Methods

Top Methods Author & Article Info References

We identified all relevant English-language articles published between 1966 and July 1992 through prominent review articles and a MEDLINE search (keywords: endoscopy, bronchoscopy, infections, transmission, and disinfection). We also manually searched bibliographies of identified articles to find additional sources. The entire search yielded 265 articles, all of which were reviewed in depth.

General Considerations

Flexible endoscopy is a common clinical procedure; an estimated 8.7 million gastrointestinal and 580 000 pulmonary flexible endoscopies were done in the United States in 1989 [1]. The risk for transmitting infections via these procedures depends on three factors: exposure of the endoscope to microorganisms, cleaning and disinfection procedures, and instrument design. Depending on the origin of the contaminating microorganisms, transmission of infection can be categorized as either patient-to-patient or environment-to-patient (Figure 1). During a procedure, an endoscope can be contaminated with whatever organisms are contained in patient secretions. Environmental contamination typically results from flushing or cleaning the endoscope with contaminated solutions. Whether contamination of the endoscope persists depends on the quantity and nature of the microorganisms. Some microbes are inherently more resistant to disinfectants Figure 2, the efficacy of which depends on type, concentration, and duration of exposure. If patient material, such as blood, feces, or secretions, remains on or in the endoscope after cleaning, the effectiveness of subsequent disinfection diminishes.

View larger version (18K): [in this window] [in a new window]   Figure 1. Nosocomial transmission of microorganisms via endoscopes. Flow diagram shows major routes and sources of nosocomial transmission of microorganisms via endoscopes.

View larger version (30K): [in this window] [in a new window]   Figure 2. Resistance of microorganisms to disinfectants. Descending order of in-vitro resistance to germicidal chemicals. Modified from reference 2; reproduced with the permission of the author and publisher. Most Resistant Least Resistant Most Resistant Least Resistant.

Because endoscopes are made of fragile, heat-sensitive materials, they are routinely decontaminated by high-level disinfection, not sterilization [2]. Gastrointestinal endoscopes, with their multiple internal channels and valves Figure 3, are more complex than the single-channeled bronchoscopes. In general, the more complex the instrument, the more crevices, joints, or surface pores there are and, hence, the more problematic cleaning and disinfection becomes [2].

View larger version (91K): [in this window] [in a new window]   Figure 3. Cross-section of typical gastrointestinal endoscope. A schematic drawing of a fiberoptic gastrointestinal endoscope shows the complex internal design.

Transmission of Specific Microorganisms

Salmonella Infections

Salmonella infections occur frequently in the United States; for example, more than 41 000 culture-positive patients were reported to the Centers for Disease Control (CDC) in 1989[3]. A chronic asymptomatic carrier state, defined by the continued fecal excretion of Salmonella organisms for more than 1 year, develops in approximately 3% of persons after S. typhi infection (typhoid fever) [4] and in fewer than 1% of persons after nontyphoidal Salmonella infection [5]. Both acutely infected persons and chronic carriers are thus potential sources of endoscopic contamination. Many disinfectants, including glutaraldehyde, phenolics, and iodophors, effectively kill salmonellae [2].

Transmission of Salmonella infections by endoscopy has occurred with many serotypes, including S. agona, S. kedougou, S. newport, S. oranienburg, S. oslo, S. typhi, and S. typhimurium [6-14]. Of the 84 patients reported to have developed such infections, 6 patients developed septicemia and 1 patient died. In most cases, the disinfectant (hexachlorophene, cetrimide, chlorhexidine, or quaternary ammonium compounds) used to clean the endoscopes had relatively little microbiocidal activity against salmonellae. In one outbreak, investigators identified inadequately disinfected colonic biopsy forceps as the source of infection [9].

Pseudomonas Infections

Pseudomonas aeruginosa flourishes in warm, damp environments. Typical environmental reservoirs include respiratory equipment, sinks, and water bottles [15]. Most acute P. aeruginosa infections, which often involve the lungs, are nosocomially acquired. Among healthy adults, P. aeruginosa can colonize many body sites, as evidenced by isolation from throat (0% to 7%), sputum (2%), and stool (3% to 24%). Hospitalized patients, as well as patients with certain chronic lung diseases, have higher colonization rates [16]. Potential sources of endoscope contamination with Pseudomonas species thus include environmental reservoirs, acutely infected patients, and colonized patients. Like Salmonella species, P. aeruginosa is susceptible to glutaraldehyde, phenolics, and iodophors [2].

Most P. aeruginosa infections transmitted by endoscopy occurred after endoscopic retrograde cholangiopancreatography and resulted from environment-to-patient transfer of organisms. In all cases, the investigators isolated P. aeruginosa from some part of the endoscope. These infections resulted in bacteremia in 45 patients, of whom 4 died [17-26]. Most infections were caused by the use of an inadequate disinfectant [21, 22], contamination of an inner channel [18, 20], or incomplete drying of the endoscope channels before overnight storage [23, 26]. One epidemic, however, occurred despite the use of glutaraldehyde after each procedure and ceased only after replacement of the endoscope [19].

The first reports of bronchoscopic transmission of P. aeruginosa involved patients who developed Pseudomonas pneumonia [27, 28]. In a subsequent report, investigators cultured P. aeruginosa from the bronchoscopic washings of 11 patients; 1 patient, who was immunosuppressed, developed severe pneumonia [29]. The investigators isolated P. aeruginosa from the aspiration-irrigation channel of the bronchoscope, and the outbreak continued until they sterilized the bronchoscope with ethylene oxide.

Mycobacteria

Approximately 22 000 new cases of active tuberculosis occur in the United States each year [30], and an estimated 10 to 15 million persons carry Mycobacterium tuberculosis in its dormant phase [31]. After many years of decline, the incidence of tuberculosis in the United States began increasing dramatically in 1986 [30], predominantly because of an increasing number of cases in HIV-infected patients [32]. Infections with mycobacteria commonly found in the environment, such as M. avium-intracellulare complex, M. chelonae, and M. fortuitum, have also recently increased [33]. Unfortunately, studies determining the sensitivity of mycobacteria to various disinfectants are conflicting. In general, cetrimide, chlorhexidine, and iodophors are considered unreliable. Glutaraldehyde is widely accepted as a mycobactericidal agent, but the time required for disinfection remains undefined [34-36].

In the first reports of the bronchoscopic transmission of M. tuberculosis, the investigators, who disinfected the bronchoscopes with iodophor, advocated using a more effective disinfectant such as glutaraldehyde [37, 38]. In three subsequent cases, the patients developed clinically apparent M. tuberculosis infection despite rigorous cleaning and disinfection of the bronchoscope; the suction valve, with its spring-operated sleeve, seemed the most likely source of contamination [39]. The investigators tested this hypothesis by contaminating bronchoscopes with M. fortuitum; after they routinely cleaned and disinfected the instrument, M. fortuitum remained in all valves.

Among other mycobacteria, M. chelonae has most commonly been associated with endoscopic transmission. In a large outbreak, M. chelonae was isolated from bronchial washings, brushings, or sputum in 72 patients [40]. Two patients developed clinical disease and one patient died. After recognizing the outbreak, the investigators changed from glutaraldehyde disinfection of the bronchoscopes to ethylene oxide sterilization. Nevertheless, they continued to isolate M. chelonae from clinical specimens until they discovered punctured suction channels in two of the bronchoscopes; M. chelonae was isolated from slimy material in the interior of both instruments.

Hepatitis B Virus

On average, more than 300 000 cases of primary hepatitis B virus (HBV) occur each year in the United States [41], and approximately 5% to 10% of patients develop persistent HBV infection. The estimated number of chronic HBV carriers in the United States ranges from 750 000 to 1 000 000 [42]. In infected persons, hepatitis B surface antigen (HBsAg) has been found in various body fluids, including serum, feces, bile, and saliva [43].

The inability to culture HBV has limited the evaluations of its environmental stability and its susceptibility to disinfectants. Alternative approaches include measuring the presence of HBsAg or inoculating chimpanzees. A study in chimpanzees showed that HBV-contaminated inanimate objects, if not properly cleaned and disinfected, can harbor and transmit the virus for up to 1 week [44]. Most of the commonly used disinfectant and sterilization procedures, however, inactivate HBV [45, 46].

Two studies have shown the potential for the endoscopic transmission of HBV. In one study, iodophor-isopropyl alcohol removed HBsAg from surfaces of endoscopes used in four HBsAg-positive patients, but not from the cytology brushes or biopsy forceps [47]. In the other study, an endoscope and a biopsy forceps were immersed for 15 minutes in gastric juice containing 1.0% serum and Iodine-125-HBsAg, and, despite subsequent disinfection with chlorhexidine and cetrimide for 15 to 20 minutes, they were both positive for Iodine-125-HBsAg [48].

One group of investigators documented the endoscopic transmission of HBV [49]. After performing endoscopy in an HBsAg- and HBeAg (hepatitis B e antigen)-positive 50-year-old patient, the investigators disinfected the endoscope with glutaraldehyde for 21 hours, cleaned the channels with water and air, and wiped the controls and connecting tube with 70% ethyl alcohol. Subsequently, they used the same endoscope on a 78-year-old woman with gastrointestinal hemorrhage; 96 days later, the woman developed jaundice and positive serologic responses for HBsAg and HBeAg. The investigators hypothesized that the air and water channels, which were cleaned with water but not liquid disinfectant, were the source of contamination.

In contrast to this isolated case of HBV transmission via endoscopy, the absence of endoscopic transmission of HBV has been described in 12 reports [48, 50-60]. Among 394 patients who were exposed to endoscopes that had recently been used on HBsAg-positive patients, none developed clinical hepatitis and 357 (91%) had serologic confirmation that they had not contracted HBV infection.

Human Immunodeficiency Virus

Currently, the CDC estimates that more than 1 000 000 people in the United States are infected with human immunodeficiency virus (HIV), corresponding to an HIV prevalence rate of 0.4% to 0.6% in the general population [61]. Among selected groups receiving medical care, however, HIV prevalence rates range from 5% to 11% [62-65]. In infected patients, HIV can be isolated from most body fluids [66], including blood, saliva, alveolar fluid, and from tissues such as lung [67] and rectal mucosa [68]. To date, direct proof of transmission exists only with blood and semen [69]. Most routinely used disinfectants inactivate HIV, even at concentrations below those present in commercial preparations [70].

Four studies have addressed the disinfection of HIV-contaminated endoscopes. In one, the investigators inoculated suspensions of high-titer HIV into the air and instrument channels of two endoscopes for 20 minutes, then rinsed one with isotonic saline and the other with 2% glutaraldehyde at 60 °C for 30 minutes [71]. Human immunodeficiency virus was recovered from the endoscope rinsed with isotonic saline but not from the one disinfected with glutaraldehyde. A different group did upper gastrointestinal endoscopy and biopsy procedures in patients with AIDS and found HIV on 7 of 20 unwashed endoscopes but on none of 20 endoscopes disinfected with 2% glutaraldehyde for 2 minutes [72]. Subsequently, the investigators contaminated endoscope suction-biopsy channels with high-titer HIV suspensions and determined that disinfection with 2% alkaline glutaraldehyde for 2 minutes successfully decontaminated the endoscopes [73]. In addition, they contaminated the air and water channels of an endoscope with HIV and found no HIV after disinfecting it with glutaraldehyde for 10 minutes. In a recent study, bronchoscopes used in patients with AIDS were tested for the presence of HIV using polymerase chain reaction techniques and culture infectivity assays [74]. All seven bronchoscopes sampled immediately after use (before cleaning) showed the presence of HIV by polymerase chain reaction, although conventional infectivity and antigen assays were negative in all instances. Cleaning in detergent, however, effectively removed HIV from the contaminated instruments. To date, the endoscopic transmission of HIV has not been reported.

Other Microorganisms

Other organisms transmitted by endoscopic procedures include Pseudomonas pseudomallei [75] and Serratia marcescens [76] after bronchoscopy, Staphylococcus epidermidis [77] and Enterobacter aerogenes [25] after endoscopic retrograde cholangiopancreatography, and Trichosporon beigelli [78] and Strongyloides stercoralis after upper endoscopy [79]. Four cases of documented endoscopic transmission of Helicobacter pylori have been reported [80, 81].

Cases of endoscopic transmission of hepatitis C virus, cytomegalovirus, herpes simplex virus, or human T-cell lymphotropic virus type I and type II have not been documented.

Surveys and Studies of Infectious Complications

Two large surveys in the 1970s suggested that infectious complications of gastrointestinal endoscopy were rare [82, 83]. In a later survey, investigators found that infectious complications occurred in 97 of 10 425 endoscopic retrograde cholangiopancreatography procedures (1.1%). Infectious complications included cholangitic sepsis and pancreatic sepsis and occurred mainly in patients with obstructed ducts; 13 patients died [84]. The cause of these infections was not determined, but the investigators speculated that contamination of endoscopes was a possible source. Moreover, cultures of cannulas inserted through meticulously cleaned endoscopes grew various microorganisms. In a recent survey of gastroenterology nurses in the United States, 116 of 2030 respondents (6%) acknowledged that infections transmitted by endoscopy had occurred at their institution [14]. The investigators did not clarify which, if any, of these infections had been reported in the literature. The infectious agents included Pseudomonas species, Salmonella species, Serratia species, Campylobacter species, Staphylococcus aureus, Mycobacterium species, Clostridium difficile, HBV, and the Creutzfeldt-Jakob agent. In a separate study that involved a retrospective analysis, investigators detected transmission of H. pylori in 3 of 281 (1.1%) endoscopic examinations [80]. In a prospective study, 97 patients had blood cultures taken before and 5 minutes after endoscopic retrograde cholangiopancreatography; in addition, aspirates were obtained from the pancreatic or biliary ducts after injection of contrast material [85]. All blood cultures were negative, but 14 patients had positive aspirate cultures with an identical strain of P. aeruginosa; the investigators isolated the same strain of P. aeruginosa from the inner channel of the endoscope.

In a retrospective survey of 24 521 bronchoscopies, only two cases of pneumonia were identified [86]. In a subsequent prospective study of 100 fiberoptic bronchoscopies, complications after the procedure included fever in 16% and parenchymal infiltrates in 6%; the infiltrates were usually transient, but in one instance a patient died of rapidly progressing pneumonia [87]. One group prospectively evaluated 43 patients after bronchoscopy and found that no patients developed fever (or other evidence of infection) within 24 hours of the procedure [88]. More recently, investigators identified bronchoscopy as an independent risk factor for nosocomial pneumonia [89]. In all of these reports, the source of the organism causing the pneumonia was not identified.

Cleaning and Disinfection Guidelines and Practices

In 1978, an ad-hoc committee formed by the Association for Practitioners in Infection Control set forth general guidelines regarding the decontamination of fiberoptic endoscopes [90]. The committee emphasized the need for meticulous mechanical cleaning of the endoscope and its channels, followed by high-level disinfection and rinsing. In addition, they recommended sterilizing accessory instruments, such as biopsy forceps and cytology brushes, after each use. The CDC developed similar guidelines in the early 1980s [91]. In 1985, in response to the increasing prevalence of HIV infection, the CDC initiated recommendations for universal precautions [92]. Applying universal precautions to fiberoptic endoscopes, the CDC recommended standardized cleaning and disinfection after every patient because these instruments were presumably used on patients with unrecognized infections as well as on those with known infections.

In 1988, The World Congress of Gastroenterology convened a Working Party on endoscopic disinfection and produced a consensus report on endoscope decontamination [93] that emphasized 10 points:

1. Every endoscopic procedure should be performed with a clean, disinfected endoscope.

2. Manual cleaning of the endoscope surface, valves, and channels, the most important step for preventing the transmission of infections during endoscopy, should occur immediately after each procedure to prevent drying of secretions or formation of a biofilm, both of which may be difficult to remove. The endoscope should be immersed in warm water and detergent, washed on the outside with disposable sponges or swabs, and brushed on the distal end with a small toothbrush. Valves should be removed, cleaned by brushing away adherent debris and flushing detergent through lumina of hollow components, then disinfected. The biopsy-suction channel should be thoroughly cleaned with a brush appropriate for the instrument and channel size.

3. Equipment that cannot be immersed should be phased out.

4. Disinfection should be done by soaking the endoscope and all internal channels for at least 5 to 10 minutes in 2% glutaraldehyde. (Many investigators have suggested 20 minutes.)

5. To reduce bacterial colonization during overnight storage, the channels should be rinsed with 70% alcohol and dried with compressed air and the endoscope then stored in a hanging position.

6. Automatic washing devices can be used for disinfection procedures, but these devices do not perform or replace manual cleaning.

7. Colonization of automatic washers with opportunistic microorganisms may occur; thus, periodic culturing of these machines is necessary.

8. Accessories that breach the mucosa, such as biopsy forceps, should be mechanically cleaned and then autoclaved after each use.

9. More research needs to be done regarding disinfection in endoscopy.

10. Equipment should be redesigned to enable a verifiable system of disinfection analogous to systems used for sterilization of surgical equipment.

Much less attention has been focused on cleaning and disinfection practices for bronchoscopes. In 1989 The Working Party of the British Thoracic Society published guidelines regarding bronchoscopy and infection control [94]. To prevent cross-contamination, this group recommended the following:

1. The bronchoscope should be cleaned and disinfected at the beginning of the first procedure of the day and immediately after each procedure.

2. After dismantling the valve, all parts of the bronchoscope should be washed and brushed in neutral detergent.

3. After each procedure, the bronchoscope should be soaked in 2% glutaraldehyde for 20 minutes. Bronchoscopes should be washed before disinfection because glutaraldehyde can act as a fixative for mucus and blood.

4. After the glutaraldehyde soak, the channel should be rinsed with sterile water or 70% alcohol but not with tap water.

5. Before bronchoscopy is done in an immunocompromised patient, the bronchoscope should be cleaned and disinfected as usual, except that a 60-minute glutaraldehyde soak should be done.

6. Nonimmersible bronchoscopes should be phased out.

7. If effective cleaning and disinfection is done, dedicated bronchoscopes for high-risk patients are unnecessary.

Although recommended cleaning and disinfection guidelines are well established, four major problems remain: glutaraldehyde-induced side effects; flaws in automatic endoscope washers; inherent difficulties in cleaning and disinfecting endoscopes because of instrument design; and lack of uniform compliance with current recommended guidelines.

Although glutaraldehyde solutions are the most widely recommended and used high-level disinfectant for endoscopes [95], the optimal immersion time remains controversial. The World Congress of Gastroenterology Working Party recommends at least a 5- to 10-minute glutaraldehyde disinfection but acknowledges that this duration may not kill mycobacteria or bacterial spores [93]. Recently, the Association for Practitioners in Infection Control recommended an immersion time of at least 20 minutes [96]. Glutaraldehyde can cause headaches, conjunctivitis, dermatitis, asthma-like responses, and nasal irritation in staff [97, 98]. Patients may also be adversely affected. One group of investigators attributed 13 cases of post-colonoscopy chemical proctitis to glutaraldehyde that was inadvertently left in the air-water channel after cleaning; at the time of colonoscopy, the glutaraldehyde was sprayed into the rectum [99]. In addition, investigators recently associated glutaraldehyde with an outbreak of abdominal cramps and bloody diarrhea in patients who had had sigmoidoscopy [100]. Because of their lack of effectiveness, many potential alternatives to glutaraldehyde, such as formaldehyde-alcohol, 3% phenolics, iodophors, isopropyl alcohol, ethyl alcohol, and a 1:16 dilution of glutaraldehyde-phenate, were recently removed from the recommended list of high-level disinfectants [96].

Automated disinfecting machines, which relieve personnel from the time-consuming manual disinfection process and from prolonged contact with glutaraldehyde, were introduced in the 1970s and were initially received with enthusiasm by endoscopy personnel [101, 102]. Critical evaluations, however, found these machines to be expensive, noisy, and heavy [103]. In addition, manual cleaning, which typically requires 10 to 15 minutes, is necessary before automatic disinfection [103]. The effectiveness of these machines depends on the water quality, water delivery system, type of disinfectant, type of flexible endoscope, exposure time, temperature, flow rates, and design of the washer [102]. Specific outbreaks have been associated with these machines; for example, a series of P. aeruginosa infections, including one death, resulted from inadequate drying of an endoscope by an automatic washer [23]. In one study, P. aeruginosa colonization (or infection) occurred in 16 of 240 patients (6.7%) who underwent endoscopic retrograde cholangiopancreatography during a 3-year period and in 99 of 1109 patients (8.9%) who underwent upper gastrointestinal procedures [101]. The investigators detected the presence of a thick biofilm of P. aeruginosa in the detergent holding tank, inlet water hose, and air vents of the automatic reprocessing machine used. They attributed the contamination to a design flaw in the washer, because contamination occurred despite flushing and disinfecting it with the manufacturer's recommended disinfectants. Other failures with these machines have occurred [23, 102, 104-106].

In a recent study, investigators prospectively compared automated and manual disinfection [107]. They did rigorous cleaning and disinfection of gastrointestinal endoscopes that included brushing and cleaning with hydrogen peroxide and detergent, automated or manual disinfection with 2% glutaraldehyde (minimum of 20 minutes), rinsing with 70% alcohol, and forced-air drying. Cultures obtained on a weekly basis showed that neither the automated nor the manual process completely decontaminated the endoscopes; 10 of the 60 endoscopes (17%) had positive cultures after cleaning and disinfection. Cultures taken from automated disinfected endoscopes typically grew Pseudomonas species and atypical mycobacteria, whereas those from manually disinfected endoscopes typically grew colonic flora (Klebsiella pneumonia and Escherichia coli). In addition, the automated disinfecting machines used in this study were heavily contaminated with multiple Pseudomonas species and atypical mycobacteria.

Because of their complex physical arrangement of various channels and valve systems, endoscopes may remain contaminated despite effective cleaning and disinfection [2, 107]. Most gastrointestinal endoscopes have three major channels: air, water, and biopsy-suction (see Figure 3) [108]. The air and water channels, which are approximately 1.0 to 1.2 mm in internal diameter, cannot be cleaned by physical means and require flushing with liquids or air [2]. Within the endoscope, lumens, crevices, joints, pores, and loosely mated or occluded surfaces provide areas that may collect patient material. One investigator, using a rigid 2-mm arthroscope placed into the biopsy-suction channel, examined and photographed the internal surfaces of endoscopes and discovered numerous abnormalities: an irregular, kinked channel surface at the junction between the suction-channel tubing and the suction-valve housing (rather than a smooth circular junction) in a new unused instrument; accumulated patient material at this junction in a used endoscope; total blockage of the channel by accumulated patient material; residual patient material on and around the valve seat in a suction-valve housing; and multiple holes in the interior surface of a suction channel with adjacent heavy encrustations of patient material [2]. The presence of such residual patient material may contribute to the failure of cleaning and disinfection procedures [109].

Seven recent studies addressed the level of adherence to recommended guidelines for cleaning and disinfection of gastrointestinal endoscopes [14, 110-115]. One survey compared the cleaning and disinfection methods used by U.S. family practice physicians who perform flexible sigmoidoscopy with those recommended by the CDC [110]. Among the 1585 physicians (67%) who responded, 32% used appropriate procedures, 54% did not follow recommended methods, and 13% used appropriate solutions but at an inadequate concentration or for an insufficient duration. The investigators concluded that many family practice physicians who do flexible sigmoidoscopy use disinfection procedures that do not inactivate HIV. In a similar survey of 74 endoscopy centers in western Europe, 30% of the centers inadequately disinfected the endoscope after procedures in patients with upper gastrointestinal bleeding and unknown HIV or HBV status [111]. For routine procedures in patients with unknown HIV or HBV status, 70% of centers did not adequately disinfect instruments after endoscopic retrograde cholangiopancreatography, 87% did not adequately disinfect instruments after colonoscopy, and 100% did not adequately disinfect instruments after upper endoscopy. After endoscopy in patients with known Clostridium difficile infection, 30% of the centers performed inadequate disinfection procedures. An Australian survey reported that only 45% of hospitals both cleaned and disinfected endoscopes satisfactorily [112]. In a recent study, the Center for Devices and Radiologic Health and the state departments of Iowa, Maryland, and Massachusetts conducted a collaborative investigation of 26 health care facilities [113]. These investigators found that 78% of the facilities failed to sterilize biopsy forceps, and they isolated at least 100 000 colonies of bacteria from 24% of cultures taken from the internal channels of 71 endoscopes. Additional studies have documented a lack of uniform compliance with recommended guidelines [14, 114, 115].

Conclusions

Our review disclosed 281 infections transmitted by gastrointestinal endoscopy (Tables 1 and 2) and 96 infections transmitted by bronchoscopy (Table 3). The clinical spectrum of these infections ranged from asymptomatic colonization to serious disease, sometimes with fatal outcomes. In addition, numerous clusters of bacterial pseudoinfections have been associated with bronchoscopic procedures [102, 116-128]. In these instances of pseudoinfection, organisms were not inoculated into patients but, rather, into contaminated patient fluids that were suctioned through a contaminated suction valve or suction tubing. Although these pseudoinfections did not involve transmission of infection, they frequently led to unnecessary diagnostic and therapeutic interventions because infection was thought to be present.

Agents repeatedly identified in transmission of infections by endoscopy include Salmonella species and P. aeruginosa in gastrointestinal endoscopy, and M. tuberculosis, atypical mycobacteria, and Pseudomonas species in bronchoscopy. In general, Salmonella species and M. tuberculosis originally contaminate endoscopes from an infected patient, whereas Pseudomonas species and atypical mycobacteria originate from aqueous reservoirs, such as those used during cleaning and disinfection (see Figure 1). It is striking that the range of bacterial pathogens associated with these reports is so narrow. This may be explained in part, by a recognition phenomenon attributable to the unique nature of Salmonella and M. tuberculosis infections. Perhaps other more common infections are transmitted by endoscopy but are not recognized. Alternatively, the ability of both Salmonella species and M. tuberculosis to produce long-term carrier states, as well as the relative resistance of mycobacteria to disinfectants, may also contribute to their frequent role in infections transmitted by endoscopy.

Aside from the single documented case of HBV infection [49] and the mention of HBV and Creutzfeldt-Jakob agent transmission in one survey [14], no cases of viral agents transmitted by endoscopy have been reported. Given the long incubation periods of these infections, it may be difficult to link their transmission with a procedure done weeks or months before.

Overall, the number of reported infections that we have described is remarkably small, considering the millions of endoscopic procedures done annually [1], suggesting a low incidence of infections transmitted by endoscopy. These recognized and reported cases, however, probably represent a minority of all infections transmitted by endoscopy, because they were primarily due to easily recognized bacterial infections characterized by short incubation periods and often occurring in large or unusual clusters. Low-level endemic transmission is almost certainly more difficult to detect and may be considerably underestimated. In addition, asymptomatic or brief self-limited illnesses would likely go unrecognized. For viral illnesses with long incubation periods, such as infections with hepatitis viruses or retroviruses, linking their transmission to a procedure done weeks or months before is difficult. Finally, from the practitioner's perspective, the incentive for recognizing and reporting these infections is minimal.

Given the above limitations and lack of prospective studies, the true incidence of infections transmitted by endoscopy is impossible to determine. To better understand the risk for transmitting infections by endoscopy, large prospective studies should be conducted in which cultures are obtained from patients and patients are monitored for clinical disease after endoscopy. In addition, such studies should compare the risk for transmitting infections in procedures with and without biopsy and should concurrently culture endoscopes and accessories to evaluate the compliance with (and effectiveness of) cleaning and disinfection guidelines. Recently, the CDC requested physicians to report episodes of endoscopy-related infections (or pseudoinfections) through state health departments to the Epidemiology Branch, Hospital Infection Program at the CDC [129]; widespread physician participation in this reporting system is essential and may provide a better estimate of the incidence of such infections.

Endoscopic transmission of infections results predominantly from three factors: improper cleaning and disinfection procedures; contamination of endoscopes by mechanical washers; and inability to decontaminate endoscopic valves or channels despite appropriate disinfection (due to the mechanical configuration of the endoscopes). Failure to follow recommended cleaning and disinfection procedures probably constitutes the single most important factor leading to endoscopic transmission of microorganisms. The reasons for poor compliance are unclear, but in one survey [14] gastroenterology nurses suggested that poor compliance results from a lack of administrative support (59%), an insufficient number of endoscopes (32%), an absence of clear-cut standards for infection control procedures (19%), a lack of involvement of infection control staff (18%), the pressure to shorten down-time of endoscopes between patients (14%), and the complexity of the equipment, which makes cleaning and disinfection difficult (11%). In the future, recommended cleaning and disinfection guidelines, such as those published by the Working Party on endoscopic disinfection [93] and the Working Party of the British Thoracic Society [94], must be strictly followed. Moreover, because clinical variables and risk-factor assessment are not reliable predictors of which patients are HIV-infected [62, 130, 131], the practice of using stringent cleaning and disinfection only after endoscopes are used on known HIV-infected patients has little rationale and should be abandoned.

Problems related to equipment design have played an important role in outbreaks of infections (or pseudoinfections) associated with the use of automated washers. These problems include the difficulty in disassembling, cleaning, and decontaminating the machine; the reuse of detergent, disinfectant, and tap water by the automated washer; the presence of tubing and reservoirs that remain filled with fluid for extended periods of time; the failure to rinse endoscopes with sterile water or alcohol after disinfection; and the failure of the automated machine, in some instances, to self-disinfect once it is contaminated [101, 129]. Given some of these difficulties with current equipment design, emphasis needs to be placed on manual cleaning before automated disinfection and on the rinsing of channels with 70% alcohol followed by suctioning with forced air after automated disinfection. Moreover, frequent monitoring of these machines is necessary. Future automated washers should be designed to ensure their resistance to internal contamination [101]. In addition, they should have an internal monitoring system, with mechanical controls and chemical monitors, that automatically cancels the cycle when parameters are not acceptable [102].

Despite appropriate cleaning and disinfection procedures, some endoscopes, because of their complex internal structure, remain contaminated [2, 107]. Further studies should be done to evaluate the effectiveness of cleaning and disinfection procedures on the internal components of endoscopes. Such studies could include contaminating an endoscope, cleaning and disinfecting it, and then dismantling and culturing the internal channels and valves. The Working Party on endoscopic disinfection [93] recommended that equipment be redesigned to enable a verifiable system of disinfection analogous to systems used for sterilization of surgical equipment. Recently, such a newly redesigned endoscope was introduced [132, 133]. This sterile sheathed endoscope consists of two major components: a reusable endoscope without channels; and a disposable, sterile sheathed set that includes a sheath, air and water channels, a biopsy-suction channel, a distal window, and a cover for the endoscope control body [132]. Thorough evaluation of this newly designed endoscope, as well as other new products, will be necessary.