Selective Digestive Decontamination — Finding the Way Forward
Michael Klompas
N Engl J Med 2026;394:1543-1545
DOI: 10.1056/NEJMe2602823
In critical care, selective digestive decontamination (SDD) is the elephant in the room. Cluster-randomized trials, individually randomized trials, and meta-analyses have shown that the use of SDD is associated with a decreased risk of ventilator-associated pneumonia, of hospital-onset bacteremia, and in some studies, of death.1,2 Nonetheless, very few intensive care units (ICUs) outside the Netherlands practice it.
SDD consists of daily antibiotic prophylaxis with an anaerobe-sparing oropharyngeal paste (e.g., colistin, tobramycin, and nystatin), a gastric slurry (a mixture of similar antibiotic agents), and typically 4 days of treatment with intravenous ceftriaxone, cefotaxime, or ciprofloxacin. The regimen is designed to reduce the oral–gastrointestinal burden of organisms and the risks of pneumonia resulting from aspiration and of bacteremia resulting from bacterial translocation across the gut wall.
Clinicians outside the Netherlands have been reluctant to adopt the use of SDD for fear that daily use of prophylactic antimicrobial agents will select for antibiotic-resistant pathogens in both exposed persons and the broader population of patients in the ICU. Evidence that this occurs is sparse,3 but critics argue that the Dutch experience cannot be generalized to settings with higher baseline antibiotic use and resistance rates.
These concerns prompted trials set in ICUs with high baseline levels of antibiotic resistance. The Resistance in Gram-Negative Organisms: Studying Intervention Strategies (R-GNOSIS) trial was conducted in 13 ICUs in Europe in which at least 5% of bloodstream infections were due to extended-spectrum beta-lactamase–producing Enterobacteriaceae.4 This trial showed that SDD had no effect on 28-day mortality or ICU-acquired antibiotic-resistant gram-negative bloodstream infections; however, the SDD regimen did not include a parenteral component, and meta-analyses suggest that a reduction in mortality is contingent on the inclusion of a parenteral agent.2
In this issue of the Journal, the investigators of the Selective Decontamination of the Digestive Tract in Intensive Care Unit (SuDDICU) collaboration now report the final results of the SuDDICU trial, a crossover, cluster-randomized trial that was conducted in 19 ICUs in Australia and in 7 ICUs in Canada5; the results from the Australian units alone were reported in 2022.6 The SuDDICU regimen included daily application of an oral antibiotic paste and gastric suspension and a 4-day course of an intravenous antibiotic. The trial involved 20,000 patients, 9289 of whom were undergoing mechanical ventilation and were eligible for SDD. The remaining 10,711 patients were not eligible for SDD but were monitored for the development of new antibiotic-resistant organisms in an observational ecologic assessment. Crossover from SDD to standard care or vice versa occurred after 12 months (with a 3-month gap between the periods).
The use of SDD as compared with standard care had no significant effect on the incidence of in-hospital death (27.9% in the SDD group and 29.5% in the standard-care group; odds ratio, 0.94; 95% confidence interval [CI], 0.84 to 1.05), nor did it appear to alter the number of days alive and free from mechanical ventilation. SDD did appear to be associated with fewer bloodstream infections than standard care (4.9% and 6.8% of the patients, respectively), with an adjusted mean difference of −1.30 percentage points (95% CI, −2.55 to −0.05), and fewer antibiotic-resistant organisms than standard care (in 16.8% and 26.8% of the patients, respectively), with an adjusted mean difference of −9.60 percentage points (95% CI, −12.40 to −6.80). No apparent between-group differences were observed in the total defined daily doses of antibiotics or in the development of new Clostridioides difficile infections (secondary microbiologic outcomes).
In the ecologic assessment, SDD was noninferior to standard care with regard to the development of new bloodstream and C. difficile infections in the unexposed ICU population. In addition, there were no apparent differences in antibiotic-resistant infections in this population, but noninferiority could not be established. In addition, in a post hoc analysis that was restricted to patients with acute brain injury (2738 patients), SDD appeared to be associated with lower mortality than standard care (odds ratio for in-hospital death, 0.80; 95% CI, 0.68 to 0.94).
The SuDDICU investigators then echoed and amplified their findings in a meta-analysis published in NEJM Evidence.7 Across 32 randomized trials that collectively enrolled 27,687 participants, SDD was associated with reductions in in-hospital death (risk ratio, 0.91; 95% credible interval, 0.82 to 0.99), new bloodstream infections (risk ratio, 0.68; 95% credible interval, 0.57 to 0.81), and positive cultures for antibiotic-resistant organisms (risk ratio, 0.63; 95% credible interval, 0.44 to 0.90).
Do all these trials now provide sufficient evidence to advise routine use of SDD? It is tempting to dismiss the SuDDICU trial results because no significant effect on mortality was observed, the point estimates differed in the results for Canada and Australia, and the duration of the trial was too short to determine the long-term effect of SDD on antibiotic resistance in the ICU. Similarly, some trials in the meta-analysis were small, old, and less rigorous than the SuDDICU trial, and the results that were pooled across cluster-randomized trials were inconsistent with those pooled across individually randomized trials.
To dismiss the results of the SuDDICU trial and meta-analysis, however, would be to ignore the accumulating evidence that SDD confers benefits without clear harms. The net effect of SDD on mortality may still be uncertain, but fewer bloodstream infections and fewer resistant infections are concrete advantages (especially as U.S. hospitals prepare to publicly report hospital-onset bacteremia and fungemia rates and seek additional ways to lower sepsis-related mortality8). The fear that SDD increases overall antibiotic use and resistance has not been borne out. Rather, increased early use of prophylaxis appears to be counterbalanced by fewer treatment courses for clinical infections.3,5
In practice, however, most clinicians will want to see more trial data to confirm these benefits, particularly in their own specific locations and patient populations. One also wonders whether similar benefits may be realized with less aggressive regimens (e.g., parenteral agents alone for fewer days). A pragmatic way forward may be to start with patients who receive intubation for acute traumatic or hemorrhagic brain injury. This population is at very high risk for nosocomial infections, and the potential mortality benefit seen in patients with acute brain injury in the SuDDICU trial echoes randomized trials suggesting that single-dose ceftriaxone prophylaxis may reduce mortality among patients with acute traumatic or hemorrhagic brain injury.9,10 Focusing on this population may help additional ICUs gain experience and may perhaps crystallize how best to deploy this controversial yet still promising practice.
