Selective Digestive Tract Decontamination—Will It Prevent Infection with Multidrug-Resistant Gram-Negative Pathogens but Still Be Applicable in Institutions where Methicillin-Resistant Staphylococcus aureus and Vancomycin-Resistant Enterococci Are Endemic?

Abstract

Infections that develop during stays in intensive care units (ICUs) represent a serious threat for critically ill patients. Respiratory tract infections in patients undergoing mechanical ventilation (ventilator-associated pneumonia [VAP]), intravascular device-associated infections, and postoperative infections are the most common among patients in ICUs. Depending on patient selection criteria and the specificity of diagnostic criteria, incidence rates of VAP as high as 70% have been reported [1]. Because of the increasing resistance of nosocomial pathogens to antibiotics (and probably also because of the insufficient feedback on ward-specific ecological profiles and susceptibility patterns from microbiology laboratories), initial antimicrobial treatment of VAP frequently is inappropriate in some settings, which decreases the likelihood that patients will survive [2, 3]. For all these reasons, the prevention of ICU-acquired infections, and of VAP in particular, has become an important measure for improving patient care in ICUs.

Early-Onset and Late-Onset Infections

ICU-acquired respiratory tract infections have a typical etiological distribution. Infections occurring within the first days of mechanical ventilation and hospital stay are most frequently caused by commensal bacteria of the respiratory tract, such as Streptococcus pneumoniae, Haemophilus influenzae, and methicillin-susceptible Staphylococcus aureus. Until now, the likelihood that healthy individuals would be colonized with methicillin-resistant S. aureus (MRSA) has been low, but this situation may be changing in the United States [4]. It is assumed that these so-called early-onset infections are already incubating at the time of intubation. Empirical antibiotic therapy for early-onset VAP in patients who are admitted directly from the community and who do not have specific risk factors for colonization with Pseudomonas aeruginosa should be primarily directed against the pathogens causing these early-onset infections.

In most patients, respiratory tract colonization with pathogens that cause early-onset infection is replaced during the first week of stay in the ICU by colonization with gram-negative enteric bacteria (e.g., Escherichia coli and Klebsiella pneumoniae), P. aeruginosa, Acinetobacter species, and MRSA. Although antibiotic treatment, which is prescribed to ∼70% of patients during their first week in an ICU, probably contributes to this colonization shift, replacement also occurs in patients who are not receiving antibiotics. The mechanisms underlying these ecological changes are poorly understood.

Principles of Selective Decontamination of the Digestive Tract (SDD)

The purposes of SDD are to treat infections that are incubating at the time of admission to the ICU and to prevent ICU-acquired infections. Incubating infections are treated by intravenous administration of antibiotics, irrespective whether there is clinical suspicion of infection, during the first 4 days of stay in the ICU. Because these infections are caused by commensal flora, they are called “primary endogenous” infections. Late-onset infections, with pathogens such as gram-negative enteric bacteria (e.g., E. coli and K. pneumoniae) and P. aeruginosa, are always preceded by colonization and, therefore, are called “secondary endogenous” infections.

Colonization and subsequent infection with ICU-acquired bacteria are prevented by topical application of antibiotics in the oropharynx and the gastrointestinal tract. For this, a paste is applied to the buccal cavity, and a solution is administered through the nasogastric tube; both the paste and the solution contain antibiotics that cover the flora typically associated with ICU-acquired infections, such as gram-negative enteric bacteria (e.g., E. coli and K. pneumoniae) and P. aeruginosa infection. Another prerequisite for the antibiotic spectrum of both systemic and topical components is that the anaerobic intestinal flora should not be disturbed. Animal studies conducted in the 1970s demonstrated that maintenance of the anaerobic intestinal flora provides protection against colonization and infection with gram-negative bacteria [5]. The “selective” part of SDD reflects this anaerobic flora-sparing effect. In the classic SDD approach, a second-generation cephalosporin was used for intravenous prophylaxis, and colistin and tobramycin were used for topical prophylaxis [6]. In anticipation of potential overgrowth with yeasts, amphotericin B was added to this topical regimen. The final parts of SDD consist of optimal hygiene in the ICU, to prevent exogenous infections, and the performance of surveillance cultures of rectal and respiratory tract samples, to monitor decontamination efficacy and the emergence of SDD-resistant pathogens.

Effects of SDD

The concept of SDD, as described above, emerged from the findings of a series of uncontrolled studies performed by Stoutenbeek et al. [6], in trauma patients in the early 1980s. Compared with historical controls, decontamination of the gastrointestinal tract had no discernible effect on the incidence of early-onset and late-onset VAP. However, the addition of oropharyngeal decontamination prevented late-onset but not early-onset VAP. Subsequently, intravenous prophylaxis was added, and the development of both early-onset and late-onset VAP was completely prevented. Since then, multiple trials have been performed among different patient populations and with many variations of the classic SDD regimen. The general picture emerging from all these studies was that SDD significantly reduced the incidence of VAP in settings where levels of antibiotic resistance were low [7], that the magnitude of reduction in the incidence of VAP was inversely related to the methodological quality of study designs [8], and that significant reductions in ICU-associated mortality rates were not demonstrated unequivocally in individual studies [7]. Significant reductions in ICU-associated mortality rates could be demonstrated only in meta-analyses [9, 10].

Because of these conflicting results, controversy has persisted as to whether SDD should be used as a routine measure of infection prevention in ICU patients. Proponents have pointed to the beneficial outcomes in individual trials and meta-analyses [11], whereas opponents have addressed the lack of sound scientific evidence on solid outcome measures (such as patient survival or length of ICU stay), unknown cost efficacy, and the constant threat of increased selection of pathogens resistant to the SDD regimen (such as MRSA and vancomycin-resistant enterococci [VRE]) [12]. As a result, the routine use of SDD has not become common practice, even in countries with low levels of antibiotic resistance, such as the Netherlands and the Scandinavian countries.

New Evidence for the Efficacy of SDD

Recently, the debate on SDD has been fueled by new evidence. Improved patient survival was demonstrated in 2 randomized trials [13, 14]. Krueger et al. [13] described a randomized, stratified, double-blind, placebo-controlled trial of the effects of a regimen combining intravenously administered ciprofloxacin during the first 4 days in the ICU and a mixture of gentamicin and colistine applied in the nostrils and the oropharynx and through the nasogastric tube, for 527 patients. Vancomycin was added to this mixture in patients receiving immunosuppressive therapy or patients with acute respiratory distress syndrome. At admission to the ICU, patients were stratified into 3 groups according to their APACHE II scores (<20, 20–29, and >29). Overall, ICU-associated survival rates in both study groups were not significantly different (relative risk [RR], 0.761; 95% CI 0.533–1.086), but ICU-associated mortality rates were significantly reduced among patients with mid-range APACHE II scores (RR, 0.508; 95% CI, 0.295–0.875). One year after inclusion, however, survival differences were reduced but still approached statistical significance (RR, 0.72; 95% CI, 0.496–1.046; P = .08) [13].

In the second trial, de Jonge et al. [14] used an innovative study design to evaluate the unitwide effects of SDD. In that study, eligible patients were randomized to 1 of 2 wards, which had identical structures, were attended by the same physicians, and used the same medical protocols but had different nursing teams. SDD was applied to all patients in 1 of the 2 units, whereas the other unit served as control unit. A total of 1090 patients were randomized during the 2-year period, and the RR of dying while in the ICU was significantly reduced in the unit where SDD was used (RR, 0.65; 95% CI, 0.44–0.85). This survival benefit was accompanied by a significant reduction in the length of ICU stay (11.6 vs. 13.4 days; P < .01) and reduced hospital-associated mortality (RR, 0.78; 95% CI, 0.63–0.95). Extensive monitoring of antibiotic resistance revealed that gram-negative bacteria resistant to tobramycin, imipenem, or ciprofloxacin were recovered less frequently from patients receiving SDD. Of note, there were no patients colonized (or infected) with MRSA, and intestinal colonization with VRE was found only sporadically.

Why were the results of this trial, the largest SDD trial to date, so outrageously beneficial? On the basis of meta-analyses, relative reductions in ICU-associated mortality rates could be expected to be ∼10%–15% [8–10], much lower than the 35% reduction reported by de Jonge et al. [14]. Poor randomization does not seem to be a likely explanation. Although not all patients could be randomized (for instance, in cases when only a single ICU bed was available), demographic variables and disease severity at admission were similar for both groups. A preexisting outcome difference between the 2 wards might have influenced the results. In fact, the RR of ICU-associated mortality during the 2 years before the study was 0.9 (95% CI, 0.7–1.1) for patients treated in the ward where SDD was administered. This preexisting difference was not incorporated in the final analysis and would have slightly reduced the efficacy of SDD. Another clear difference between this trial and prior studies was the unitwide application of SDD, instead of randomization per patient. In the latter design, patients receiving and not receiving SDD are treated within the same unit. If cross-transmission is an important mode of bacterial spread, the choice of study design may have considerable effects on study outcome. For instance, with randomization per patient, patients receiving SDD will indirectly protect control patients from becoming colonized and infected, whereas the patients receiving SDD remain at risk to acquire infections exogenously from nondecontaminated patients. The inevitable effect would be a reduction in the true effects of SDD. Unfortunately, the relevance of cross-transmission has not been quantified in any SDD trial. However, the emergence of antibiotic-resistant pathogens in ICUs worldwide strongly suggests that cross-transmission is relevant in many settings.

Sdd and Antibiotic Resistance

Importantly, the study by de Jonge et al. [14] was performed in a setting where levels of antibiotic resistance are low and where MRSA is completely absent. Because the antimicrobial spectrum of the classic SDD regimen lacks coverage of most gram-positive bacteria, increased rates of colonization and infection with enterococci and MRSA can be expected and have been reported [15–17]. In a surgical ICU in Austria, the proportion of MRSA among all S. aureus isolates increased from 17% to 81% during a 5-year period when SDD was practiced. The increase was comparable among patients receiving and not receiving SDD, suggesting that cross-transmission frequently occurred [15]. Therefore, SDD has been considered to be contraindicated in settings where MRSA is endemic. Alternately, vancomycin has been added to the topical components of SDD to control for the emergence of MRSA [13, 18]. Clearly, topical prophylactic use of vancomycin is not an attractive option in settings where both MRSA and VRE are endemic. Horizontal transfer of the vancomycin-resistance gene between both species has been demonstrated recently [19]. On the other hand, if confirmed, the 35% reduction in the RR for ICU-associated mortality found by de Jonge et al. [14] would be tremendous. This benefit must be balanced against the additional infections and attributable mortality due to MRSA and VRE infection. Estimates of attributable mortality due to MRSA infections have ranged from considerable (OR, 1.93) [20] to low [21–23].

Little is known about the effects of SDD in settings where VRE is endemic. In an American ICU, where SDD was used infrequently between 1996 and 2000 and where 18.5% of all patients staying in the ICU for at least 3 days were colonized with VRE, administration of SDD alone was not associated with VRE carriage. However, the combination of SDD with either vancomycin use or combined vancomycin and ceftazidime use was significantly associated with VRE colonization, compared with administration of these antibiotics without SDD [24]. As for MRSA, increased selection of VRE must be balanced against the potential beneficial effects of SDD, because vancomycin resistance has been identified as an independent predictor of death due to enterococcal bacteremia [25].

Selection of antibiotic-resistant gram-negative bacteria has also been considered to be a serious adverse effect of SDD. The results of the study by de Jonge et al. [14], however, suggest the opposite: patients receiving SDD had lower rates of acquisition of antibiotic-resistant gram-negative pathogens, which suggests that SDD would be an effective measure for the control of antibiotic resistance. In fact, SDD has been used before to control nosocomial outbreaks [26]. It is unclear whether acquisition of antibiotic-resistant pathogens in the study by de Jonge et al. [14] resulted from mutations, horizontal gene transfer, selection of preexisting strains, or clusters of cross-transmission due to lapses in infection control. In the latter case, SDD compensates for poor infection control practices. However, on the basis of all the SDD literature, there is no evidence that, in a setting with low levels of antibiotic resistance, SDD will facilitate emergence of antibiotic resistance in the short term. Naturally, the long-term effects remain unknown.

Individual Components of SDD

As described above, SDD has 3 antibacterial components: systemic prophylaxis, oropharyngeal decontamination, and gastrointestinal decontamination. The earliest experiments with SDD in critically ill trauma patients suggested that gastrointestinal decontamination had no effect on rates of respiratory infection, whereas oropharyngeal decontamination and systemic prophylaxis prevented late-onset and early-onset pneumonia, respectively [6]. These observations, which were initially obtained in a series of uncontrolled studies, have been confirmed in randomized, double-blind, placebo-controlled trials. Emerging questions were whether a short course of systemic antibiotics or just oropharyngeal decontamination could be as effective as the full SDD regimen.

The concept of a short course of intravenous antibiotics has been tested in 1 randomized, controlled trial of 100 comatose patients [27]. Two doses of cefuroxim (1500 mg) administered near the time of intubation reduced the incidence of VAP from 50% among untreated patients to 24% among patients receiving prophylaxis with cefuroxim. More than 70% of VAP episodes were early onset. Further supporting evidence comes from at least 2 risk-factor analyses for the development of VAP [28, 29]. In both studies, the receipt of systemic antibiotics during the first days after admission to an ICU was independently associated with a reduced risk of developing VAP.

Four studies evaluated the effects of oropharyngeal decontamination on the development of VAP, and all 4 found a significant risk reduction [30–33]. In the largest and most recent trial, oropharyngeal decontamination resulted in a 62% reduction in the RR of developing VAP, without affecting gastric and intestinal colonization, thus underscoring the pivotal role of oropharyngeal colonization in the development of VAP [30]. Alternatively, oropharyngeal application of chlorhexidine might be useful for preventing VAP, although this concept has, so far, been studied only in patients with a low risk for this complication [34].

Until now, there has been no evidence that gastrointestinal decontamination is necessary for the observed effects of SDD. In fact, preoperative intestinal decontamination had no effects on postoperative infection rates among patients undergoing cardiopulmonary bypass surgery [35]. Whether intestinal decontamination is necessary for the prevention of antibiotic resistance development, as suggested by some, remains to be determined.

Conclusion

After 20 years, the use of SDD to treat patients in ICUs still is a controversial issue, although recent studies have clearly underscored the potential benefits of this approach. There are now 2 studies that have demonstrated substantial improvements in outcome. However, because of methodological concerns, these results need confirmation. Importantly, there is no evidence that SDD will, in the short term, increase levels of antibiotic resistance in settings with low baseline levels of resistance. In the Netherlands, where levels of antibiotic resistance have remained exceptionally low throughout the years, the recent findings have led to a multicenter study comparing SDD and oropharyngeal decontamination with the current standard of care. The results of that trial are expected in 2006.

The situation is less clear for settings where antibiotic-resistant pathogens, such as MRSA, VRE, and multidrug-resistant gram-negative bacteria, are endemic. The efficacy of SDD has not been demonstrated in such settings, and some studies have clearly demonstrated enhanced selection of these pathogens. Whether the benefits of SDD, as found in settings with low levels of resistance, outweigh enhanced selection and infection risks with resistant pathogens, remains to be established. Moreover, the addition of topical antibiotics targeted towards these resistant pathogens, such as vancomycin, should be carefully balanced against the risk of resistance selection.

Acknowledgments

Potential conflicts of interest. M.J.M.B.: no conflicts.

References