; in one study, the number of deaths attributed to pneumonia reflected 60% of all deaths resulting from nosocomial infections (17,35,74-80). Patients receiving mechanically assisted ventilation have higher mortality rates than do patients not receiving ventilation support; however, other factors (e.g., the patient's underlying disease{s} and organ failure) are stronger predictors of death in patients who have pneumonia (34,74).

Analyses of pneumonia-associated morbidity have indicated that pneumonia could prolong hospitalization by 4-9 days (79-83); in the United States, a conservative estimate of the direct cost of this prolonged hospitalization is $1.2 billion per year (83). Nosocomial pneumonia is a major infection-control problem because of its reported frequency, associated high fatality rate, and attendant costs.

IV. Pathogenesis

Bacteria can invade the lower respiratory tract by aspiration of oropharyngeal organisms, inhalation of aerosols containing bacteria, or, less frequently, by hematogenous spread from a distant body site (Figure_1). In addition, bacterial translocation from the gastrointestinal tract has been hypothesized recently as a mechanism for infection. Of these routes, aspiration is believed to be the most important for both nosocomial and community-acquired pneumonia.

In radioisotope-tracer studies, 45% of healthy adults were found to aspirate during sleep (84). Persons who swallow abnormally (e.g., those who have depressed consciousness, respiratory tract instrumentation and/or mechanically assisted ventilation, or gastrointestinal tract instrumentation or diseases) or who have just undergone surgery are particularly likely to aspirate (6,34,35,63,85-87).

The high incidence of gram-negative bacillary pneumonia in hospitalized patients might result from factors that promote colonization of the pharynx by gram-negative bacilli and the subsequent entry of these organisms into the lower respiratory tract (33,88-91). Although aerobic gram-negative bacilli are recovered infrequently or are found in low numbers in pharyngeal cultures of healthy persons (88,92), the likelihood of colonization substantially increases in comatose patients, in patients treated with antimicrobial agents, and in patients who have hypotension, acidosis, azotemia, alcoholism, diabetes mellitus, leukocytosis, leukopenia, pulmonary disease, or nasogastric or endotracheal tubes in place (33,91,93,94).

Oropharyngeal or tracheobronchial colonization by gram-negative bacilli begins with the adherence of the microorganisms to the host's epithelial cells (90,95-97). Adherence may be affected by multiple factors associated with the bacteria (e.g., presence of pili, cilia, capsule, or production of elastase or mucinase), host cell (e.g., surface proteins and polysaccharides), and environment (e.g., pH and presence of mucin in respiratory secretions) (89,90,95,98-107). Although the exact interactions between these factors have not been fully elucidated, studies indicate that certain substances (e.g., fibronectin) can inhibit the adherence of gram-negative bacilli to host cells (98,100,108). Conversely, certain conditions (e.g., malnutrition, severe illness, or postoperative state) can increase adherence of gram-negative bacteria (89,98,102,107,109).

The stomach also might be an important reservoir of organisms that cause nosocomial pneumonia (34,110-114). The role of the stomach as such a reservoir might differ depending on the patient's underlying conditions and on prophylactic or therapeutic interventions (22,111,115-118). In healthy persons, few bacteria entering the stomach survive in the presence of hydrochloric acid at pH less than 2 (119,120). However, when gastric pH increases from the normal levels to greater than or equal to 4, microorganisms are able to multiply to high concentrations in the stomach (117,119,121-123). This can occur in elderly patients (121); in patients who have achlorhydria (119), ileus, or upper gastrointestinal disease; and in patients receiving enteral feeding, antacids, or histamine-2 {H-2} antagonists (111,117,118, 123-125). Other factors (e.g., duodeno-gastric reflux and the presence of bile) may contribute to gastric colonization in patients who have impaired intestinal motility; these other factors need further investigation (116).

Bacteria also can enter the lower respiratory tract of hospitalized patients through inhalation of aerosols generated primarily by contaminated respiratory-therapy or anesthesia-breathing equipment (126-129). Outbreaks related to the use of respiratory-therapy equipment have been associated with contaminated nebulizers, which are humidification devices that produce large amounts of aerosol droplets less than 4 um via ultrasound, spinning disk, or the Venturi mechanism (126,129,130). When the fluid in the reservoir of a nebulizer becomes contaminated with bacteria, the aerosol produced may contain high concentrations of bacteria that can be deposited deep in the patient's lower respiratory tract (126,130,131). Contaminated aerosol inhalation is particularly hazardous for intubated patients because endotracheal and tracheal tubes provide direct access to the lower respiratory tract. In contrast to nebulizers, bubble-through or wick humidifiers primarily increase the water-vapor (or molecular-water) content of inspired gases. Although heated bubble-through humidifiers generate aerosol droplets, they do so in quantities that may not be clinically important (127,132); wick humidifiers do not generate aerosols.

Bacterial pneumonia has resulted, in rare instances, from hematogenous spread of infection to the lung from another infection site (e.g., pneumonia resulting from purulent phlebitis or right-sided endocarditis). Another mechanism, translocation of viable bacteria from the lumen of the gastrointestinal tract through epithelial mucosa to the mesenteric lymph nodes and to the lung, has been demonstrated in animal models (133). Translocation is postulated to occur in patients with immunosuppression, cancer, or burns (133); however, data are insufficient to describe this mechanism in humans (134).

V. Risk Factors and Control Measures

Several large studies have examined the potential risk factors for nosocomially acquired bacterial pneumonia (Table_2) (6,34,35,135, 136). Although specific risk factors have differed between study populations, they can be grouped into the following general categories:

1. host factors (e.g., extremes of age and severe underlying conditions, including immunosuppression); b) factors that enhance colonization of the oropharynx and/or stomach by microorganisms (e.g., administration of antimicrobials, admission to an ICU, underlying chronic lung disease, or coma); c) conditions favoring aspiration or reflux (e.g., endotracheal intubation, insertion of nasogastric tube, or supine position); d) conditions requiring prolonged use of mechanical ventilatory support with potential exposure to contaminated respiratory equipment and/or contact with contaminated or colonized hands of HCWs; and e) factors that impede adequate pulmonary toilet (e.g., undergoing surgical procedures that involve the head, neck, thorax, or upper abdomen or being immobilized as a result of trauma or illness) (6,33-35,62,73, 74,135).

1. Oropharyngeal, Tracheal, and Gastric Colonization

   The association between colonization of the oropharynx (88,137), trachea (138), or stomach (110,111,117,123) and predisposition to gram-negative bacillary pneumonia prompted efforts to prevent infection by using either prophylactic local application of antimicrobial agent(s) (139,140) or local bacterial interference (141,142). Although early studies suggested that the first method (i.e., use of aerosolized antimicrobials) could eradicate common gram-negative pathogens from the upper respiratory tract (138), superinfection occurred in some patients receiving this therapy (139-141,143,144). The second method (i.e., bacterial interference {with alpha-hemolytic streptococci}) has been used successfully by some investigators to prevent oropharyngeal colonization by aerobic gram-negative bacilli (141). However, the efficacy of this method for general usage has not been evaluated.

   In many studies, the administration of antacids and H-2 blockers for prevention of stress bleeding in critically ill, postoperative, and/or mechanically ventilated patients has been associated with gastric bacterial overgrowth (34,112,113, 118,122,123,145-147). Sucralfate, a cytoprotective agent that has little effect on gastric pH and may have bactericidal properties of its own, has been suggested as a potential substitute for antacids and H-2 blockers (148-150). The results of clinical trials comparing the risk for pneumonia in patients receiving sucralfate with that in patients treated with antacids and/or H-2 blockers have been variable (112,118,147,148,151-153). In most randomized trials, ICU patients receiving mechanically assisted ventilation who were treated either with only antacids or with antacids and H-2 blockers had increased gastric pH, high bacterial counts in the gastric fluid, and increased risk for pneumonia in comparison with patients treated with sucralfate (112,118,147,148,151). In one study of a large number of patients, the incidence of early-onset pneumonia (i.e., onset occurring less than or equal to 4 days after intubation) did not differ between patient groups, but late-onset pneumonia occurred in 5% of 76 patients treated with sucralfate, 16% of 69 treated with antacids, and 21% of 68 treated with an H-2 blocker (147). Conversely, a meta-analysis of data from eight earlier studies (154) and a later study comparing sucralfate with ranitidine (153) did not indicate a strong association between nosocomial pneumonia and drugs that increase gastric pH. Additional studies, in which bronchoscopy with either PSB or BAL is used to more reliably diagnose pneumonia, are being conducted to compare the efficacy of sucralfate and ranitidine.

   Selective decontamination of the digestive tract (SDD) is another strategy designed to prevent bacterial colonization and lower respiratory tract infection in mechanically ventilated patients (155-179). SDD is aimed at preventing oropharyngeal and gastric colonization with aerobic gram-negative bacilli and Candida sp. without altering the anaerobic flora (Table_3). Various SDD regimens use a combination of locally administered nonabsorbable antibiotic agents, such as polymyxin and an aminoglycoside (either tobramycin, gentamicin, or, rarely, neomycin) or a quinolone (either norfloxacin or ciprofloxacin) coupled with either amphotericin B or nystatin. The local antimicrobial preparation is applied as a paste to the oropharynx and administered either orally or via the nasogastric tube four times a day. In addition, in many studies, a systemic (intravenous) antimicrobial (e.g., cefotaxime or trimethoprim) is administered to the patient.

   Although most studies (155-158,160-167,169,170,175-177), including two meta-analyses (171,178), have demonstrated a decrease in the rates of nosocomial respiratory infections after SDD, these studies have been difficult to assess because they have differed in design and study population and many have had short follow-up periods (Table_3). In most of these studies, the diagnosis of pneumonia was based on clinical criteria; bronchoscopy with BAL or PSB was used in only a few studies (159,162,173,175-177,179).

   Two recently published reports of large, double-blind, placebo- controlled trials demonstrated no benefit from SDD (173,174). One of these studies, which was conducted in France, noted that the incidence of gram-negative bacillary pneumonia decreased significantly after SDD, but this decrease was not accompanied by a decrease in pneumonia from all causes (173). In the other study, no differences were noted between patients randomly assigned to SDD or placebo treatment conditions; however, both patient groups also received simultaneous treatment with intravenous cefotaxime (174).

   Although an earlier meta-analysis indicated a trend toward decreased mortality in patients administered SDD (171), a more recent and more extensive analysis highlights the equivocal effect of SDD on patient mortality, as well as the high cost of using SDD to prevent nosocomial pneumonia or death resulting from nosocomial pneumonia (i.e., to prevent one case of nosocomial pneumonia, six patients {range: five to nine patients} would have to be administered SDD; to prevent one death, 23 patients {range: 13-39 patients}) (178). Furthermore, both the development of antimicrobial resistance and superinfection with gram-positive bacteria and other antibiotic- resistant nosocomial pathogens are public health concerns (156,158, 159,161,175,180). Thus, currently available data do not justify the routine use of SDD for prevention of nosocomial pneumonia in ICU patients. SDD may be ultimately useful for specific subsets of ICU patients, such as patients with trauma or severe immunosuppression (e.g., bone-marrow-transplant recipients).

   A new approach advocated to prevent oropharyngeal colonization in patients receiving enteral nutrition is to reduce bacterial colonization of the stomach by acidifying the enteral feed (181). Although the absence of bacteria from the stomach has been confirmed in patients given acidified enteral feeding, the effect on the incidence of nosocomial pneumonia has not been evaluated (181).

2. Aspiration of Oropharyngeal and Gastric Flora

   Clinically important aspiration usually occurs in patients who a) have a depressed level of consciousness; b) have dysphagia resulting from neurologic or esophageal disorders; c) have an endotracheal (nasotracheal or orotracheal), tracheostomal, or enteral (nasogastric or orogastric) tube in place; and/or d) are receiving enteral feeding (35,84,85,182-186). Placement of an enteral tube may increase nasopharyngeal colonization, cause reflux of gastric contents, or allow bacterial migration via the tube from the stomach to the upper airway (183,186-188). When enteral feedings are administered, gross contamination of the enteral solution during preparation (189-191) and elevated gastric pH (70,192,193) may lead to gastric colonization with gram-negative bacilli. In addition, gastric reflux and aspiration might occur because of increased intragastric volume and pressure (70,117,183).

   Although prevention of pneumonia in such patients may be difficult, methods that make regurgitation less likely (e.g., placing the patient in a semirecumbent position {i.e., by elevating the head of the bed} and withholding enteral feeding if the residual volume in the stomach is large or if bowel sounds are not heard upon auscultation of the abdomen) may be beneficial (185,194-197). Conversely, equivocal results have been obtained by a) administering enteral nutrition intermittently in small boluses rather than continuously (70,193); b) using flexible, small-bore enteral tubes (186,198); or c) placing the enteral tube below the stomach (e.g., in the jejunum) (199,200).

3. Mechanically Assisted Ventilation and Endotracheal Intubation

   Patients receiving continuous, mechanically assisted ventilation have 6-21 times the risk for acquiring nosocomial pneumonia compared with patients not receiving ventilatory support (34,63,65,75). One study indicated that the risk for developing ventilator-associated pneumonia increased by 1% per day (5). This increased risk was attributed partially to carriage of oropharyngeal organisms upon passage of the endotracheal tube into the trachea during intubation, as well as to depressed host defenses secondary to the patient's severe underlying illness (6,34,35,201). In addition, bacteria can aggregate on the surface of the tube over time and form a glycocalyx (i.e., a biofilm) that protects the bacteria from the action of antimicrobial agents or host defenses (202). Some researchers believe that these bacterial aggregates can become dislodged by ventilation flow, tube manipulation, or suctioning and subsequently embolize into the lower respiratory tract and cause focal pneumonia (203,204). Removing tracheal secretions by gentle suctioning and using aseptic techniques to reduce cross-contamination to patients from contaminated respiratory therapy equipment or contaminated or colonized hands of HCWs have been used traditionally to help prevent pneumonia in patients receiving mechanically assisted ventilation.

   The risk for pneumonia also is increased by the direct access of bacteria to the lower respiratory tract, which often occurs because of leakage around the endotracheal cuff (86,205), thus enabling pooled secretions above the cuff to enter the trachea (206). In one study, the occurrence of nosocomial pneumonia was delayed and decreased in intubated patients whose endotracheal tubes had a separate dorsal lumen that allowed drainage (i.e., by suctioning) of secretions in the space above the endotracheal tube cuff and below the glottis (206). However, additional studies are needed to determine the cost-benefit ratio of using this device.

4. Cross-Colonization Via Hands of HCWs

   Pathogens that cause nosocomial pneumonia (e.g., gram-negative bacilli and S. aureus) are ubiquitous in hospitals, especially in intensive- or critical-care areas (207,208). Transmission of these microorganisms to patients frequently occurs via an attending HCW's hands that have become contaminated or transiently colonized with the microorganisms (209-215). Procedures such as tracheal suctioning and manipulation of the ventilator circuit or endotracheal tubes increase the opportunity for cross-contamination (215,216). The risk for cross-contamination can be reduced by using aseptic techniques and sterile or disinfected equipment when appropriate (65) and by eliminating pathogens from the hands of HCWs (65,215,217-219).

   In theory, adequate handwashing is an effective way of removing transient bacteria from the hands (218,219); however, personnel compliance with handwashing recommendations has been generally poor (220-223). For this reason, the routine use of gloves has been advocated to help prevent cross-contamination (224,225). The routine use of gloves, in addition to the use of gowns, was associated with a decrease in the incidence of nosocomial RSV infection (226) and other infections acquired in ICUs (227). However, nosocomial pathogens can colonize gloves (228), and outbreaks have been traced to HCWs who did not change gloves after having contact with one patient and before providing care to another (229,230). In addition, gloved hands can be contaminated through leaks in the gloves (231).

5. Contamination of Devices Used on the Respiratory Tract

   Devices used on the respiratory tract for respiratory therapy (e.g., nebulizers), diagnostic examination (e.g., bronchoscopes and spirometers), and administration of anesthesia are potential reservoirs and vehicles for infectious microorganisms (65,232-236). Routes of transmission might be from device to patient (127,129, 234-244), from one patient to another (245,246), or from one body site to the lower respiratory tract of the same patient via hand or device (233,246-248). Contaminated reservoirs of aerosol-producing devices (e.g., nebulizers) can allow the growth of hydrophilic bacteria that subsequently can be aerosolized during use of the device (126,129,130,242). Gram-negative bacilli (e.g., Pseudomonas sp., Xanthomonas sp., Flavobacterium sp., Legionella sp., and nontuberculous mycobacteria) can multiply to substantial concentrations in nebulizer fluid (241,249-251) and increase the risk for pneumonia in patients using such devices (127-130,241,242, 252,253).

   Proper cleaning and sterilization or disinfection of reusable equipment are important components of a program to reduce infections associated with respiratory therapy and anesthesia equipment (234, 235,237-240,242,254-259). Many devices or parts of devices used on the respiratory tract have been categorized as semicritical in the Spaulding classification system for appropriate sterilization or disinfection of medical devices because they come into direct or indirect contact with mucous membranes but do not ordinarily penetrate body surfaces (Appendix A), and the associated risk for infection in patients after the use of such devices is less than that associated with devices that penetrate normally sterile tissues (260). Thus, if sterilization of these devices by steam autoclave or ethylene oxide is not possible or cost-effective (261), they can be subjected to high-level disinfection by pasteurization at 75 C for 30 minutes (262-265) or by use of liquid chemical disinfectants approved by the Environmental Protection Agency (EPA) as sterilants/disinfectants and approved for use on medical instruments by the Food and Drug Administration (225, 266-268).

   If a respiratory device needs rinsing to remove a residual liquid chemical sterilant/disinfectant after chemical disinfection, sterile water is preferred because tap or locally prepared distilled water might contain microorganisms that can cause pneumonia (249,250, 269-272). In some hospitals, a tap-water rinse followed by air- drying with or without an alcohol rinse (i.e., to hasten drying) is used (273). In theory, if complete drying is achieved after a tap-water rinse, the risk for nosocomial pneumonia associated with the use of the device is probably low. Air drying reduces the level of microbial contamination of the hands of HCWs after washing, and air drying also reduces contamination of gastrointestinal endoscopes (274-276). However, many semicritical items used on the respiratory tract (e.g., corrugated tubing, jet or ultrasonic nebulizers, and bronchoscopes) are difficult to dry, and the degree of dryness of a device is difficult to assess (265). Data are insufficient regarding the safety of routinely using tap water for rinsing (followed by drying) reusable semicritical respiratory devices after their disinfection or between their uses on the same patient (242,258,273, 277).

1. Mechanical Ventilators, Breathing Circuits, Humidifiers, Heat-Moisture Exchangers, and In-Line Nebulizers

1. Mechanical ventilators. The internal machinery of mechanical ventilators used for respiratory therapy is not considered an important source of bacterial contamination of inhaled gas (278). Thus, routine sterilization or high-level disinfection of the internal machinery is considered unnecessary. Using high-efficiency bacterial filters at various positions in the ventilator breathing circuit had been advocated previously (279,280). Filters interposed between the machinery and the main breathing circuit can eliminate contaminants from the driving gas and prevent retrograde contamination of the machine by the patient; however, these filters also might alter the functional specifications of the breathing device by impeding high gas flows (279-281). Placement of a filter or condensate trap at the expiratory-phase tubing of the mechanical-ventilator circuit may help prevent cross- contamination of the ventilated patient's immediate environment (247,282), but the importance of such filters in preventing nosocomial pneumonia needs further evaluation.

2. Breathing circuits, humidifiers, and heat-moisture exchangers. In the United States, most hospitals use ventilators with either bubble-through or wick humidifiers that produce either insignificant (132,283) or no aerosols, respectively, for humidification. Thus, these devices probably do not pose an important risk for pneumonia in patients. In addition, bubble-through humidifiers are usually heated to temperatures that reduce or eliminate bacterial pathogens (283,284). Sterile water, however, is still usually used to fill these humidifiers (285) because tap or distilled water might contain microorganisms, such as Legionella sp., that are more heat-resistant than other bacteria (252,271).

   The potential risk for pneumonia in patients using mechanical ventilators that have heated bubble-through humidifiers stems primarily from the condensate that forms in the inspiratory- phase tubing of the ventilator circuit as a result of the difference in the temperatures of the inspiratory-phase gas and ambient air; condensate formation increases if the tubing is unheated (286). The tubing and condensate can rapidly become contaminated, usually with bacteria that originate in the patient's oropharynx (286). In one study, 33% of inspiratory circuits were colonized with bacteria via this route within 2 hours, and 80% within 24 hours, after initiation of mechanical ventilation (286). Spillage of the contaminated condensate into the patient's tracheobronchial tree, as can occur during procedures in which the tubing is moved (e.g., for suctioning, adjusting the ventilator setting, or feeding or caring for the patient), may increase the risk for pneumonia in the patient (286). Thus, in many hospitals, HCWs are trained to prevent such spillage and to drain the fluid periodically. Microorganisms contaminating ventilator-circuit condensate can be transmitted to other patients via the hands of HCWs handling the fluid, especially if the HCW neglects washing hands after handling the condensate.

   The role of ventilator-tubing changes in preventing pneumonia in patients using mechanical ventilators with bubble-through humidifiers has been investigated. Initial studies of in-use contamination of mechanical ventilator circuits with humidifiers have indicated that neither the rate of bacterial contamination of inspiratory-phase gas nor the incidence of pneumonia was significantly increased when tubing was changed every 24 hours rather than every 8 or 16 hours (287). A later study indicated that changing the ventilator circuit every 48 hours rather than every 24 hours did not result in an increase in contamination of the inspiratory-phase gas or tubing of the ventilator circuits (288). In addition, the incidence of nosocomial pneumonia was not significantly higher when circuits were changed every 48 hours rather than every 24 hours (288). More recent reports suggest that the risk for pneumonia may not increase when the interval for circuit change is prolonged beyond 48 hours. Another study indicated that the risk for pneumonia was not significantly higher when the circuits were never changed for the duration of use by the patient (eight {29%} of 28 patients) rather than when the circuits were changed every 48 hours (11 {31%} of 35 patients) (289).

   These findings indicate that the recommended daily change in ventilator circuits may be extended to greater than or equal to 48 hours. This change in recommendation could result in substantial savings for U.S. hospitals by reducing the number of circuits used and the amount of personnel time required to change the circuits (285,288). The maximum time, however, that a circuit can be safely left unchanged on a patient has not been determined.

   Condensate formation in the inspiratory-phase tubing of a ventilator breathing circuit can be decreased by elevating the temperature of the inspiratory-phase gas with a heated wire in the inspiratory-phase tubing. However, in one report, three cases of endotracheal- or tracheostomy-tube blockage by dried secretions of the patient were attributed to the decrease in the relative humidity of inspired gas that resulted from the elevation of the gas temperature (290). Until additional information regarding the frequency of such cases is available, HCWs who provide care to patients requiring mechanical ventilation should be aware of the advantages and potential complications associated with using heated ventilator tubing.

   Condensate formation can be eliminated by using a heat-moisture exchanger (HME) or a hygroscopic condenser humidifier (i.e., an "artificial nose") (291-296). An HME recycles heat and moisture exhaled by the patient and eliminates the need for a humidifier. In the absence of a humidifier, no condensate forms in the inspiratory-phase tubing of the ventilator circuit. Thus, bacterial colonization of the tubing is prevented, and the need to change the tubing on a periodic basis is obviated (216). Some models of HMEs are equipped with bacterial filters, but the advantage of using such filters is unknown. HMEs can increase the dead space (i.e., the area of the lung in which air is not exchanged) and resistance to breathing, might leak around the endotracheal tube, and might result in drying of sputum and blockage of the tracheobronchial tree (297). Although recently developed HMEs that have humidifiers increase airway humidity without increasing colonization of bacteria (293, 298), additional studies are needed to determine whether the incidence of pneumonia is decreased (299-302).

3. Small-volume ("in-line") medication nebulizers. Small-volume medication nebulizers that are inserted in the inspiratory circuit of mechanical ventilators can produce bacterial aerosols (242). If such devices become contaminated by condensate in the inspiratory tubing of the breathing circuit, they can increase the patient's risk for pneumonia because the nebulizer aerosol is directed through the endotracheal tube and bypasses many of the normal host defenses against infection (286).

2. Large-Volume Nebulizers. Nebulizers with large-volume (greater than 500 cc) reservoirs, including those used in intermittent positive-pressure breathing (IPPB) machines and ultrasonic or spinning-disk room-air humidifiers, pose the greatest risk for pneumonia to patients, probably because of the large amount of aerosols they generate (237-241,252,303). These reservoirs can become contaminated by the hands of HCWs, unsterile humidification fluid, or inadequate sterilization or disinfection between uses (126). Once introduced into the reservoir, various bacteria, including Legionella sp., can multiply to sufficiently large numbers within 24 hours to pose a risk for infection in patients who receive inhalation therapy (128,129,241,253,303). Sterilization or high-level disinfection of these nebulizers can eliminate vegetative bacteria from their reservoirs and make them safe for patient use (260). However, unlike nebulizers attached to IPPB machines, room-air humidifiers have a high cost-benefit ratio: evidence of clinical benefits from their use in hospitals is lacking, and the potential cost of daily sterilization or disinfection of, and use of sterile water to fill, such devices is substantial.

3. Hand-Held Small-Volume Medication Nebulizers. Small-volume medication nebulizers used to administer bronchodilators, including nebulizers that are hand-held, can produce bacterial aerosols. Hand-held nebulizers have been associated with nosocomial pneumonia, including Legionnaires disease, resulting from either contamination with medications from multidose vials (304) or Legionella-contaminated tap water used for rinsing and filling the reservoir (258).

4. Suction Catheters, Resuscitation Bags, Oxygen Analyzers, and Ventilator Spirometers. Tracheal suction catheters can introduce microorganisms into a patient's lower respiratory tract. Two types of suction-catheter systems are used in U.S. hospitals: the open single-use catheter system and the closed multi-use catheter system. Studies comparing the two systems have involved low numbers of patients; the results of these studies suggest that the risk for catheter contamination or pneumonia does not differ between patients on whom the single-use suction method is used and those on whom the closed multi-use catheter system is used (305-307). Although advantages of cost and decreased environmental contamination have been attributed to use of the closed-suction system (308,309), larger studies are needed to compare the advantages and disadvantages of both systems (310).

   Reusable resuscitation bags are particularly difficult to clean and dry between uses; microorganisms in secretions or fluid left in the bag may be aerosolized and/or sprayed into the lower respiratory tract of the patient on whom the bag is used; in addition, contaminating microorganisms might be transmitted from one patient to another via hands of HCWs (311-313). Oxygen analyzers and ventilator spirometers have been associated with outbreaks of gram-negative respiratory tract colonization and pneumonia resulting from patient-to-patient transmission of organisms via hands of HCWs (233,245). These devices require either sterilization or high-level disinfection between uses on different patients. Education of physicians, respiratory therapists, and nursing staff regarding the associated risks and appropriate care of these devices is essential.

5. Anesthesia Equipment. The contributory role of anesthesia equipment in outbreaks of nosocomial pneumonia was reported before hospitals implemented routine after-use cleaning and disinfection/sterilization of reusable anesthesia-equipment components that could become contaminated with pathogens during use (314,315).

1. Anesthesia machine. The internal components of anesthesia machines, which include the gas sources and outlets, gas valves, pressure regulators, flowmeters, and vaporizers, are not considered an important source of bacterial contamination of inhaled gases (316). Thus, routine sterilization or high-level disinfection of the internal machinery is unnecessary.

2. Breathing system or patient circuit. The breathing system or patient circuit (including the tracheal tube or face mask, inspiratory and expiratory tubing, y-piece, CO2 absorber and its chamber, anesthesia ventilator bellows and tubing, humidifier, adjustable pressure-limiting valve, and other devices and accessories), through which inhaled and/or exhaled gases flow to and from a patient, can become contaminated with microorganisms that might originate from the patient's oropharynx or trachea. Recommendations for in-use care, maintenance, and reprocessing (i.e., cleaning and disinfection or sterilization) of the components of the breathing system have been published (317,318). In general, reusable components of the breathing system that directly touch the patient's mucous membranes (e.g., face mask or tracheal tube) or become readily contaminated with the patient's respiratory secretions (e.g., y-piece, inspiratory and expiratory tubing, and attached sensors) are cleaned and subjected to high-level disinfection or sterilization between patients. The other parts of the breathing system (e.g., CO2 absorber and its chamber), for which an appropriate and cost-effective schedule of reprocessing has not been firmly determined (319), are changed, cleaned, and sterilized or subjected to high-level disinfection periodically in accordance with published guidelines (317,318) and/or the manufacturers' instructions.

   Using high-efficiency bacterial filters at various positions in the patient circuit (e.g., at the y-piece or on the inspiratory and expiratory sides of the patient circuit) has been advocated (317,320,321) and shown to decrease contamination of the circuit (321-323). However, the use of bacterial filters to prevent nosocomial pulmonary infections has not been proven to be effective and requires additional analysis (324-326).

6. Pulmonary Function Testing Apparatus.

1. Internal parts of pulmonary function testing apparatus. The internal parts of pulmonary function testing apparatus usually are not considered an important source of bacterial contamination of inhaled gas (327). However, because of concern about possible carry-over of bacterial aerosols from an infectious patient-user of the apparatus to the next patient (246,328), placement of bacterial filters (i.e., that remove exhaled bacteria) between the patient and the testing equipment has been advocated (246,329). More studies are needed to evaluate the need for and efficacy of these filters in preventing nosocomial pneumonia (330).

2. Tubing, rebreathing valves, and mouthpieces. Tubing, connectors, rebreathing valves, and mouthpieces could become contaminated with patient secretions during use of the pulmonary function testing apparatus. Thus, these items should be cleaned and subjected to high-level disinfection or sterilization between uses on different patients.

6. Thoracoabdominal Surgical Procedures

   Certain patients are at high risk for developing postoperative pulmonary complications, including pneumonia. These persons include those who are obese or are greater than 70 years of age or who have chronic obstructive pulmonary disease (331-334). Abnormal results from pulmonary function tests (especially decreased maximum expiration flow rate), a history of smoking, the presence of tracheostomy or prolonged intubation, or protein depletion that can cause respiratory-muscle weakness are also risk factors (62,68,136). Patients who undergo surgery of the head, neck, thorax, or abdomen might have impairment of normal swallowing and respiratory clearance mechanisms as a result of instrumentation of the respiratory tract, anesthesia, or increased use of narcotics and sedatives (332,335, 336). Patients who undergo upper abdominal surgery usually have diaphragmatic dysfunction that results in decreased functional residual capacity of the lungs, closure of airways, and atelectasis (337,338).

   Interventions aimed at reducing the postoperative patient's risk for pneumonia have been developed (339). These include deep breathing exercises, chest physiotherapy, use of incentive spirometry, IPPB, and continuous positive airway pressure by face mask (339-349). Studies evaluating the relative efficacy of these modalities reported variable results and were difficult to compare because of differences in outcome variables assessed, patient populations studied, and study design (339,341,342,348-350). Nevertheless, many studies have reported that deep breathing exercises, use of incentive spirometry, and IPPB are advantageous maneuvers, especially in patients who had preoperative pulmonary dysfunction (342,343,345,346,348-350). In addition, control of pain that interferes with cough and deep breathing during the immediate postoperative period decreases the incidence of pulmonary complications after surgery. Several methods of controlling pain have been used; these include both intramuscular or intravenous (including patient-controlled) administration of analgesia and regional (e.g., epidural) analgesia (351-358).

7. Other Prophylactic Measures

1. Vaccination of Patients. Although pneumococci are not a major cause of nosocomial pneumonia, these organisms have been identified as etiologic agents of serious nosocomial pulmonary infection and bacteremia (359-361). The following factors place patients at high risk for complications from pneumococcal infections: age greater than or equal to 65 years of age, chronic cardiovascular or pulmonary disease, diabetes mellitus, alcoholism, cirrhosis, cerebrospinal fluid leaks, immunosuppression, functional or anatomic asplenia, or infection with human immunodeficiency virus (HIV). Pneumococcal vaccine is effective in preventing pneumococcal disease (362,363). Because two thirds or more of patients with serious pneumococcal disease have been hospitalized at least once within the 5 years preceding their pneumococcal illness, offering pneumococcal vaccine in hospitals (e.g., at the time of patient discharge) should contribute substantially to preventing the disease (362,364).

2. Prophylaxis with Systemic Antimicrobial Agents. The systemic administration of antimicrobials is commonly used to prevent nosocomial pneumonia -- especially for patients who are receiving mechanical ventilation, are postoperative, and/or are critically ill (365-367). However, the efficacy of this practice is questionable, and superinfection, which is possible as a result of any antimicrobial therapy, could occur (74,91,366-371).

3. Use of "Kinetic Beds" or Continuous Lateral Rotational Therapy (CLRT) for Immobilized Patients. Use of kinetic beds, or CLRT, is a maneuver for prevention of pulmonary and other complications resulting from prolonged immobilization or bed rest, such as in patients with acute stroke, critical illness, head injury or traction, blunt chest trauma, and/or mechanically assisted ventilation (372-377). This procedure involves the use of a bed that turns continuously and slowly (from less than or equal to 40 for CLRT to greater than or equal to 40 for kinetic therapy) along its longitudinal axis. Among the hypothesized benefits are improved drainage of secretions within the lungs and lower airways, increased tidal volume, and reduction of venous thrombosis with resultant pulmonary embolization (378-381). However, the efficacy of CLRT in preventing pneumonia needs further evaluation because studies have yielded variable results (372-376). In addition, the studies either involved small numbers of patients (373), lacked adequate randomization (372), had no clear definition of pneumonia (372), did not distinguish between community-acquired and nosocomial pneumonia (373,377), or did not adjust for possible confounding factors (e.g., mechanical ventilation, endotracheal intubation, nasogastric intubation, and enteral feeding) (372).

   
   

LEGIONNAIRES DISEASE

1. Epidemiology

   Legionnaires disease is a multisystem illness, with pneumonia, caused by Legionella sp. (382). Since the etiologic agent of Legionnaires disease was identified, numerous nosocomial outbreaks of the disease have been reported, thus enabling researchers to study the epidemiology of epidemic legionellosis. In contrast, the epidemiology of sporadic (i.e., nonoutbreak-related) nosocomial Legionnaires disease has not been well defined. However, when one case is identified, the presence of additional cases should be suspected. Of 196 cases of nosocomial Legionnaires disease reported in England and Wales during 1980-1992, 69% occurred during 22 nosocomial outbreaks (defined as two or more cases occurring at a hospital during a 6-month period) (383). Nine percent of cases occurred greater than 6 months before or after a hospital outbreak, and another 13% occurred in hospitals in which other sporadic cases, but no outbreaks, were identified. Only 9% occurred at institutions in which no outbreaks or additional sporadic cases were identified.

   In North America, the overall proportion of nosocomial pneumonias caused by Legionella sp. has not been determined, although the reported proportions from individual hospitals have ranged from zero to 14% (384-386). Because diagnostic tests for Legionella sp. infection are not performed routinely on all patients who have hospital-acquired pneumonia in most U.S. hospitals, this range probably underestimates the incidence of Legionnaires disease.

   Legionella sp. are commonly found in various natural and man-made aquatic environments (387,388) and may enter hospital water systems in low or undetectable numbers (389,390). Cooling towers, evaporative condensers, heated potable-water-distribution systems within hospitals, and locally produced distilled water can provide a suitable environment for legionellae to multiply. Factors known to enhance colonization and amplification of legionellae in man-made water environments include temperatures of 25-42 C (391-395), stagnation (396), scale and sediment (392), and the presence of certain free-living aquatic amoebae that are capable of supporting intracellular growth of legionellae (397,398).

   A person's risk for acquiring legionellosis after exposure to contaminated water depends on a number of factors, including the type and intensity of exposure and the person's health status (399-401). Persons who are severely immunosuppressed or who have chronic underlying illnesses, such as hematologic malignancy or end-stage renal disease, are at a markedly increased risk for legionellosis (401-404). Persons in the later stages of acquired immunodeficiency syndrome (AIDS) also are probably at increased risk for legionellosis, but data are limited because of infrequent testing of patients (401). Persons who have diabetes mellitus, chronic lung disease, or nonhematologic malignancy; those who smoke cigarettes; and the elderly are at moderately increased risk (382). Nosocomial Legionnaires disease also has been reported among patients in pediatric hospitals (405,406).

   Underlying disease and advanced age are risk factors not only for acquiring Legionnaires disease but also for dying as a result of the illness. In a multivariate analysis of 3,524 cases reported to CDC from 1980 through 1989, immunosuppression, advanced age, end-stage renal disease, cancer, and nosocomial acquisition of disease were each independently associated with a fatal outcome (401). The mortality rate was 40% among 803 persons who had nosocomially acquired cases, compared with 20% among 2,721 persons who had community-acquired cases (401); this difference probably reflected the increased severity of underlying disease in hospitalized patients.

2. Diagnosis

   The clinical spectrum of disease caused by Legionella sp. is broad and ranges from asymptomatic infection to rapidly progressive pneumonia. Legionnaires disease cannot be distinguished clinically or radiographically from pneumonia caused by other agents (407,408), and evidence of infection with other respiratory pathogens does not exclude the possibility of concomitant Legionella sp. infection (409-411).

   The diagnosis of legionellosis may be confirmed by any one of the following: culture isolation of Legionella from respiratory secretions or tissues, microscopic visualization of the bacterium in respiratory secretions or tissue by immunofluorescent microscopy, or, for legionellosis caused by Legionella pneumophila serogroup 1, detection of L. pneumophila serogroup-1 antigens in urine by radioimmunoassay, or observation of a four-fold rise in L. pneumophila serogroup-1 antibody titer to greater than or equal to 1:128 in paired acute and convalescent serum specimens by use of an indirect immunofluorescent antibody (IFA) test (412,413). A single elevated antibody titer does not confirm a case of Legionnaires disease because IFA titers greater than or equal to 1:256 are found in 1%-16% of healthy adults (410, 414-417).

   Because the above tests complement each other, performing each test when Legionnaires disease is suspected increases the probability of confirming the diagnosis (418). However, because none of the laboratory tests is 100% sensitive, the diagnosis of legionellosis is not excluded even if one or more of the tests are negative (413,418). Of the available tests, the most specific is culture isolation of Legionella sp. from any respiratory tract specimen (419,420).

Modes of Transmission

Inhalation of aerosols of water contaminated with Legionella sp. might be the primary mechanism by which these organisms enter a patient's respiratory tract (382). In several hospital outbreaks, patients were considered to be infected through exposure to contaminated aerosols generated by cooling towers, showers, faucets, respiratory therapy equipment, and room-air humidifiers (11,241,258, 421-427). In other studies, aspiration of contaminated potable water or pharyngeal colonizers was proposed as the mode of transmission to certain patients (425,428-430). However, person-to-person transmission has not been observed.

IV. Definition of Nosocomial Legionnaires Disease

The incubation period for Legionnaires disease is usually 2-10 days (431); thus, for the purposes of this document and the accompanying HICPAC recommendations, laboratory-confirmed legionellosis that occurs in a patient who has been hospitalized continuously for greater than or equal to 10 days before the onset of illness is considered a definite case of nosocomial Legionnaires disease, and laboratory-confirmed infection that occurs 2-9 days after hospital admission is a possible case of the disease.

V. Prevention and Control Measures

1. Prevention of Legionnaires Disease in Hospitals with No Identified Cases (Primary Prevention)

   Prevention strategies in health-care facilities in which no cases of nosocomial legionellosis have been identified have differed depending on the immunologic status of the patients, the design and construction of the facility, the resources available for implementing prevention strategies, and state and local regulations.

   At least two strategies are practiced with regard to the most appropriate and cost-effective means of preventing nosocomial legionellosis, especially in hospitals in which no cases or only sporadic cases of the illness have been detected. However, a study comparing the cost-benefit ratios of these strategies has not been conducted.

   The first approach is based on periodic, routine culturing of water samples from the hospital's potable water system for the purpose of detecting Legionella sp. (432,433). When greater than or equal to 30% of the samples obtained are culture-positive for Legionella sp., the hospital's potable water system is decontaminated (433), and diagnostic laboratory tests for legionellosis are made available to clinicians in the hospital's microbiology department so that active surveillance for cases can be implemented (433,434). This approach is based on the premise that no cases of nosocomial legionellosis can occur if Legionella sp. is not present in the potable water system, and, conversely, if Legionella sp. are cultured from the water, cases of nosocomial legionellosis could occur (428,435). Proponents of this strategy indicate that when physicians are informed that the potable water system of the hospital is culture-positive for Legionella sp., they are more inclined to conduct the necessary tests for legionellosis (434). A potential advantage of using this approach in hospitals in which no cases of nosocomial legionellosis have occurred is that routinely culturing a limited number of water samples is less costly than routinely performing laboratory diagnostic testing for all patients who have nosocomial pneumonia.

   The main argument against this approach is that, in the absence of cases, the relationship between the results of water cultures and the risk for legionellosis remains undefined. The bacterium has been frequently present in water systems of buildings (436), often without being associated with known cases of disease (271,385,437, 438). In a study of 84 hospitals in Quebec, 68% of the water systems were found to be colonized with Legionella sp., and 26% were colonized at greater than 30% of sites sampled; however, cases of Legionnaires disease were reported rarely from these hospitals (271). Similarly, at one hospital in which active surveillance for legionellosis and environmental culturing for Legionella sp. were done, no cases of legionellosis occurred in a urology ward during a 3.5-month period when 70% of water samples from the ward were culture-positive for L. pneumophila serogroup 1 (385). Interpretation of the results of routinely culturing the water might be confounded by differing results among the sites sampled within a single water system and by fluctuations in the concentration of Legionella sp. at the same site (439,440). In addition, the risk for illness after exposure to a given source might be influenced by a number of factors other than the presence or concentration of organisms; these factors include the degree to which contaminated water is aerosolized into respirable droplets, the proximity of the infectious aerosol to the potential host, the susceptibility of the host, and the virulence properties of the contaminating strain (441-443). Thus, data are insufficient to assign a level of risk for disease even on the basis of the number of colony-forming units detected in samples from the hospital environment. By routinely culturing water samples, many hospital administrators will have to initiate water-decontamination programs if Legionella sp. are identified. Because of this problem, routine monitoring of water from the hospital's potable water system and from aerosol-producing devices is not widely recommended (444).

   The second approach to preventing and controlling nosocomial legionellosis involves a) maintaining a high index of suspicion for legionellosis and appropriately using diagnostic tests for legionellosis in patients who have nosocomial pneumonia and who are at high risk for developing the disease and dying from the infection (385,445), b) initiating an investigation for a hospital source of Legionella sp. upon identification of one case of definite or two cases of possible nosocomial Legionnaires disease, and c) routinely maintaining cooling towers and using only sterile water for the filling and terminal rinsing of nebulization devices.

   Measures used in hospitals in which cases of nosocomial legionellosis have been identified include either a) routine maintenance of potable water at greater than or equal to 50 C or less than 20 C at the tap or b) chlorination of heated water to achieve 1-2 mg/L of free residual chlorine at the tap, especially in areas where immunosuppressed and other high-risk patients are located (385,428,439,446-449). However, the cost-benefit ratio of such measures in hospitals in which no cases of legionellosis have been identified needs additional evaluation.

2. Prevention of Legionnaires Disease in Hospitals with Identified Cases (Secondary Prevention)

   The indications for a full-scale environmental investigation to search for and subsequently decontaminate identified sources of Legionella sp. in hospital environments have not been clarified, and these indications probably differ depending on the hospital. In hospitals in which as few as one to three nosocomial cases are identified during a period of several months, intensified surveillance for Legionnaires disease has frequently identified numerous additional cases (403,422,425,447). This finding suggests the need for a low threshold for initiating an investigation after laboratory confirmation of cases of nosocomial legionellosis. However, when developing a strategy for responding to such an identification, infection-control personnel should consider the level of risk for nosocomial acquisition of, and mortality from, Legionella sp. infection at their particular hospital.

   An epidemiologic investigation conducted to determine the source of Legionella sp. involves several important steps. First, microbiologic and medical records should be reviewed. Second, active surveillance should be initiated to identify all recent or ongoing cases of legionellosis. Third, potential risk factors for infection (including environmental exposures such as showering or use of respiratory-therapy equipment) should be identified by creating a line listing of cases, analyzing the collected information (by time, place, and person), and comparing case-patients with appropriate controls. Fourth, water samples should be collected from environmental sources implicated by the epidemiologic investigation and from other potential sources of aerosolized water. Fifth, subtype-matching between legionellae isolated from patients and environmental samples should be conducted (427,450-452). This last step can be crucial in supporting epidemiologic evidence of a link between human illness and a specific source (453).

   In some hospitals in which the heated-water system was identified as the source of the organism, the system was decontaminated by pulse (one-time) thermal disinfection or superheating (i.e., flushing each distal outlet of the hot-water system for at least 5 minutes with water at greater than or equal to 65 C) and hyperchlorination (flushing all outlets of the hot-water system with water containing greater than or equal to 10 mg/L of free residual chlorine) (449, 454-456). After either of these procedures, most hospitals either a) maintain heated water at greater than or equal to 50 C or less than 20 C at the tap or b) chlorinate heated water to achieve 1-2 mg/L of free residual chlorine at the tap (385,428,439,446-449). Additional measures (e.g., physical cleaning or replacement of hot-water storage tanks, water-heaters, faucets, and showerheads) may be required because scale and sediment might accumulate in this equipment and protect organisms from the biocidal effects of heat and chlorine (392,449). Alternative methods for controlling and eradicating legionellae in water systems (e.g., treating water with ozone, ultraviolet light, or heavy metal ions) have limited the growth of legionellae under laboratory and/or operating conditions (457-462). However, additional data are needed regarding the efficacy of these methods before they can be considered standard precautions. Measures for decontaminating hospital cooling towers have been published previously (463).

   Additional preventive measures have been used to protect severely immunocompromised patients. At one hospital, immunosuppressed patients were restricted from taking showers, and, for these patients, only sterile water was used for drinking or flushing nasogastric tubes (429). In another hospital, a combined approach consisting of continuous heating, particulate filtration, ultraviolet treatment, and monthly pulse hyperchlorination of the water supply to the bone-marrow transplant unit was used to decrease the incidence of Legionnaires disease (458).

   The decision to search for hospital environmental sources of Legionella sp. and the choice of procedures to use to eradicate such contamination should take into account the type of patient population served by the hospital. Furthermore, decision makers should consider a) the high cost of an environmental investigation and of instituting control measures to eradicate Legionella sp. from sources in the hospital (464,465) and b) the differential risk, based on host factors, for acquiring nosocomial legionellosis and of having severe and fatal infection with the microorganism.

ASPERGILLOSIS

1. Epidemiology

   Aspergillus sp. are ubiquitous fungi that commonly occur in soil, water, and decaying vegetation. Aspergillus sp. have been cultured from unfiltered air, ventilation systems, contaminated dust dislodged during hospital renovation and construction, horizontal surfaces, food, and ornamental plants (466).

   Aspergillus fumigatus and Aspergillus flavus are the most frequently isolated Aspergillus sp. in patients who have laboratory-confirmed aspergillosis (467). Nosocomial aspergillosis has been recognized increasingly as a cause of severe illness and mortality in highly immunocompromised patients (e.g., patients undergoing chemotherapy and/or organ transplantation, including bone-marrow transplantation for hematologic and other malignant neoplasms) (468-472).

   The most important nosocomial infection caused by Aspergillus sp. is pneumonia (473,474). Hospital outbreaks of pulmonary aspergillosis have occurred primarily in granulocytopenic patients, especially those in bone-marrow transplant units (473-480). Although invasive aspergillosis has been reported in recipients of solid-organ (e.g., heart and kidney) transplants (481-485), the incidence of Aspergillus sp. infections in these patients has been lower than in recipients of bone-marrow transplants, probably because granulocytopenia is less severe in solid-organ transplant recipients and the use of corticosteroids, especially in kidney transplant recipients, has decreased with the introduction of cyclosporine (483,486). The efficacy of infection- control measures, such as provision of protected environments and prophylaxis with antifungal agents, in preventing aspergillosis in solid-organ transplant recipients has not been well evaluated (483,484,486,487). In one study of heart-transplant recipients, using only protective isolation of patients did not prevent fungal infections (488).

   The reported attributable mortality from invasive pulmonary aspergillosis has differed depending on the patient population studied. Rates have been as high as 95% in recipients of allogeneic bone-marrow transplants and patients who have aplastic anemia, compared with rates of 13%-80% in leukemic patients (489-491).

2. Pathogenesis

   In contrast to most bacterial pneumonias, the primary route of acquiring Aspergillus sp. infection is by inhalation of the fungal spores. In severely immunocompromised patients, primary Aspergillus sp. pneumonia results from invasion of local lung tissue (467,474,492). Subsequently, the fungus might disseminate via the bloodstream to involve multiple other deep organs (467,474,493). A role for nasopharyngeal colonization with Aspergillus sp., as an intermediate step before invasive pulmonary disease, has been proposed but remains to be elucidated (494-496). Conversely, colonization of the lower respiratory tract by Aspergillus sp. has predisposed patients, especially those with preexisting lung disease (e.g., chronic obstructive lung disease, cystic fibrosis, or inactive tuberculosis), to invasive pulmonary and/or disseminated infection (467,474,497).

Diagnosis

Diagnosing pneumonia caused by Aspergillus sp. is often difficult without performing invasive procedures. Although bronchoalveolar lavage has been a useful screening test (498-500), lung biopsy is still considered the most reliable technique (501). Histopathologic demonstration of tissue invasion by fungal hyphae has been required in addition to isolation of Aspergillus sp. from respiratory tract secretions because the latter, by itself, may indicate colonization (502). However, when Aspergillus sp. is grown from the sputum of a febrile, granulocytopenic patient who has a new pulmonary infiltrate, it is highly likely that the patient has pulmonary aspergillosis (495,503). Routine blood cultures are remarkably insensitive for detecting Aspergillus sp. (504), and systemic antibody responses in immunocompromised patients are probably unreliable indicators of infection (505-507). Antigen-based serologic assays are being developed in an attempt to allow for the rapid and specific diagnosis of Aspergillus sp. infections; however, the clinical usefulness of such assays has not been determined (508,509).

IV. Risk Factors and Control Measures

The primary risk factor for invasive aspergillosis is severe and prolonged granulocytopenia, both disease- and therapy-induced (510). Because bone-marrow-transplant recipients experience the most severe degree of granulocytopenia, they probably constitute the population at highest risk for developing invasive aspergillosis (490,511). The tendency of bone-marrow-transplant recipients to contract severe granulocytopenia (i.e., less than 1,000 polymorphonuclears/uL) is associated with the type of graft they receive. Although both autologous and allogeneic bone-marrow-transplant recipients are severely granulocytopenic for up to 4 weeks after the transplant procedure, acute or chronic graft-versus-host disease also could develop in allogeneic-transplant recipients. The latter might occur up to several months after the procedure, and the disease and/or its therapy (which often includes high doses of corticosteroids, cyclosporine, and other immunosuppressive agents) might result in severe granulocytopenia. Consequently, in developing strategies to prevent invasive Aspergillus sp. infection in bone-marrow-transplant recipients, infection-control personnel should consider exposures of the patient to the fungus both during and subsequent to the immediate post-transplantation period. After hospital discharge, patients (especially allogeneic-transplant recipients) might continue to manifest severe granulocytopenia and, therefore, are susceptible to fungal exposures at home and in ambulatory-care settings. To help address the problem of invasive aspergillosis in bone-marrow-transplant recipients, various studies are in progress to evaluate newer methods of a) enhancing host resistance to invasive fungal (and other) infections and b) eliminating or suppressing respiratory fungal colonization of the upper respiratory tract. These methods include, respectively, the use of granulocyte-colony-stimulating factors and intranasal application of amphotericin B or oral or systemic antifungal drug prophylaxis (466,512-515). For solid-organ transplant recipients, risk factors for invasive aspergillosis have not been studied as extensively. In one study of liver-transplant recipients, risk factors for invasive infection with Aspergillus sp. included preoperative and postoperative receipt of steroids and antimicrobial agents and prolonged duration of transplant surgery (516).

The presence of aspergilli in the hospital environment is the most important extrinsic risk factor for opportunistic invasive Aspergillus sp. infection (517,518). Environmental disturbances caused by construction and/or renovation activities in and around hospitals markedly increase the airborne Aspergillus sp. spore counts in such hospitals and have been associated with nosocomial aspergillosis (476,478,479,519-522). Aspergillosis in immunosuppressed patients also has been associated with other hospital environmental reservoirs. Such reservoirs include contaminated fireproofing material, damp wood, and bird droppings in air ducts (478,523,524).

A single case of nosocomial Aspergillus sp. pneumonia is often difficult to link to a specific environmental exposure. However, additional cases may remain undetected without an active search that includes an intensive retrospective review of microbiologic, histopathologic, and postmortem records; notification of clinicians caring for high-risk patients; and establishment of a system for prospective surveillance for additional cases. When additional cases are detected, the likelihood is increased that a hospital environmental source of Aspergillus sp. can be identified (476,478,519-524). Previous investigations have demonstrated the importance of construction activities and/or fungal contamination of hospital air-handling systems as major sources for outbreaks (473,476,478,519-523). New molecular typing techniques (i.e., karyotyping {525} and DNA endonuclease profiling, which is now available for A. fumigatus {526}) may substantially aid in identifying the source of an outbreak.

Outbreaks of invasive aspergillosis reinforce the importance of maintaining an environment as free as possible of Aspergillus sp. spores for patients who have severe granulocytopenia. To achieve this goal, specialized services in many large hospitals -- particularly bone-marrow transplant services -- have installed "protected environments" for the care of their high-risk, severely granulocytopenic patients and have increased their vigilance during hospital construction and routine maintenance of hospital air- filtration and ventilation systems to prevent exposing high-risk patients to bursts of fungal spores (476,478,519-523,527-532).

Although the exact configuration and specifications of the protected environments might differ between hospitals, such patient-care areas are built to minimize fungal spore counts in air by maintaining a) filtration of incoming air by using central or point-of-use high- efficiency particulate air (HEPA) filters that are capable of removing 99.97% of particles greater than or equal to 0.3 um in diameter; b) directed room airflow (i.e., from intake on one side of the room, across the patient, and out through the exhaust on the opposite side of the room); c) positive room-air pressure relative to the corridor; d) well-sealed rooms; and e) high rates of room-air changes (range: 15 to greater than 400 per hour), although air-change rates at the higher levels might pose problems of patient comfort (473,528-530,532-534). The oldest and most studied protected environment is a room with laminar airflow. Such an environment consists of a bank of HEPA filters along an entire wall of the room; air is pumped by blowers through these filters and into the room at a uniform velocity (90 plus or minus 20 feet/minute), forcing the air to move in a laminar, or at least unidirectional, pattern (535). The air usually exits at the opposite end of the room, and ultra-high air-change rates (i.e., 100-400 air changes per hour) are achieved (473,527). The net effects are essentially sterile air in the room, minimal air turbulence, minimal opportunity for microorganism build-up, and a consistently clean environment (473).

The laminar-airflow system is effective in decreasing or eliminating the risk for nosocomial aspergillosis in high-risk patients (473,528,532,534). However, such a system is costly to install and maintain. Less expensive alternative systems with lower air-change rates (i.e., 10-15 air changes per hour) have been used in some hospitals (529,530,536). However, studies comparing the efficacy of these alternative systems with laminar-airflow rooms in eliminating Aspergillus sp. spores and preventing nosocomial aspergillosis are limited. One hospital that employed cross-flow ventilation, point-of-use HEPA filters, and 15 air changes per hour reported that cases of nosocomial aspergillosis had occurred in patients housed in these rooms, although this rate was low (i.e., 3.4%) (530,536). However, these infections had been caused by A. flavus, a species that was not cultured from the room air, suggesting that the patients were probably exposed to fungal spores when they were allowed outside their rooms (530).

Copper-8-quinolinolate was used on environmental surfaces contaminated with Aspergillus sp. to control one reported outbreak of aspergillosis (537), and it has been incorporated in the fireproofing material of a newly constructed hospital to help decrease the environmental spore burden (530); however, its general applicability has not been established.

VIRAL PNEUMONIAS

Viruses can be an important and often unappreciated cause of nosocomial pneumonia (538-540). In one prospective study of endemic nosocomial infections, approximately 20% of pneumonia cases resulted from viral infections (539). Although the early diagnosis and treatment of viral pneumonia infections have been possible in recent years (541-544), many hospitalized patients remain at high risk for developing severe and sometimes fatal viral pneumonia (538,545-552). These data and reports of well-documented outbreaks involving nosocomial viral transmission (553-556) indicate that measures to prevent viral transmission should be instituted.

Nosocomial respiratory viral infections a) usually follow community outbreaks that occur during a particular period every year (555,557-560), b) confer only short-term immunity (561), c) affect both healthy and ill persons (547,548,554,562-564), and d) have exogenous sources. A number of viruses -- including adenoviruses, influenza virus, measles virus, parainfluenza viruses, RSV, rhinoviruses, and varicella-zoster virus -- can cause nosocomial pneumonia (548,555,556,565-571,572); however, adenoviruses, influenza viruses, parainfluenza viruses, and RSV reportedly have accounted for most (70%) nosocomial pneumonias caused by viruses (573).

Influenza and RSV infections contribute substantially to the morbidity and mortality associated with viral pneumonia, and the epidemiology of both viral infections has been well researched; for these reasons, this section concerning viral pneumonias focuses on the principles of, and approaches to, the control of these two types of infection. Recommendations for preventing nosocomial pneumonia caused by infection with other viral pathogens were published previously (224).

RSV INFECTION

1. Epidemiology

   RSV infection is most common during infancy and early childhood, but it can also occur in adults (562,565,574,575). Infection usually causes mild or moderately severe upper respiratory illness. However, both life-threatening pneumonia and bronchiolitis have occurred in immunocompromised patients, the elderly, and children who have chronic cardiac and pulmonary disease (547,549,564,565, 576,577).

   Recent surveillance of 10 U.S. hospital laboratories in which cultures for RSV are performed suggests that community outbreaks occur on a seasonal basis from December through March; these outbreaks last 3-5 months and are associated with an increased number of hospitalizations and deaths among infants and young children (578). During community outbreaks of RSV, children who have respiratory symptoms at the time of hospital admission are often reservoirs for RSV (553,555).

2. Diagnosis

   The clinical characteristics of RSV infection, especially in neonates, are often indistinguishable from those of other viral respiratory tract infections (565,566). Culture of RSV from respiratory secretions is the standard for diagnosis. Rapid antigen-detection kits that use direct immunofluorescence or enzyme-linked immunosorbent assays can provide results within hours. The benefit of using these tests to identify infected patients depends on the sensitivity and specificity of the test. The reported sensitivity and specificity of RSV enzyme immunoassays vary between 80% and 95% and may be even lower in actual practice (579-582). In general, once laboratory-confirmed cases of RSV infection are identified in a hospital, a presumptive diagnosis of RSV infection in subsequent cases with manifestations suggestive of RSV infection may be acceptable for infection-control purposes.

Modes of Transmission

RSV is present in large numbers in the respiratory secretions of symptomatic persons infected with the virus, and it can be transmitted directly via large droplets during close contact with such persons or indirectly via RSV-contaminated hands or fomites (553,583,584). The portal of entry is usually the conjunctiva or the nasal mucosa (585). Inoculation by RSV-contaminated hands is the usual way of depositing the virus onto the eyes or nose (553,583-585). Hands can become contaminated by handling either the respiratory secretions of infected persons or contaminated fomites (583,584).

In nosocomial RSV outbreaks for which the viral isolates were typed, more than one strain of RSV often was identified (554,563,586), suggesting multiple sources of the virus. Potential sources include patients, HCWs, and visitors. Because infected infants shed large amounts of virus in their respiratory secretions and can easily contaminate their immediate surroundings, they are a major reservoir for RSV (587). HCWs might become infected after exposure in the community (588) or in the hospital and subsequently transmit infection to patients, other HCWs, or visitors (566,589).

IV. Control Measures

Different combinations of control measures, ranging from the simple to the complex, have been effective in varying degrees in preventing and controlling nosocomial RSV infection (226,589-596). Successful programs have shared two common elements: implementation of contact-isolation precautions and compliance with these precautions by HCWs. In theory, strict compliance with handwashing recommendations could prevent most nosocomial RSV infections; however, studies have indicated that such compliance among HCWs is poor (221,222). Thus, other preventive measures are usually necessary to prevent RSV infection.

The wearing of gloves and gowns has been associated with decreased incidence of nosocomial RSV (226). The wearing of gloves has helped decrease transmission of RSV, probably because the gloves remind HCWs to comply with handwashing and other precautions and deter them from touching their eyes or nose. However, the benefits derived from wearing gloves are offset if the gloves are not changed after contact with an infected patient or with contaminated fomites and if hands are not washed adequately after glove removal (229). The wearing of both gloves and gowns during contact with RSV-infected infants or their immediate environment has been successful in preventing infection (226). In addition, the use of eye-nose goggles rather than masks has protected HCWs from infection; however, eye-nose goggles are not widely available and are inconvenient to wear (593,597).

Additional measures may be indicated to control ongoing nosocomial transmission of RSV or to prevent transmission to patients at high risk for serious complications resulting from the infection (e.g., patients whose cardiac, pulmonary, or immune systems are compromised). The following additional control measures have been used in various combinations: a) using private rooms for infected patients OR cohorting infected patients, with or without preadmission screening by rapid laboratory diagnostic tests; b) cohorting HCWs; c) excluding HCWs who have symptoms of upper respiratory tract infection from caring for uninfected patients at high risk for severe or fatal RSV infection (e.g., infants); d) limiting visitors; and e) postponing admission of patients at high risk for complications from RSV infection (224,590, 592,594,596). Although the exact role of each of these measures has not been determined, their use for controlling RSV outbreaks seems prudent.

INFLUENZA

1. Epidemiology

   Pneumonia that occurs in patients who have influenza can be caused by the influenza virus, a secondary bacterial infection, or a combination of both (598-600). Influenza-associated pneumonia can occur in any person but is more common in infants and young children, in persons greater than 65 years of age, and in persons of any age who are immunosuppressed or have certain chronic medical conditions (e.g., severe underlying heart or lung disease) (575,601-603).

   Influenza typically occurs on a seasonal basis during December-April; during this period, peak influenza activity in an affected community usually lasts 6-8 weeks (604,605). Nosocomial outbreaks can occur in a community affected by an influenza epidemic; these outbreaks are often characterized by abrupt onset and rapid transmission (606-608). Most reported institutional outbreaks of influenza have occurred in nursing homes; however, hospital outbreaks in pediatric and chronic-care wards and in medical and neonatal intensive-care units have been reported (556,609-612).

   Influenza is believed to be spread from person to person by a) direct inhalation of droplet nuclei or small-particle aerosols or b) direct deposition of virus-laden large droplets onto the mucosal surfaces of the upper respiratory tract of a person during close contact with an infected person (613-616). The extent to which transmission might occur by contact with virus-contaminated hands or fomites is unknown; however, such contact is not the primary mode of transmission (617).

   The most important reservoirs of influenza virus are infected persons. Although the period of greatest communicability is during the first 3 days of illness, the virus can be shed both before the onset of symptoms and for greater than or equal to 7 days afterward (556,604, 618).

2. Diagnosis

   Influenza is clinically indistinguishable from other febrile respiratory illnesses; however, during outbreaks with laboratory- confirmed cases, a presumptive diagnosis of the infection can be made for illnesses that have similar manifestations (619). Historically, diagnosis of influenza was made by virus isolation from nasopharyngeal secretions or by serologic conversion, but recently developed rapid diagnostic tests that are similar to culture in sensitivity and specificity now enable early diagnosis and treatment of cases and provide a basis for prompt initiation of antiviral prophylaxis as part of outbreak control (620-625).

Prevention and Control of Influenza

The most effective measure for reducing the impact of influenza is the vaccination of persons at high risk for complications of the infection before the influenza season begins each year. High-risk persons include persons 6 months-18 years of age who are receiving long-term aspirin therapy and persons who either a) are greater than or equal to 65 years of age; b) are in long-term-care units; or c) have either chronic disorders of the pulmonary or cardiovascular systems, diabetes mellitus, renal dysfunction, hemoglobinopathies, or immunosuppression (611,626-628). Patients who have musculoskeletal disorders that impede adequate respiration also may be at high risk for complications resulting from influenza. When high vaccination rates are achieved in closed or semi-closed settings, the risk for outbreaks is reduced because of induction of herd immunity (629,630).

When an institutional outbreak is caused by influenza type A, antiviral agents can be used both for treatment of ill persons and as prophylaxis for others (631). Two related antiviral agents, amantadine hydrochloride and rimantadine hydrochloride, are effective against influenza type A but not against influenza type B (543,632-634). These agents can be used in the following ways to prevent illness caused by influenza A virus: a) as short-term prophylaxis for high-risk persons after late vaccination; b) as prophylaxis for persons for whom vaccination is contraindicated; c) as prophylaxis for immunocompromised persons who might not produce protective levels of antibody in response to vaccination; d) as prophylaxis for unvaccinated HCWs who provide care to patients at high risk for infection, either for the duration of influenza activity in the community or until immunity develops after vaccination; and e) as prophylaxis when vaccine strains do not closely match the epidemic virus strain (631).

Amantadine has been available in the United States for many years; rimantadine has been approved for use since 1993. Both drugs protect against all naturally occurring strains of influenza A virus; thus, antigenic changes in the virus that might reduce vaccine efficacy do not alter the effectiveness of amantadine or rimantadine. Both drugs are 70%-90% effective in preventing illness if administered before exposure to influenza A virus (632,635). In addition, they can reduce the severity and duration of illness caused by influenza A virus if administered within 24-48 hours after onset of symptoms (636,637). These drugs can limit nosocomial spread of influenza type A if they are administered to all or most patients when influenza type A illnesses begin in a facility (609,638,639).

Compared with rimantadine, amantadine has been associated with a higher incidence of adverse central nervous system (CNS) reactions (e.g., mild and transitory nervousness, insomnia, impaired concentration, mood changes, and lightheadedness). These symptoms have been reported in 5%-10% of healthy young adults receiving 200 mg of amantadine per day (543,632). In the elderly, CNS side effects may be more severe; in addition, dizziness and ataxia occur more frequently among persons in this age group than among younger persons (640,641). Dose reductions of both amantadine and rimantadine are recommended for certain patients, such as persons greater than or equal to 65 years of age and/or those who have renal insufficiency. The drug package inserts for amantadine and rimantadine contain important information regarding administration of these drugs. Guidelines for the use of amantadine and rimantadine and considerations for the selection of these drugs were published previously by the Advisory Committee on Immunization Practices (ACIP) (631).

The emergence of amantadine- and rimantadine-resistant strains of influenza A virus has been observed in persons who have received these drugs for treatment of the infection (642,643). Because of the potential risk for transmitting resistant viral strains to contacts of persons receiving amantadine or rimantadine for treatment (643,644), infected persons taking either drug should avoid, as much as possible, contact with others during treatment and for 2 days after discontinuing treatment (644,645). This is particularly important if the contacts are uninfected high-risk persons (644,646).

The primary focus of efforts to prevent and control nosocomial influenza is the vaccination of high-risk patients and HCWs before the influenza season begins (628,647,648). The decision to use amantadine or rimantadine as an adjunct to vaccination in the prevention and control of nosocomial influenza is based partially on results of virologic and epidemiologic surveillance in the hospital and the community. When outbreaks of influenza type A occur in a hospital, and antiviral prophylaxis of high-risk persons and/or treatment of cases is undertaken, administration of amantadine or rimantadine should begin as early in the outbreak as possible to reduce transmission (609,638,631).

Measures other than vaccination and chemoprophylaxis have been recommended for controlling nosocomial influenza outbreaks. Because influenza can be transmitted during contact with an infected person, the following procedures have been recommended: observing contact- isolation precautions, placing patients who have symptoms of influenza in private rooms, cohorting patients who have influenza-like illness, and wearing a mask when entering a room in which a person who has suspected or confirmed influenza is housed (224). Handwashing and the wearing of gloves and gowns by HCWs during the patient's symptomatic period also have been recommended; however, the exact role of these measures in preventing influenza transmission has not been determined (224,608,649). Although influenza can be transmitted via the airborne route, the efficacy of placing infected persons in rooms that have negative air pressure in relation to their immediate environment has not been assessed. In addition, this measure may be impractical during institutional outbreaks that occur during a community epidemic of influenza because many HCWs and newly admitted patients could be infected with the virus; thus, the hospital would face the logistical problem of accommodating all ill persons in rooms that have special ventilation. Although the effectiveness of the following measures has not been determined, their implementation could be considered during severe outbreaks: a) curtailment or elimination of elective admissions, both medical and surgical; b) restriction of cardiovascular and pulmonary surgery; c) restriction of hospital visitors, especially those who have acute respiratory illnesses; and d) restriction of HCWs who have an acute respiratory illness from the workplace (649).

Part II. Recommendations for Preventing Nosocomial Pneumonia

INTRODUCTION

These recommendations are presented in the following order based on the etiology of the infection: bacterial pneumonia, including Legionnaires disease; fungal pneumonia (i.e., aspergillosis); and virus-associated pneumonia (i.e., RSV and influenza infections). Each topic is subdivided according to the following general approaches for nosocomial infection control:

1. Staff education and infection surveillance;

2. Interruption of transmission of microorganisms by eradicating infecting microorganisms from their epidemiologically important reservoirs and/or preventing person-to-person transmission; and

3. Modifying host risk for infection.

As in previous CDC guidelines, each recommendation is categorized on the basis of existing scientific evidence, theoretical rationale, applicability, and economic impact (224,225,650-654). However, the previous CDC system of categorizing recommendations has been modified as follows:

CATEGORY IA Strongly recommended for all hospitals and strongly

supported by well-designed experimental or epidemiologic studies.

CATEGORY IB Strongly recommended for all hospitals and viewed as

effective by experts in the field and a consensus of HICPAC. These recommendations are based on strong rationale and suggestive evidence, even though definitive scientific studies may not have been done.

CATEGORY II Suggested for implementation in many hospitals. These

recommendations may be supported by suggestive clinical or epidemiologic studies, a strong theoretical rationale, or definitive studies applicable to some but not all hospitals.

NO RECOMMENDATION; Practices for which insufficient evidence or UNRESOLVED ISSUE consensus regarding efficacy exists.

BACTERIAL PNEUMONIA

1. Staff Education and Infection Surveillance

1. Staff education

   Educate HCWs regarding nosocomial bacterial pneumonias and infection-control procedures used to prevent these pneumonias (655-661). CATEGORY IA

2. Surveillance

1. Conduct surveillance of bacterial pneumonia among ICU patients at high risk for nosocomial bacterial pneumonia (e.g., patients receiving mechanically assisted ventilation and selected postoperative patients) to determine trends and identify potential problems (6,34,35,62,63,662-664). Include data regarding the causative microorganisms and their antimicrobial susceptibility patterns (2,3). Express data as rates (e.g., number of infected patients or infections per 100 ICU days or per 1,000 ventilator-days) to facilitate intrahospital comparisons and determination of trends (66,665-667). CATEGORY IA

2. Do not routinely perform surveillance cultures of patients or of equipment or devices used for respiratory therapy, pulmonary- function testing, or delivery of inhalation anesthesia (65,668, 669). CATEGORY IA

2. Interrupting Transmission of Microorganisms

1. Sterilization or disinfection and maintenance of equipment and devices

1. General measures

1. Thoroughly clean all equipment and devices before sterilization or disinfection (266,267,670). CATEGORY IA

2. Sterilize or use high-level disinfection for semicritical equipment or devices (i.e., items that come into direct or indirect contact with mucous membranes of the lower respiratory tract) (Appendix A). High-level disinfection can be achieved either by wet heat pasteurization at 76 C for 30 minutes or by using liquid chemical disinfectants approved as sterilants/ disinfectants by the Environmental Protection Agency and cleared for marketing for use on medical instruments by the Office of Device Evaluation, Center for Devices and Radiologic Health, Food and Drug Administration (260,262,264,267,671). Follow disinfection with appropriate rinsing, drying, and packaging, taking care not to contaminate the items in the process. CATEGORY IB

3. (1) Use sterile (not distilled, nonsterile) water for rinsing

   reusable semicritical equipment and devices used on the respiratory tract after they have been disinfected chemically (241,249,250,258,269). CATEGORY IB

   (2) No Recommendation for using tap water (as an alternative

   to sterile water) to rinse reusable semicritical equipment and devices used on the respiratory tract after such items have been subjected to high-level disinfection, regardless of whether rinsing is followed by drying with or without the use of alcohol (241,249,250,258,269,273,277). UNRESOLVED ISSUE

4. Do not reprocess equipment or devices that are manufactured for a single use only, unless data indicate that reprocessing such items poses no threat to the patient, is cost-effective, and does not change the structural integrity or function of the equipment or device (672,673). CATEGORY IB

2. Mechanical ventilators, breathing circuits, humidifiers, and nebulizers

1. Mechanical ventilators

   Do not routinely sterilize or disinfect the internal machinery of mechanical ventilators (126,128,674). CATEGORY IA

2. Ventilator circuits with humidifiers

   (1) Do not routinely change more frequently than every 48

   hours the breathing circuit, including tubing and exhalation valve, and the attached bubbling or wick humidifier of a ventilator that is being used on an individual patient (34,283,288). CATEGORY IA

   (2) No Recommendation for the maximum length of time after

   which the breathing circuit and the attached bubbling or wick humidifier of a ventilator being used on a patient should be changed (289). UNRESOLVED ISSUE

   (3) Sterilize reusable breathing circuits and bubbling or

   wick humidifiers or subject them to high-level disinfection between their uses on different patients (259,260,262,264,267). CATEGORY IB

   (4) Periodically drain and discard any condensate that

   collects in the tubing of a mechanical ventilator, taking precautions not to allow condensate to drain toward the patient. Wash hands after performing the procedure or handling the fluid (215,282,286). CATEGORY IB

   (5) No Recommendation for placing a filter or trap at the

   distal end of the expiratory-phase tubing of the breathing circuit to collect condensate (247,282). UNRESOLVED ISSUE

   (6) Do not place bacterial filters between the humidifier

   reservoir and the inspiratory-phase tubing of the breathing circuit of a mechanical ventilator. CATEGORY IB

   (7) Humidifier fluids

   (a) Use sterile water to fill bubbling humidifiers (132,

   241,249,250,286). CATEGORY II

   (b) Use sterile, distilled, or tap water to fill wick

   humidifiers (249, 250,286). CATEGORY II

   (c) No Recommendation for preferential use of a closed,

   continuous-feed humidification system. UNRESOLVED ISSUE

3. Ventilator breathing circuits with hygroscopic condenser-humidifiers or heat-moisture exchangers

   (1) No Recommendation for preferential use of hygroscopic

   condenser-humidifier or heat-moisture exchanger rather