The role of barrier precautions in infection control

JournaE

of Hospital

The role

Infection

(1991)

of barrier

18 (Supplement

A), 515-523

precautions

in infection

control

11. A. Goldmann Hospital Epidemiology Department and Division of Infectious Diseases, Children’s Hospital and the Department of Pediatrics, Harvard Medical School, Boston, Massachusetts, USA Summary: Barrier precautions are a fundamental component of any infection control strategy and a critical aspect of all isolation systems. Because many infections are transmitted from patient-to-patient via the hands of personnel, gloves and gowns are widely recommended to provide an extra measure of protection against cross-infection. It is not clear whether gloves are superior to handwashing (if performed obsessionally) in this respect, and there is little evidence that gowns confer additional benefit. These concerns notwithstanding, barrier precautions can substantially reduce the risk of some infections, such as respiratory syncytial virus disease. On the other hand, the modes of transmission of many infections are complex (e.g. with rotavirus) or controversial (e.g. with rhinovirus), and, even though hands are involved in transmission, barrier precautions alone may not suffice to prevent spread. Moreover, neither gloves nor gowns can prevent nosocomial infections caused by endogenous microbial flora; perhaps this explains the limited efficacy of barrier precautions in reducing the endemic rate of infection due to bacteria such as Pseudomonas aeruginosa in intensive care units. Barrier precautions may also fail if colonized patients are not identified promptly. One potential solution to this problem is ‘body substance isolation’ (BSI), in which all patients are considered to be potential carriers of nosocomial pathogens whether or not they have been cultured or have developed a clinical infection. In BSI barrier techniques are used when any potentially contaminated patient material is handled. BSI also provides barrier protection from bloodborne pathogens for personnel. Proponents claim that BSI obviates the need for more complex precaution systems, but BSI does not take into account the complexity of disease transmission and is largely untested. ‘Universal precautions’ were designed to protect personnel, not to provide barriers to cross-infection, and are prone to misinterpretation and misapplication. Keywords:

Precautions;

isolation;

nosocomial

infection;

cross-infection.

Introduction

The practice of barrier precautions to prevent cross-infection, particularly the use of gloves, has all the characteristics of a typical ritual. The high priests of infection control proclaim its virtues and infection control practitioners accept it without question, despite very little scientific or epidemiological documentation of efficacy. Indeed, glove manufacturing is a recession-proof industry. Children’s Hospital, Boston, a 340-bed Correspondence to: Dr Donald A. Goldmann, Longwood Avenue, Boston, MA 02115. 01954701/91/06A515+09

Division

$03.00/O

of Infectious

Diseases,

Children’s

0 1991 The Hospital

515

Hospital, Infection

300 Society

516

D. A. Goldmann

university-affiliated referral centre, spent $391 500 on gloves in 1989, and there is no reason to believe that the crisis in hospital financing in the USA will lead to retrenchment in this particular area. Although many of the gloves that are purchased by hospitals are now used for personal protection as required by universal precautions,’ rather than to prevent person-to-person spread of nosocomial pathogens, the continued devotion of hospital personnel to barrier precautions is obvious. The rationale

for the use of gloves and gowns

If we begin by acknowledging that many organisms of nosocomial importance are transmitted from person-to-person via the hands of health care staff, then-whether or not we have the evidence to prove it-barrier precautions make sense. Of course, it is hard to imagine that thorough handwashing, especially if chlorhexidine gluconate or another suitable agent is used, would be less efficacious than gloves, which are, after all, considerably more expensive. Situations in which gloves may be superior to handwashing, such as patient care by clinical personnel who have chronic hand colonization with a nosocomial pathogen2 probably occur rarely. Thus, when we advocate the use of gloves, we may be hoping for some positive psychological or behavioural effect. Perhaps wearing gloves reminds personnel that patients harbour dangerous microbes and that aseptic technique is critical. Perhaps compliance is improved because personnel believe that they will be easier to spot if they forget to wear gloves than if they bypass the sink. Unfortunately, appropriate studies have not been performed to determine whether the use of gloves improves compliance with infection control practices. As for gowns, which are often lumped together indiscriminately with gloves as if personnel handled patients with their chests or abdomens, there is little evidence that they confer additional benefit. Certainly, the habit of donning an overgown merely to enter a nursery or intensive care unit is not such overgowns might supported in the medical literature. 3*4 If anything, if not changed between patients.’ abet the transfer of organisms Nonetheless, neonatal intensive care units cling to this ritual, even when the physicians and nurses who work in these units do not believe that overgowns are indicated. 3 Gowns make sense only if worn for individual patient contacts when soiling of clothes is likely and perhaps when caring for patients with burns or desquamative skin conditions. It is not known whether paper or non-woven synthetic polymers are superior to cloth for this purpose.‘j Evidence

of the effectiveness

of barrier

precautions

For the sake of argument, however, let us assume that gloves provide a rational barrier to the transmission of microorganisms in the hospital, and

Barrier

precautions

517

let us acknowledge that gowns continue to be used along with gloves for barrier precautions even though they are probably not essential most of the time. Given these assumptions, do the published data support the efficacy of barrier precautions? Respiratory syncytial virus (RSV) provides a convenient model for studying this issue. Nosocomial RSV is a major problem in paediatrics, causing apnoea and unexpected death in neonates, severe and prolonged infections in immunosuppressed patients, severe respiratory compromise in children with underlying cardiac or pulmonary disease, and occasionally life threatening bronchiolitis even in normal infants.7 Nosocomial RSV infections are common and predictable, occurring in up to 45% of infants hospitalized for 1 week or longer and in nearly 100% of children who remain in the hospital for more than a month during the epidemic season.* that RSV is transmitted from There is considerable evidence patient-to-patient by healthcare staff who have contaminated their hands personnel frequently with the secretions of infected babies .9 In addition, infect themselves by inoculating virus into their eyes or nose. Since RSV tends to produce the symptolms of a severe cold in adults,” personnel who elect to continue working, despite their infection, serve as efficient vectors in the transmission of RSV to babies under their care. Thus, because the hands play such a key role in the spread of RSV, it should be possible to block transmission by using simple barrier techniques to protect the hands from viral contamination. When we examined compliance with barrier precautions at Children’s Hospital, however, we found that personnel on our infant/toddler ward wore gloves and gowns for only 39% of contacts with patients infected with RSV.” In an effort to improve performance, we openly monitored compliance with precautions on this ward over a 4-week period. Not surprisingly, compliance improved dramatically, with personnel wearing gloves and gowns for 81% of contacts at the start of the monitoring period and in nearly 9.5% of contacts towards the end. We feared that compliance would diminish once open monitoring ceased, but instead compliance remained at very high levels for nearly 1.5 years. Apparently, the beneficial impact of improved compliance on the RSV infection rate, as noted below, motivated nurses to maintain their excellent performance standards-an attitude which was reinforced by the vigorous efforts of the head nurse. To determine the magnitude of the impact of increased compliance with barrier precautions on the nosocomial RSV infection rate, we measured the incidence density (cases per 1000 patient days) before and after the open monitoring period. ‘i Intuitively, the risk of nosocomial RSV infection should be influenced by exposure of susceptible children to other patients on the ward who already were excreting virus. Therefore, we calculated a weekly index of risk based on the number of infected patients on the ward, as described previously. ” After making this adjustment, we found that the summary risk of nosocomial RSV infection was 2.9 times greater (95%

518

D. A. Goldmann

confidence interval 1.5-5.7) in the low compliance period than in the high compliance period. As expected, we noted a strong association between increasing infection rate and increasing level of exposure to RSV. Interestingly, however, the slope of this relationship was 4*5-fold steeper when compliance was poor. In other words, gloves and gowns worked best when the burden of RSV on the wards was greatest, and barrier precautions can be expected to have the largest impact when they are needed most-at the height of the seasonal RSV epidemic. Klein et aL6 have taken a different approach to evaluating the efficacy of barrier precautions. They performed a randomized prospective trial of barrier precautions in a paediatric intensive care unit. Rather than studying a specific pathogen, such as RSV, all nosocomial bacterial and fungal infections were monitored. Gowns and gloves were used for all patients randomized to the barrier precautions group, regardless of whether the child had signs or symptoms of an infection. In essence, the precautions used in this study were similar to those specified in so-called ‘body substance isolation’ which is discussed below. Barrier precautions appeared to have a major impact on the transmission of nosocomial pathogens in this study. The rate of nosocomial colonization was 31% in the barrier precautions group and 42% in the standard care group, with colonization occurring significantly earlier (median 7 ZX. 12 days) in patients receiving standard care. The nosocomial infection rate was 4.4 infections per 100 days in patients on precautions and 8.6 per 100 days in patients on standard care (P = 0.008). Actuarial analysis showed that the barrier precautions group had a significantly longer infection-free interval, and the benefit of precautions appeared to increase the longer patients remained in the unit. Almost all of the observed nosocomial bacteraemias occurred in the standard care group, although this difference did not reach statistical significance. On the other hand, it must be recognized that this was a rather small study with some limitations. There were nine cases of tracheobronchitis, which is a problematic diagnosis of questionable clinical significance in ventilated children. There were three cases each of sinusitis and otitis-infections which probably are related more to mechanical factors than to colonization with specific nosocomial organisms per se. Body

substance

isolation

Despite these rather minor reservations, the study by Klein et aL6 was a pioneering evaluation of a new system of precautions that has become increasingly popular in the USA-body substance isolation (BSI).12 This system was developed to address some of the problems that have been encountered over the past two decades as hospitals have struggled to implement the traditional category-specific and disease-specific isolation protocols espoused by the Centers for Disease Control (CDC).i3 Some of the advantages and potential disadvantages of these CDC systems are

ISarrier

519

precautions

outlined in Table I. BSI was designed to avoid some of these problems while providing healthcare staff with relatively straightforward, uniform procedures for limiting cross-infection and at the same time protecting personnel from exposure to bloodborne pathogens. The underlying rationale of BSI is powerful and persuasive. Hospitalized patients, it is argued, all harbour potentially pathogenic microorganisms, including nosocomial strains, which reach high concentrations in oral precautions secretions, stool, and other ‘body substances’. Since traditional focus on patients with doculmented infections, they essentially deal with only the tip of a vast microbiological iceberg. True, these systems do include provisions for isolating patients who are colonized but not infected such as methicillin-resistant with certain antibiotic-resistant strains, Staphylococcus aureu~ and aminoglycoside-resistant Gram-negative rods. However, this approach ignores pathogens that may be equally virulent but do not happen to express the resistance phenotype that the hospital epidemiologist or microbiologist finds fashionable at the moment. The key elements of BSI can be summarized as follows.‘2 1. Non-sterile gloves are worn for contact with mucous membranes, non-intact skin, blood, secr’etions, and ‘moist body substances’ for all patients; gloves are changed before contact with other patients. 2. Handwashing apparently is indicated between patient contacts,i4 although the original description of BSI explicitly stated that ‘handwashing is unnecessary . . . unless hands become visibly soiled due to punctures in the gloves’.12 3. Other barriers, such as gowns and goggles, are worn as necessary to prevent soiling of clothing or splashes into mucous membranes. 4. All laboratory specimens are considered to be potentially infectious. 5. Traditional precautions (single room, negative air pressure, and masks) are used for patients with airborne diseases, such as tuberculosis, Table

I. Advantages

and disadvantages of traditional

isolation systems

System

Advantages

Disadvantages

Category-specific

Simplicity

Infections with different specific modes of transmission are lumped single category

Disease-specific

in a

Precautions are based on the transmission of specific diseases and may appear more rational to caregivers

Requires a greater degree of clinical judgment than category-specific isolation

Only those procedures needed to contain a specific disease are implemented, potentially reducing cost

Difficult to apply to patients whose signs and symptoms do not suggest a specific diagnosis

520

D. A. Goldmann

although detailed instructions for pathogens that might be spread by droplets or droplet nuclei are not provided. Although the study by Klein et ~1.~ did not claim to be an evaluation of BSI per se, it is clear that the procedures used were very similar. This paediatric intensive care unit study remains the only rigorous published demonstration of the efficacy of BSI. The originators of the BSI system have published only uncontrolled descriptive data.14 Following implementation of BSI at the Harborview Medical Center, a 330-bed municipal teaching hospital affiliated with the University of Washington, compliance with recommended gloving procedures increased from 61% to 81%, with both doctors and nurses showing considerable improvement. Since nosocomial infection surveillance was not performed routinely in this institution, the isolation rate of specific ‘marker’ organisms was monitored as an index of the success of BSI in limiting transmission of nosocomial Gram-negative rods were pathogens. Serratia spp. and amikacin-resistant chosen as ‘markers’ because of their prevalence on the hospital wards, particularly in critical care units. The rate of serratia isolations fell from 2.0 per 1000 patient days in the preintervention year to 1 .l in the year following implementation, and the rate of amikacin-resistant Gram-negative rod isolations fell from 0.58 to O-42. However, a progressive decline in the recovery of both ‘markers’ continued in the subsequent 2 years, suggesting that other interventions or long-term secular trends may have been responsible for at least part of the observed changes. Even though efficacy data are sparse, infection control professionals are bound to be beguiled by BSI’s ‘back to basics’ appeal and ease of ‘Barrier precautions for all’ seems so rational that many implementation. programmes may be inclined to accept BSI without demanding further studies. However, there are grounds for caution. BSI is rather costly in comparison with conventional isolation systems, although the recent introduction of universal precautions for bloodborne pathogens has reduced the incremental cost of BSI considerably. In fact, confusion between BSI and universal precautions has not only had fiscal consequences but has also led to practices that paradoxically could increase the risk of cross-infection. Because universal precautions were implemented with the express purpose of protecting personnel from bloodborne pathogens, not to prevent transmission of organisms from patient-to-patient, this system does not reinforce attitudes required for vigorous adherence to barrier precautions. Although not yet subjected to careful study, numerous anecdotal reports suggest that personnel tend to disregard the rather subtle but critical differences between universal precautions and BSI, going from patient to patient wearing the same pair of gloves. The potential adverse consequences of this practice were emphasized in a recent report by Maki et aZ.,15 in studies of methicillin-resistant S. aweus transmission in a surgical intensive care unit. Even if personnel do remove gloves after caring for individual patients, they may be lured into neglecting handwashing by a false sense of

Barrier

precautions

521

security. Such lapses are dangerous since hands may be contaminated in the process of glove removal.16 The improper use of gloves may be remedied by intensive educational efforts and behavioural modification techniques. Unfortunately, however, even perfect application of barrier precautions may produce disappointing results in the very areas of thle hospital where we count on BSI to have its greatest impact, such as critical care units. Many, perhaps most, of the infections in these units arise from the patient’s endogenous flora and cannot be controlled by barrier methods. For example, patients admitted to intensive care units are frequently colonized by small numbers of Pseudomonas aeruginosa which may not be detectable by standard culture procedures. Under the selective pressure of broad-spectrum antibiotics to which almost all intensive care unit patients are exposed, P. aeruginosa proliferates to become a major component of the patient’s flora, often developing resistance to multiple antibiotics in the process.16 In a very careful systematic study of pseudomonas colonization in intensive care, Weinstein’s group 16,17demonstrated that only a small number of patients acquired their strain as a result of cross-infection, and only very few infections could have been prevented by barrier precautions. In contrast, the spread of aminoglycoside-resistant enterobacteriaceae was reduced by barrier precautions, presumably because these organisms are primarily transmitted from patient to patient on the hands of the staff.” In a different clinical setting, even brief exposure to first generation cephalosporin surgical prophylaxis led to outgrowth of Enterobacter spp. which were present at undetectable levels before surgery.” Similarly, Kernodle et cA.,~’ using extremely sensitive techniques, found a striking prevalence of methicillin-resistant coagulase-negative staphylococcal skin colonization in patients about to undergo cardiac surgery; cephalosporin prophylaxis was associated with easily detectable colonization postoperatively. Thus, barrier precautions, regardless of how well enforced, can be expected to have only a limited effect on the endemic rate of nosocomial infections caused by many pathogens, although epidemic spread might be prevented under some circumstances. In addition to these potential problems, BSI is based on oversimplified assumptions concerning the epidemiology of infectious diseases. For many pathogens the route of transmission is either complex or incompletely understood. Returning to the example of RSV cited earlier, it seems clear that direct contact with the patient is not required to contaminate the hands of personnel. RSV survives quite well on environmental surfaces, and personnel who never touch a patient are perfectly capable of acquiring RSV through contact with contaminated objects in the immediate vicinity of an infected baby. 9 Education cloncerning the importance of environmental contamination was, therefore, an integral part of our evaluation of gloves and gowns in controlling the spread of RSV on our infant/toddler ward. Personnel were specifically instructed to wear gloves for contact with any

522

D. A. Goldmann

fomite that might have been exposed to secretions from an infected infant. Control of other nosocomial pathogens, such as methicillin-resistant S. aureus and Clostridium di$icile, may also require extension of barrier techniques to the patient’s environment. BSI does not deal effectively with these nuances of microbial ecology. Microbes, after all, will not let themselves be crammed into an epidemiological niche just to make isolation precautions a little more convenient. For example, efforts to control rotavirus on paediatric wards have long been based on the assumption that this virus, like other enteric pathogens, must be spread by the faecal-oral route. When rotavirus transmission was noted despite enteric contact precautions, further investigation revealed that many children were intermittent, often asymptomatic excretors. 21,22In other words, rotavirus ought to be a perfect candidate for BSI. Unfortunately, rotavirus survives well and retains its infectivity for long periods in the environment23 and is probably spread by the respiratory route in the acute phase of infection,24 so BSI is unlikely to be a totally effective strategy. Infection control personnel who believe that we know enough about the transmission of infectious diseases to bundle most of them into a simple category may find it sobering to contemplate the status of our knowledge concerning the transmission of a major community nuisance, rhinovirus. Despite decades of investigation on the Salisbury Plain and in Wisconsin and Virginia, it is still not clear whether this virus is spread primarily through the air or by direct contact. 25,26Proponents of BSI probably would advise us to wear gloves when shaking hands, but perhaps we would be better off conducting our conversation in a well ventilated room. As always in infection control, the answers may not be as simple as they seem. References Universal precautions for prevention of transmission of human 1. Update. immunodeficiency virus, hepatitis B virus, and other bloodborne pathogens in health care settings. MM W R 1988; 37: 377-382. of personnel in the 2. Knittle MA, Eitzman DV, Bear H. Role of hand contamination epidemiology of gram-negative nosocomial infections. J Pediatr 1975; 86: 433437. LG. Overgown use for infection control in nurseries and neonatal 3. Cloney DL, Donowitz intensive care units. AJDC 1986; 140: 680-683. 4. Donowitz LG. Failure of the overgown to prevent nosocomial infection in a pediatric intensive care unit. Pediatrics 1986; 77: 35-38. of gowns in an intensive care unit. J Hasp Infect 1981; 2: 5. Nystrom B. The contamination 167-170. of nosocomial infection during pediatric 6. Klein BS, Perloff WH, Maki DG. Reduction intensive care by protective isolation. N Engl r Med 1989; 320: 1714-1721. 7. Goldmann DA. Nosocomial viral infections: recent development and new strategies. Eur J Microbial Inject Dis 1989; 8: 75-81. 8. Hall CB, Douglas RG, German JM, Measner MK. Nosocomial respiratory syncytial virus infections. N Engl J Med 1975; 293: 1343-l 346. of respiratory syncytial virus. J Pediatr 9. Hall CB, Douglas RG. Modes of transmission 1981; 99: 10&103. 10. Hall WJ, Hall CB, Speers DM. Respiratory syncytial virus infection in adults: clinical, virologic, and serial pulmonary function studies. Ann Intern Med 1978; 99: 203-205.

Barrier

precautions

523

11. Leclair JM, Freeman J, Sullivan BF, Crowley CM, Goldmann DA. Prevention of nosocomial respiratory syncytial virus infections through compliance with glove and gown isolation precautions. N Engl J Med 1987; 317: 329-334. 12. Lynch P, Jackson MM, Cummings MJ, Stamm WE. Rethinking the role of isolation practices in the prevention of nosocomial infections. Ann Intern Med 1987; 107:

243-246. 13. Garner JS, Simmons BP. Centers for Disease Control: guideline for isolation precautions in hospitals. In: ‘Guidelines for the Prevention and Control of Nosocomial Infections. Atlanta, GA: Centers for Disease Control 1981; 1-81. 14. Lynch P, Cumming MJ, Roblerts PL et al. Implementing and evaluating a system of generic infection precautions: body substance isolation. Am J Infect Control 1990; 1: 1-12. 15. Maki DG, McCormick RD, Zilz MA, Stolz SM, Alvarado CJ. An MRSA outbreak in a SICU during universal precautions: new epidemiology for nosocomial MRSA; downside for universal precautions (UPS). Abstract no. 41.3rd Decenniel Conference on Nosocomial Infections, Atlanta, GA, July 31-August 3 1990. 16. Olson B, Weinstein RA, Nathan C, Chamberlin W, Kabins SA. Occult aminoglycoside resistance in Pseudomonas aeruginosa: epidemiology and implications for therapy and control. J Infect Dis 1985; 152: 769-774. 17. Olson B, Weinstein RA, Nathan C, Chamberlin W, Kabins SA. Epidemiology of endemic Pseudomonas aeruginosa: why infection control efforts have failed. J Infect Dis 1984; 150: 808-816. 18. Weinstein RA, Kabins SA. Strategies for prevention and control of multiple drug-resistant nosocomial infection. Am J Med 1981; 70: 449-454. 19. Flynn DM, Weinstein RA, Nathan C, Gaston MA, Kabins SA. Patients’ endogenous flora as the service of ‘nosocomial’ Enterobacter in cardiac surgery. J Infect Dis 1987; 156: 363-368. 20. Kernodle DS, Barg NL, Kaiser AB. Low-level colonization of hospitalized patients with methicillin-resistant coagulase-negative staphylococci and emergence of the organisms during surgical antimicrobial prophylaxis. Antimicrob Agents Chemother 1988; 32: 202-208. 21. Eiden J’S, Verleur DG, Vonderfecht SL, Yolken RH. Duration and pattern of asymptomatic rotavirus shedding by hospitalized children. Pediatr Infect DisJ 1988; 7: 564-569. 22. Champsaur H, Questiaux E, F’revot J et al. Rotavirus carriage, asymptomatic infection, and disease in the first two years of life. I. Virus shedding. J Infect Dis 1984; 149:

667-674. 23. Keswick BH, Pickering

LK, DuPont HL et al. Survival and detection of rotaviruses on environmental surfaces in day care centers. Appl Environ Microbial 1983; 46: 813-816. 24. Prince DS, Astry C, Vonderfecht S et al. Aerosol transmission of experimental rotavirus infection. Pediatr Infect Dis 1986; 5: 218-222. 25. Dick EC, Jennings LC, Mink KA, Wartgow CD, Inhorn SL. Aerosol transmission of rhinovirus colds. J Infect Dis ‘1987; 156: 442448. 26. Gwaltney JM, Jr., Moskals,ki PB, Hendley JO. Hand-to-hand transmission of rhinovirus colds. Ann Intern &fed 1978; 88: 463467.