© 2002 American Thoracic Society
In Defense of EvidenceThe Continuing Saga of Selective Decontamination of the Digestive TractDepartment of Surgery Toronto General Hospital Toronto, Ontario, Canada Nosocomial infection is a common complication of critical illness (1). Developing in the most vulnerable of patients, it is associated with significantly increased morbidity and mortality. Fear of the consequences of failing to treat a suspected nosocomial infection has provided a potent impetus for early empiric therapy in the intensive care unit (ICU), although the benefits of such an approach are unproven. Ironically, the most convincing evidence that ICU-acquired infection is a cause rather than a consequence of critical illness derives from a body of work that has been largely ignored—that evaluating selective digestive tract decontamination (SDD). There are more than 30 published randomized controlled trials that evaluate the efficacy of SDD in preventing ICU-acquired infection and its attendant mortality. A meta-analysis of these studies, incorporating 5,727 patients, found that the combination of a short course (48 to 96 hours) of a systemic antimicrobial agent (usually a third-generation cephalosporin) and a mixture of topical antibiotics that selectively eliminate aerobic Gram-negative organisms and fungi (classically, tobramycin, polymyxin, and amphotericin) reduces the odds ratio for ventilator-associated pneumonia to 0.35 (0.29–0.41) and for mortality to 0.80 (0.69–0.93) or a 6% reduction from 30–24% (2). Benefits are greatest in surgical patients, in whom the risk of pneumonia is reduced to 0.19 (0.15–0.26), the risk of bacteremia to 0.51 (0.34–0.75), and risk of mortality to 0.60 (0.41–0.88) (3). The mechanism of this effect is less clear. It may be a consequence of the elimination of a gastrointestinal reservoir of bacteria, although enteral decontamination alone does not significantly improve mortality (2). A recent report in this journal suggested that oropharyngeal decontamination, without concomitant digestive tract decontamination, can reduce rates of ventilator-associated pneumonia, although it did not in this admittedly underpowered trial significantly improve survival (4, 5). In this issue of AJRCCM (pp. 1029–1037), Krueger and colleagues reported the results of a stratified, randomized, double-blind, placebo-controlled trial of up to 4 days of parenteral ciprofloxacin, combined with topical polymyxin and tobramycin for the duration of the ICU stay (6). The 546 predominantly surgical patients comprise the largest single study of SDD yet reported, and the results are convincing. Rates of respiratory, bloodstream, and urinary tract infections were reduced, and significantly fewer of the treated patients developed new organ dysfunction. Stratifying by baseline severity of illness, the authors found that benefits were most marked for patients with Acute Physiology and Chronic Health Evaluation II scores between 20 and 30. By a strict intent-to-treat analysis of the entire population, ICU mortality was reduced significantly (19.0% in the placebo group versus 27.5% for controls, p = 0.03), a number needed to treat of 12 (12 patients must receive SDD to achieve 1 extra survivor). Compared with such recent therapeutic successes as pressure-limited ventilation in acute lung injury (number needed to treat = 11) (7), or activated protein C for severe sepsis (number needed to treat = 16) (8), this is an impressive clinical benefit. A rigorous surveillance protocol of microbiological monitoring revealed reduced Gram-negative colonization of the digestive tract, with no evidence of emerging resistance, but heavy colonization by Gram-positive organisms and fungi in both groups. Subsequent use of systemic antimicrobial agents was reduced in patients receiving SDD, although use in both groups was high. One might anticipate that a well-performed, adequately powered, randomized, blinded, placebo-controlled trial that demonstrated a beneficial impact on mortality, a strong biologic effect on the specific target of therapy, and absence of harm, and that did so without significantly increasing costs, would lead to a substantial change in clinical practice. This is unlikely to happen. Resistance to the concept of SDD, particularly in North America, has been strong, even in the face of mounting evidence of benefit from individual trials and meta-analyses. In part, this reflects the reluctance of clinicians to abandon strongly held views and the lack of ready availability of a commercial preparation of the agents used. More substantively, however, it arises from persistent concerns about the potential effects of widespread SDD use on antibiotic resistance. Fears about resistance appear to arise from an early cohort study of 292 ICU patients treated with inhaled polymyxin to prevent nosocomial pneumonia. Ten patients developed pneumonia caused by polymyxin-resistant organisms, and their mortality rate was higher than that of historical controls (9); however, no controlled study to date has provided convincing evidence that the SDD regimen induces resistance in Gram-negative species; indeed, the technique has even been used to decontaminate patients colonized with resistant Gram-negative organisms (10, 11). Less clear, however, are the long-term effects of SDD on the emergence of resistance in Gram-positive organisms (12). Gram-positive infections appear to be more common when the regimen is used, and the safety of the use of SDD in settings where methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococci are prevalent has not been established (13). A recent double-blind randomized trial found that Gram-positive colonization and infection during SDD could be significantly reduced by the addition of nasal mupirocin to the regimen (14). Limiting antibiotic pressures to minimize the development of resistance is a crucial concern: whether SDD contributes to or reduces these pressures remains an open question. The available evidence clearly supports the use of SDD as an effective strategy that can reduce morbidity and mortality in selected groups of critically ill patients, specifically those with severe, but salvageable disease, who have been admitted with a surgical diagnosis, but the gulf between evidence and conventional wisdom is striking. Perhaps a North American study of SDD will be needed to resolve the debate finally. Such a study must recruit a population most likely to benefit, be adequately powered to detect or exclude a clinically important effect on mortality, and use a factorial design, comparing not only the full SDD regimen against placebo, but also including patients who receive topical or systemic antibiotics alone. It must further include long-term surveillance to detect potential changes in colonizing flora. At conventional levels of power and significance, approximately 2,400 patients would need to be studied to exclude a 5% mortality benefit for SDD in a population whose placebo mortality was 25%. This is eminently doable; the potential benefits for patients are immense. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
