Pneumonia Prevention to Decrease Mortality in Intensive Care Unit: a Systematic Review and Meta-Analysis

MAJOR ARTICLE

Pneumonia Prevention to Decrease Mortality in Intensive Care Unit: A Systematic Review and Meta-analysis

Antoine Roquilly,1 Emmanuel Marret,3 Edward Abraham,4 and Karim Asehnoune1,2 1Service d’Anesthésie Réanimation Hôtel-Dieu, Nantes University Hospital, 2UPRES EA3826, Faculty of Médecine of Nantes, and 3Department of Anesthesiology and Critical Care, Tenon University Hospital, University Pierre et Marie Curie, Paris, France; and 4Wake Forest School of Medicine, Winston-Salem, North Carolina (See the Editorial Commentary by Klompas on pages 76–8.)

Background. To determine the strategies of prevention of hospital-acquired pneumonia that reduce mortality in intensive care unit (ICU). Methods. We followed PRISMA (Preferred Reported Items for Systemic Reviews and Meta-Analyses) guidelines. We searched MEDLINE and the Cochrane Controlled Trials Register (through 10 June 2014) as well as reference lists of articles. We included all randomized controlled trials conducted in critically ill adult patients hospitalized in ICUs and evaluating digestive prophylactic methods (selective digestive decontamination [SDD], acidification of gastric content, early enteral feeding, prevention of microinhalation); circuit prophylactic methods (closed suctioning systems, early tracheotomy, aerosolized antibiotics, humidification, lung secretion drainage, silver-coated endotracheal tubes) or oropharyngeal prophylactic methods (selective oropharyngeal decontamination, patient position, sinusitis prophylaxis, subglottic secretion drainage, tracheal cuff monitoring). One reviewer extracted data that were checked by 3 others. The primary outcome was the mortality rate in the ICU. Results. We identified 157 randomized trials to pool in a meta-analysis. The primary outcome was available in 145 studies (n = 37 156). The risk ratio (RR) for death was 0.95 (95% confidence interval [CI], .92–.99; P = .02) in the intervention groups. In subgroup analysis, only SDD significantly decreased mortality compared with control (n = 10 227; RR, 0.84 [95% CI, .76–.92; P < .001]). The RR for in-ICU death was 0.78 (95% CI, .69–.89; P <.001; I2 = 33%) in trials investigating SDD with systemic antimicrobial therapy and 1.00 (.84–1.21; P = .96; I2 = 0%) without systemic antimicrobial therapy. Conclusions. Selective digestive decontamination with systemic antimicrobial therapy reduced mortality and should be considered in critically ill patients at high risk for death. Keywords. hospital-acquired pneumonia/prevention; mortality; selective digestive decontamination; mechanical ventilation.

Hospital-acquired pneumonia (HAP), and notably ven- important outcomes, including duration of mechanical tilator-associated pneumonia, developing as a conse- ventilation, length of stay in the intensive care unit quence of lung bacterial colonization, alters clinically (ICU), and mortality rates [1–3]. HAP is associated with use of antibiotics that may increase the risk of multiple-drug-resistant bacteria in ICUs, and the increase in Received 22 January 2014; accepted 4 August 2014; electronically published 24 medical costs is estimated to be to $20 000 per episode of September 2014. Correspondence: Karim Asehnoune, MD, PhD, CHU de Nantes, Service d’Anes- HAP [4]. Given HAP-associated morbidity, the preven- thésie Réanimation, 1 Pl Alexis Ricordeau, 44093 Nantes Cedex 1, France (karim. tion of HAP has been the focus of numerous studies in [email protected]). ® critically ill patients and remains a controversial issue. Clinical Infectious Diseases 2015;60(1):64–75 © The Author 2014. Published by Oxford University Press on behalf of the Infectious Bacterial colonization of the oropharynx and subse- Diseases Society of America. All rights reserved. For Permissions, please e-mail: quent microaspirations are initial events that lead to [email protected]. DOI: 10.1093/cid/ciu740 HAP [5–7]. The prevention of such events was therefore

64 • CID 2015:60 (1 January) • Roquilly et al proposed as a method for reducing the rate of HAP and presum- therapies, ulcer prophylaxis, aerosolized antibiotics, closed suc- ably for reducing associated morbidity and mortality. Three main tioning systems, early tracheotomy, humidification, phytotherapy approaches have been evaluated: (1) diminishing the microaspi- (ginger extract), physiotherapy, positive end-expiratory pressure ration of digestive flora; (2) reducing the volume of oropharyn- (PEEP), tracheal saline instillation, silver-coated endotracheal geal secretions aspirated into the lungs; and (3) inhibiting tubes), selective oropharyngeal decontamination (SOD), patient overgrowth or alterations in the microbiome in the oropharynx position, sinusitis prophylaxis, subglottic secretion drainage, and or larynx. Several meta-analyses of approaches to reducing the in- tracheal cuff monitoring. Studies using a cluster-randomization cidence of HAP conclude to a reduction in the risk of infection of procedure were included in the main analysis, but a sensitive such strategies. However, international recommendations for the analysis with exclusion of such studies was planned a priori. prevention of HAP provide different conclusions [8–11]anddo Pediatric patients were excluded from the study. not state which intervention is mandatory or superior to the others. The combination of several strategies for preventing HAP Data Extraction and Quality Assessment failed to improve mortality rates [12],and nonadherence to inter- One author (A. R.) checked all titles and abstracts of the articles national guidelines for HAP prevention is common [13]. identified from the database research and examined in full all We have hypothesized that all the preventive strategies do not randomized trials potentially eligible for the review. Quality as- equally alter the risk of death. Because many interventions are sessment for each study was performed by 2 unblinded investi- often poorly applied in clinical practice, it is important to deter- gators (A. R. and K. A.). Any disagreement among the 2 authors mine the most effective interventions that should be implemented was resolved by discussion. Persistent disagreement was settled in critically ill patients. We thus performed a systematic review to by discussion with other authors (E. M. and E. A.) after separate determine which method is the most effective for decreasing mor- review of the report. tality rates. The rates of HAP, duration of mechanical ventilation, One author (A. R.) designed a standard data extraction form, and ICU length of stay were also evaluated as secondary criteria. and other authors (E. M., E. A., and K. A.) amended and vali- dated the design of the form before abstraction of data. One au- MATERIALS AND METHODS thor (A. R.) extracted the following data from each eligible study: first author, year of publication, quality assessment by Data Source and Searches the Cochrane Collaboration’s risk of bias tool, type of interven- We followed PRISMA (Preferred Reported Items for Systemic tion, inclusion criteria, criteria used for HAP diagnosis, number Reviews and Meta-Analyses) guidelines were followed during of patients, number of HAP cases, duration of mechanical ven- the design and implementation of this meta-analysis (Supple- tilation and of ICU stay, and ICU mortality rate. Data were ex- mentary Table 1). We attempted to identify all relevant studies tracted from the tables, figures, text of the manuscript, and/or published in English regardless of publication status (published from previous meta-analysis that included the selected trial. or in press). We considered abstracts presented at scientific meetings <3 years earlier (Society Of Critical Care Medicine, Data Synthesis and Analysis European Society of Intensive Care Medicine, Societé Française The primary evaluation criterion was the rate of in-ICU deaths, that d’Anesthesie Reanimation, Societé de Reanimation de Langue is, the ICU mortality rate. When no information was available on Française). PubMed (MEDLINE/Index Medicus) and the in-ICU deaths, the rate of in-hospital death was considered if pro- Cochrane Controlled Trials Register were searched for studies vided. When trials had 2 control arms, the numbers of deaths in the published from January 1969 through 10 June 2014. The Med- control arms were pooled. The other end points analyzed included ical Subject Heading terms used for the search were pneumonia the incidence of HAP, the duration of mechanical ventilation, and and intensive care units with the limit “adult 19+ years.” The the ICU length of stay (or duration of hospitalization, as provided). “related articles” hyperlinks in MEDLINE were explored for additional references. The reference lists of all selected trials and Statistical Analyses previous published meta-analysis were checked for additional All statistical analyses were performed using Review Manager references. We contacted authors to identify unpublished data. software (version 5.1.6; Cochrane Collaboration; Nordic Cochrane Centre) or Stata software (version 10.1; StataCorp). Study Selection For dichotomous data (mortality and HAP), we calculated the The authors selected all randomized trials that evaluated any of risk ratio (RR) with 95% confidence intervals (CIs). To estimate the following strategies in adult patients (aged ≥18 years) hos- the clinical relevance of a beneficial effect on mortality, we cal- pitalized in ICUs: acidified enteral feeding, selective digestive culated the number needed to treat with 95% CIs, using RR and decontamination (SDD), early enteral feeding, postpyloric control event rates if the RR was significant. For continuous enteral feeding, decreased gastric retention, probiotic/symbiotic data, weighted mean differences with 95% CIs were calculated.

Pneumonia Prevention • CID 2015:60 (1 January) • 65 Heterogeneity was determined using the I2 statistics. All pooled the control group and the effect of SDD. Differences were con- estimates used the random effects model. A sensitivity analysis was sidered statistically significant at P < .05. achieved with the exclusion of cluster-randomized trial [14]. In exploratory analyses, we compared the effects of SDD on Role of the Funding Source mortality rates, based on the administration of systemic antimi- None of the institutions/sponsors of individual author had any crobial therapy (or not), population (trauma and surgical vs role in the design and conduct of the study; collection, management, medical or mixed population), sample size (>200 or <200 pa- analysis, and interpretation of the data; or preparation, review, or tients), and risk of bias (double blinding and randomization). approval of the manuscript. The corresponding author had full ac- To evaluate publication bias for trials that studied SDD, we con- cess to all the data in the study and had final responsibility for the structed a funnel plot and carried out Egger regression intercept decision to submit the manuscript for publication. and Begg rank correlation tests to assess asymmetry. The precision of each trial was evaluated according to the standard error RESULTS of logarithm RR. Egger test and Begg rank test results were considered significant at P < .10 because these tests have low power. Identification of the Trials We finally performed a meta-regression analysis to explore the Figure 1 presents the overall search approach. Our initial possible interaction between the mortality rate (baseline risk) in searches identified 1113 studies in MEDLINE and 465 studies

Figure 1. Literature search and study selection. Asterisks indicate studies providing the primary outcome, intensive care unit (ICU) mortality rate. Ab- breviations: PEEP, positive end-expiratory pressure; SDD, selective digestive decontamination; SOD, selective oropharyngeal decontamination. aTotal 146 studies because one study comparing 1 control group to 2 intervention groups (1 digestive method and 1 oro-pharyngeal method) was counted twice.

66 • CID 2015:60 (1 January) • Roquilly et al in the Cochrane Controlled Trials Register. After duplicate titles groups combined for analysis. For 2 trials with a 2 × 2 design, and abstracts were removed, 1220 citations were excluded (1041 the 3 intervention groups (two groups with one of the inter- nonrandomized trials and 179 irrelevant citations). The remain- ventions and one group with the association of the two ing 278 studies were considered potentially eligible and were interventions) were combined and compared with the double- accessed for full-text review. After exclusion of 117 trials (103 dummy (none of the intervention) group. One trial with a nonrandomized, 8 without outcomes of interest, 6 with ab- 2 × 2 design compared patients who received continuous sub- stracts >3 years old), we included 157 evaluable trials, of glottic secretion drainage or control and ulcer prophylaxis which 145 provided the mortality rate. with either H2-receptor antagonist or aluminate, no placebo was used for ulcer prophylaxis, and only 2 arms were thus con- Description of Recorded Outcomes and Tested Interventions sideredfortheanalyses(subglotticsecretiondrainagevs Supplementary Table 2 summarizes the characteristics of the control). included clinical trials. Fifteen trials (9.6%) had >2 arms: 7 (5%) compared 1 control group with 2 interventional groups Description of Participants combined for analysis, 1 (1%) compared 1 control group with Fifteen studies (9.6%) included medical ICU patients, 46 2 interventional groups considered separately for the analysis; (29.2%) included surgical or trauma ICU patients, and 96 5 (3.2%) compared 1 interventional group with 2 control (61.1%) included mixed ICU populations.

Figure 2. Hospital mortality rates in critically ill patients receiving a strategy for preventing hospital-acquired pneumonia. All pooled estimates used the random effects model. Boldface P values indicate signiﬁcant differences (P < .05). Abbreviations: CI, conﬁdence interval; ET, endotracheal tube; NA, not applicable; PEEP, positive end-expiratory pressure; RCT, randomized controlled trial; SDD, selective digestive decontamination; SOD, selective oropharyngeal decontamination.

Pneumonia Prevention • CID 2015:60 (1 January) • 67 Risk of Bias in the Included Studies enteral feeding (P = .02), PEEP versus no PEEP (P = .02), tra- The risks for bias are presented in Supplementary Table 3.Allstu- cheal saline instillation versus standard secretion drainage dies were randomized; the risk of bias for allocation concealment (P = .009), aerosolized antibiotic versus placebo (P =.04),sil- was low in 98 studies (62%). The risks of performance (double- ver-coated versus classic endotracheal tube (P = .002), SOD ver- blind study) and detection bias (blinding of outcome assessment) sus standard oral care (P < .001), and subglottic versus routine were low in 52 (33.1%) and 89 (56.7%) of the studies, respectively. secretion drainage (P < .001). The duration of mechanical ventilation was reported in 85 Effect of Interventions studies (23 691 patients) and showed high heterogeneity across Primary Outcome: In-ICU Deaths studies (I2 = 84%). The duration of mechanical ventilation was We computed data from 145 randomized trials (37 156 pa- reduced in the intervention groups compared with the control tients). The in-ICU mortality rate was 4079 of 18 838 (21.6%) groups (weighted mean difference, −0.75 days (95 CI%, −1.16 in the experimental groups versus 4160 of 18 318 (22.7%) in to −.35; P < .001). When each intervention is analyzed sepa- controls (RR, 0.95 [95% CI, .92–.99; P = .02; I2 = 4%]; number rately (Figure 5), the duration of mechanical ventilation was re- needed to treat, 89 [95% CI, 50–443]). The results for each in- duced in trials evaluating SDD versus standard care or placebo tervention analyzed separately are provided in Figure 2 (and (P = .003) and physiotherapy versus standard care (P = .03). Supplementary Figure 1 for forest plots of each intervention The duration of the ICU stay was reported in 82 studies separately). Except SDD, no tested strategy significantly de- (22 718 patients). It was reduced in the intervention groups creased ICU mortality rates. compared with the control groups (weighted mean difference, In the 30 trials (10 227 patients) investigating SDD versus −1.13 days [95 CI%, −1.70, −.57; P <.001;I2 = 89%]). When standard care or placebo, the in-ICU mortality rate for treated each intervention is analyzed separately (Figure 6), the duration vs control patients was 1051 of 5326 (19.7%) versus 1143 of of ICU stay was reduced in trials evaluating phytotherapy (gin- 4901 (23.3%) (RR, 0.84 [95% CI, .76–.92]; number needed to ger extract) versus placebo (P < .001) or SOD versus standard treat, 27 [95% CI, 18–54]) (Figure 3A)[14–43]. The benefitof oral care (P = .03). SDD for in-ICU mortality rates remains significant after exclusion of the cluster-randomized trial (603 of 3209 [18.8%] vs 686 DISCUSSION of 2880 [23.8%]; RR, 0.82 [95% CI, .68–.95, P < .001]) (Supple- mentary Table 4). The funnel plot showed no evidence for For this systematic review, we performed an extensive literature asymmetry and does not suggest publication bias or other search with few limits regarding publication status, thereby bias (Egger test, P = .17; Begg test, P = .69) (Figure 3B). In sub- minimizing the risk of missing important studies. This meta- group analyses, the RR for in-ICU death was 0.78 (95% CI, analysis shows that SDD is the main intervention decreasing .69–.90; P < .001; I2 = 40%) in trials investigating SDD with sys- the mortality rate in critically ill patients. temic antimicrobial therapy and 1.01 (95% CI, .85–1.21; P = .91; It is widely accepted that HAP increases morbidity and mor- I2 = 0%) in trials without systemic antimicrobial therapy (Sup- tality rates in critically ill patients [45, 46]. In the current results, plementary Tables 4 and 5). Given the risk of selection of resis- there is a paradox between the sharp reduction in the rate of tant bacteria with antimicrobial therapies [44], defining the HAP and the small reductions in death rates and time on ven- population in which SDD is more effective is an important in tilators [47]. This paradox may be attributable to the systemic assessing the use of such therapy. We thus investigated the as- diffusion of prophylactic antibiotics used for the SDD [48, sociation between the effects of SDD and mortality rate in the 49], which can alter the accuracy of the bacteriological samples control group (a proxy for baseline risk). In meta-regression and thus increase the rate of false-negatives in the diagnosis of analysis, there was a trend toward a positive interaction between HAP. Moreover, the diagnosis of HAP has a low specificity and the rate of death in the control groups and benefit from SDD high interobserver variability [50], and there are major discrep- (P = .07; Figure 3C). ancies between the clinical diagnoses of HAP and the autopsy findings in critically ill patients [51, 52]. We thus used mortality Secondary Outcomes rate as the primary outcome in this meta-analysis, because it al- The HAP rate was reported in 148 trials (28 856 patients); HAP lows us to control the risk of detection bias associated with the was diagnosed in 2156 of 14 457 patients (14.9%) assigned to diagnosis of HAP, notably in open-label trials. treatment versus 2994 of 13 799 (21.7%) who received control Previous reviews and meta-analyses have already reported a therapy (RR, 0.69 [95% CI, .63–.75; P < .001]; χ2 =391 reduction in the risk of death with SDD [53]. In clinical practice, [P < .001; I2 = 63%]). When each intervention is analyzed sepa- however, SDD is rarely used in ICUs [54] and was not even con- rately (Figure 4), the RR for HAP was reduced in trials evaluat- sidered in a recent European survey that evaluated the approach- ing SDD versus placebo (P < .001), postpyloric versus gastric es for HAP prevention in ICUs [55]. The most likely explanation

68 • CID 2015:60 (1 January) • Roquilly et al Figure 3. Intensive care unit (ICU) mortality rates in trials evaluating selective digestive decontamination. A, Forest plot for mortality in randomized clinical trials evaluating the effects of digestive decontamination in critically ill patients [14–43]. B, Funnel plot for mortality rates in randomized controlled trials depicts the effects of digestive decontamination on mortality. The x-axis represents the effect; the y-axis, precision. C, Meta-regression analysis exploring the interaction between survival beneﬁt with digestive decontamination and the incidence of mortality in the control group (y = −0.83x + 1.107; P = .07). Abbreviations: CI, conﬁdence interval; M-H, Mantel-Haenszel; RR, risk ratio; SE (Log RR), standard error of the logarithm of the risk ratio.

Pneumonia Prevention • CID 2015:60 (1 January) • 69 70 • I 056 1January) (1 2015:60 CID • oulye al et Roquilly

Figure 4. Hospital-acquired pneumonia in critically ill patients receiving a strategy for preventing hospital acquired pneumonia. All pooled estimates used the random effects model. Boldface P values indicate significant differences (P < .05). Abbreviations: CI, confidence interval; ET, endotracheal tube; NA, not applicable; PEEP, positive end-expiratory pressure; RCT, randomized controlled trials; SDD, selective digestive decontamination; SOD, selective oropharyngeal decontamination. Figure 5. Duration of mechanical ventilation in critically ill patients receiving a strategy for preventing hospital-acquired pneumonia. All pooled estimates used the random effects model. Boldface P values indicate significant differences (P < .05). Abbreviations: CI, confidence interval; ET, endotracheal tube; NA, not applicable; PEEP, positive end-expiratory pressure; RCT, randomized controlled trial; SDD, selective digestive decontamination; SOD, selective oropharyngeal decontamination; WMD, weighted mean difference. for the infrequent use of SDD is concern about the risk for emergence of resistance needs to be better investigated before emerging antimicrobial resistance during treatment [44, 56, we can rule out this potential side effect of SDD. 57]. However, in the largest trial of SDD [14], the percentages Given concerns regarding acquisition of bacterial resistance, of resistant bacteria causing ICU-acquired bacteremia were not it has been suggested that SDD may be used in selected popu- significantly altered in patients treated with SDD, compared with lations, but little information is available regarding the popula- controls. Moreover, in a recent meta-analysis including microbi- tions that could be eligible for SDD. In particular, medical ological data issue from 35 studies, no relationship was found versus surgical reasons for ICU admission are not reliable crite- between the use of SDD and development of antimicrobial- ria in selecting patients for SDD [59]. The result of the meta- resistance in pathogens [58]. In the subgroup of patients with in- regression (Figure 2C) suggests that SDD is more effective in fection due to gram-negative bacteria, SDD was associated with a patients with a high risk of death. In 2 systematic reviews, decreased risk of resistance to cephalosporins and quinolones SDD seemed particularly effective in patients with multiple [58]. This surprising result may be explained by the reduction organ dysfunction syndrome or with a high risk for bacteremia in the total daily doses of antibiotics that has been reported [60]. Taken together, these data suggest that the use of SDD with SDD [14]. The lack of a proof for increased antimicrobial could be limited to patients with a high risk of death, such as resistance in numerous studies suggests that the perceived risk those presenting with acute organ failure on admission. of selection of resistant bacteria does not justify the low rate of Several regimens of SDD have been proposed, and the use of SDD use in clinical practice. However, the long-term risk of a systemic antimicrobial therapy is one of the main differences

Pneumonia Prevention • CID 2015:60 (1 January) • 71 Figure 6. Duration of ICU stay for critically ill patients receiving a strategy for preventing hospital-acquired pneumonia. All pooled estimates used the random effects model. Boldface P values indicate signiﬁcant differences (P < .05). Abbreviations: CI, conﬁdence interval; ET, endotracheal tube; ICU, intensive care unit; NA, not applicable; PEEP, positive end-expiratory pressure; RCT, randomized controlled trial; SDD, selective digestive decontamination; SOD, selective oropharyngeal decontamination; WMD, weighted mean difference.

between them. We found that systemic antimicrobial therapy is This meta-analysis has some limitations. First, our study essential to a decrease in the risk of mortality. The recent dem- combined different types of interventions. This strategy of anal- onstration that the distal respiratory flora is not distinct from ysis has already been used to estimate the attributable mortality the upper airway microbiome in healthy lungs [61, 62] suggests of HAP [46, 63], and it will enable intensivists to prioritize strat- that HAP results from an increased population of distal lung egies for implementation in clinical practice. Second, the num- bacteria rather than colonization from the digestive tract. ber of patients in the SDD analyses is greater than the number This hypothesis can explain why the application of an oral for any other intervention, which results in a higher power for paste of nonabsorbable antibiotics, which has no effect on SDD than for other strategies. Thus, the current meta-analyses the lung microbiome, is not sufficient to prevent HAP and re- cannot rule out the efficiency of strategies with fewer included duce the risk of death. Finally, the current results argue that patients. Third, the prevention of ventilator-associated pneu- protocols of SDD should always include a short course of sys- monia was analyzed together with that of HAP. However, path- temic antimicrobial therapy and question the importance of ogens involved in HAP are similar regardless of whether or not the oral paste. HAP is acquired during mechanical ventilation, thereby

72 • CID 2015:60 (1 January) • Roquilly et al suggesting that preventive strategies should be efficient with in the critically ill patient. The Canadian Critical Trials Group. Am J 1999 – both HAP and ventilator-acquired pneumonia [64]. Finally, Respir Crit Care Med ; 159:1249 56. 3. Bekaert M, Timsit JF, Vansteelandt S, et al. Attributable mortality of the existing literature on SDD has mainly considered short- ventilator-associated pneumonia: a reappraisal using causal analysis. term impacts on morbidity and mortality rates. If SDD eventu- Am J Respir Crit Care Med 2011; 184:1133–9. ally leads to increased prevalence of resistant bacteria, then the 4. Safdar N, Dezfulian C, Collard HR, Saint S. Clinical and economic con- fi sequences of ventilator-associated pneumonia: a systematic review. Crit short-term bene ts could be reversed by increases in failure of Care Med 2005; 33:2184–93. curative antimicrobial therapy. 5. Bonten MJ, Gaillard CA, de Leeuw PW, Stobberingh EE. Role of colo- In conclusion, the current meta-analysis, which includes 157 nization of the upper intestinal tract in the pathogenesis of ventilator- 1997 – randomized studies, highlights the benefits preventing HAP in associated pneumonia. Clin Infect Dis ; 24:309 19. 6. Ewig S, Torres A, El-Ebiary M, et al. Bacterial colonization patterns in fi decreasing death rates in critically ill patients. When speci c in- mechanically ventilated patients with traumatic and medical head inju- terventions are considered separately, SDD with systemic anti- ry: incidence, risk factors, and association with ventilator-associated 1999 – microbial therapy should be considered first in critically ill pneumonia. Am J Respir Crit Care Med ; 159:188 98. 7. Bonten MJM. Ventilator-associated pneumonia and the gastropulmo- patients, mainly in those with acute organ failure and at high nary route of infection: a pendulum. Am J Respir Crit Care Med risk for death. Ongoing trials should (1) enhance the selection 2011; 184:991–3. of patients who may benefit from SDD and determine the min- 8. American Thoracic Society, Infectious Diseases Society of America. fi Guidelines for the management of adults with hospital-acquired, venti- imal ef cient duration of systemic antimicrobial therapy; lator-associated, and healthcare-associated pneumonia. Am J Respir (2) compare systemic antimicrobial therapy, with or without Crit Care Med 2005; 171:388–416. digestive decontamination; (3) determine the optimal antimi- 9. Torres A, Carlet J. Ventilator-associated pneumonia. European Task 2001 crobial therapy in patients treated with SDD who develop a Force on ventilator-associated pneumonia. Eur Respir J ; 17: 1034–45. subsequent infection; and (4) assess the long-term risk for 10. Healthcare Infection Control Practices Advisory Committee, Centers emergence of multiple-drug-resistant bacteria. for Disease Control and Prevention. Guidelines for preventing healthcare-associated pneumonia, 2003 recommendations of the CDC and the Healthcare Infection Control Practices Advisory Committee. Respir Supplementary Data Care 2004; 49:926–39. 11. Dodek P, Keenan S, Cook D, et al. Evidence-based clinical practice guideline for the prevention of ventilator-associated pneumonia. Ann Supplementary materials are available at Clinical Infectious Diseases online Intern Med 2004; 141:305–13. (http://cid.oxfordjournals.org). Supplementary materials consist of data 12. Bouadma L, Deslandes E, Lolom I, et al. Long-term impact of a multi- provided by the author that are published to benefit the reader. The posted faceted prevention program on ventilator-associated pneumonia in a materials are not copyedited. The contents of all supplementary data are the medical intensive care unit. Clin Infect Dis 2010; 51:1115–22. sole responsibility of the authors. Questions or messages regarding errors 13. Rello J. Why do physicians not follow evidence-based guidelines for should be addressed to the author. preventing ventilator-associated pneumonia? : a survey based on the opinions of an international panel of intensivists. Chest 2002; Notes 122:656–61. 14. de Smet AMGA, Kluytmans JAJW, Cooper BS, et al. Decontamination Author contributions. A. R., E. M., and K. A. designed and organized of the digestive tract and oropharynx in ICU patients. N Engl J Med research; acquired, analyzed and interpreted data; and wrote the manuscript. 2009; 360:20–31. Emmanuel Marret performed the statistical analysis. E. A. provided critical 15. Aerdts SJ, Clasener HA, van Dalen R, et al. Prevention of bacterial col- revision of the manuscript for intellectual content. E. A. and K. A. supervised onization of the respiratory tract and stomach of mechanically ventilat- the study. K. A. had full access to all the data in the study and takes respon- ed patients by a novel regimen of selective decontamination in sibility for the integrity of the data and the accuracy of the data analysis. All combination with initial systemic cefotaxime. J Antimicrob Chemother authors have agreed to submit data for publication. 1990; 26(suppl A):59–76. Disclaimer. The funders had no role in study design, data collection 16. Arnow PM, Carandang GC, Zabner R, Irwin ME. Randomized and analysis, decision to publish, or preparation of the manuscript. controlled trial of selective bowel decontamination for prevention of Financial support. This work was supported by institutional and infections following liver transplantation. Clin Infect Dis 1996; 22: departmental funds. 997–1003. Potential conflicts of interest. E. M. has been a consultant for Bayer. 17. Blair P, Rowlands BJ, Lowry K, Webb H, Armstrong P, Smilie J. Selective K. A. has been on the speakers’ bureau for Fresenius and B. Braun Medical decontamination of the digestive tract: a stratified, randomized, pro- and has been a consultant to Astellas Pharma. All other authors report no spective study in a mixed intensive care unit. Surgery 1991; 110:303–9. potential conflicts. 18. Cerra FB, Maddaus MA, Dunn DL, et al. Selective gut decontamination All authors have submitted the ICMJE Form for Disclosure of Potential reduces nosocomial infections and length of stay but not mortality or Conflicts of Interest. Conflicts that the editors consider relevant to the con- organ failure in surgical intensive care unit patients. Arch Surg 1992; tent of the manuscript have been disclosed. 127:163–7. 19. Cockerill FR, Muller SR, Anhalt JP, et al. Prevention of infection in crit- References ically ill patients by selective decontamination of the digestive tract. Ann Intern Med 1992; 117:545–53. 1. Mizgerd JP. Acute lower respiratory tract infection. N Engl J Med 2008; 20. de Jonge E, Schultz MJ, Spanjaard L, et al. Effects of selective decon- 358:716–27. tamination of digestive tract on mortality and acquisition of resistant 2. Heyland DK, Cook DJ, Griffith L, Keenan SP, Brun-Buisson C. The at- bacteria in intensive care: a randomized controlled trial. Lancet 2003; tributable morbidity and mortality of ventilator-associated pneumonia 362:1011–6.

Pneumonia Prevention • CID 2015:60 (1 January) • 73 21. de La Cal MA, Cerdá E, García-Hierro P, et al. Survival benefit in crit- 39. Stoutenbeek CP, van Saene HK, Little RA, Whitehead A; Working ically ill burned patients receiving selective decontamination of the di- Group on Selective Decontamination of the Digestive Tract. The effect gestive tract: a randomized, placebo-controlled, double-blind trial. Ann of selective decontamination of the digestive tract on mortality in mul- Surg 2005; 241:424–30. tiple trauma patients: a multicenter randomized controlled trial. Inten- 22. Ferrer M, Torres A, González J, et al. Utility of selective digestive decon- sive Care Med 2007; 33:261–70. tamination in mechanically ventilated patients. Ann Intern Med 1994; 40. Ulrich C, Harinck-de Weerd JE, Bakker NC, Jacz K, Doornbos L, 120:389–95. de Ridder VA. Selective decontamination of the digestive tract with nor- 23. Flaherty J, Nathan C, Kabins SA, Weinstein RA. Pilot trial of selective floxacin in the prevention of ICU-acquired infections: a prospective decontamination for prevention of bacterial infection in an intensive randomized study. Intensive Care Med 1989; 15:424–31. care unit. J Infect Dis 1990; 162:1393–7. 41. Unertl K, Ruckdeschel G, Selbmann HK, et al. Prevention of coloniza- 24. Fox MA, Peterson S, Fabri BM, van Saene HK. Selective decontamination and respiratory infections in long-term ventilated patients by local tion of the digestive tract in cardiac surgical patients. Crit Care Med antimicrobial prophylaxis. Intensive Care Med 1987; 13:106–13. 1991; 19:1486–90. 42. Verwaest C, Verhaegen J, Ferdinande P, et al. Randomized, controlled 25. Gastinne H, Wolff M, Delatour F, Faurisson F, Chevret S; French Study trial of selective digestive decontamination in 600 mechanically ventilat- Group on Selective Decontamination of the Digestive Tract. A con- ed patients in a multidisciplinary intensive care unit. Crit Care Med trolled trial in intensive care units of selective decontamination of the 1997; 25:63–71. digestive tract with nonabsorbable antibiotics. N Engl J Med 1992; 43. Wiener J, Itokazu G, Nathan C, Kabins SA, Weinstein RA. A random- 326:594–9. ized, double-blind, placebo-controlled trial of selective digestive decon- 26. Godard J, Guillaume C, Reverdy ME, et al. Intestinal decontamination tamination in a medical-surgical intensive care unit. Clin Infect Dis in a polyvalent ICU: a double-blind study. Intensive Care Med 1990; 1995; 20:861–7. 16:307–11. 44. Bouza E, Granda MJP, Hortal J, Barrio JM, Cercenado E, Muñoz P. Pre- 27. Hammond JM, Potgieter PD, Saunders GL, Forder AA. Double-blind emptive broad-spectrum treatment for ventilator-associated pneumonia study of selective decontamination of the digestive tract in intensive in high-risk patients. Intensive Care Med 2013; 39:1547–55. care. Lancet 1992; 340:5–9. 45. Rincón-Ferrari MD, Flores-Cordero JM, Leal-Noval SR, et al. Impact of 28. Kerver AJ, Rommes JH, Mevissen-Verhage EA, et al. Prevention of ventilator-associated pneumonia in patients with severe head injury. colonization and infection in critically ill patients a prospective ran- J Trauma 2004; 57:1234–40. domized study. Crit Care Med 1988; 16:1087–93. 46. Melsen WG, Rovers MM, Koeman M, Bonten MJM. Estimating the at- 29. Korinek AM, Laisne MJ, Nicolas MH, Raskine L, Deroin V, Sanson- tributable mortality of ventilator-associated pneumonia from random- Lepors MJ. Selective decontamination of the digestive tract in neurosur- ized prevention studies. Crit Care Med 2011; 39:2736–42. gical intensive care unit patients: a double-blind, randomized, placebo- 47. Klompas M. The paradox of ventilator-associated pneumonia preven- controlled study. Crit Care Med 1993; 21:1466–73. tion measures. Crit Care 2009; 13:315. 30. Krueger WA, Lenhart FP, Neeser G, et al. Influence of combined 48. Gastinne H, Wolff M, Lachatre G, Boiteau R, Savy FP. Antibiotic levels intravenous and topical antibiotic prophylaxis on the incidence of in bronchial tree and in serum during selective digestive decontamina- infections, organ dysfunctions, and mortality in critically ill surgical tion. Intensive Care Med 1991; 17:215–8. patients: a prospective, stratified, randomized, double-blind, placebo- 49. Mol M, van Kan HJM, Schultz MJ, de Jonge E. Systemic tobramycin controlled clinical trial. Am J Respir Crit Care Med 2002;166: concentrations during selective decontamination of the digestive tract 1029–37. in intensive care unit patients on continuous venovenous hemofiltra- 31. Lingnau W, Berger J, Javorsky F, Lejeune P, Mutz N, Benzer H. Selective tion. Intensive Care Med 2008; 34:903–6. intestinal decontamination in multiple trauma patients: prospective, 50. Klompas M. Interobserver variability in ventilator-associated pneumo- controlled trial. J Trauma 1997; 42:687–94. nia surveillance. Am J Infect Control 2010; 38:237–9. 32. Mandelli M, Mosconi P, Langer M, Cigada M. Prevention of pneumonia 51. Fábregas N, Ewig S, Torres A, et al. Clinical diagnosis of ventilator as- in an intensive care unit: a randomized multicenter clinical trial. Inten- sociated pneumonia revisited: comparative validation using immediate sive Care Unit Group of Infection Control. Crit Care Med 1989; post-mortem lung biopsies. Thorax 1999; 54:867–73. 17:501–5. 52. Tejerina E, Esteban A, Fernández-Segoviano P, et al. Accuracy of clinical 33. Pneumatikos I, Koulouras V, Nathanail C, Goe D, Nakos G. Selective definitions of ventilator-associated pneumonia: comparison with autop- decontamination of subglottic area in mechanically ventilated patients sy findings. J Crit Care 2010; 25:62–8. with multiple trauma. Intensive Care Med 2002; 28:432–7. 53. Liberati A, D’Amico R, Pifferi S, Torri V, Brazzi L, Parmelli E. Antibiotic 34. Quinio B, Albanèse J, Bues-Charbit M, Viviand X, Martin C. Selective prophylaxis to reduce respiratory tract infections and mortality in decontamination of the digestive tract in multiple trauma patients: a adults receiving intensive care. Cochrane Database Syst Rev 2009;3: prospective double-blind, randomized, placebo-controlled study. CD000022. Chest 1996; 109:765–72. 54. Bastin AJ, Ryanna KB. Use of selective decontamination of the digestive 35. Rocha LA, Martín MJ, Pita S, et al. Prevention of nosocomial infection tract in United Kingdom intensive care units. Anaesthesia 2009; in critically ill patients by selective decontamination of the digestive 64:46–9. tract: a randomized, double blind, placebo-controlled study. Intensive 55. Ricard J-D, Conti G, Boucherie M, et al. A European survey of nosoco- Care Med 1992; 18:398–404. mial infection control and hospital-acquired pneumonia prevention 36. Rodríguez-Roldán JM, Altuna-Cuesta A, López A, et al. Prevention practices. J Infect 2012; 65:285–91. of nosocomial lung infection in ventilated patients: use of an anti- 56. Salgado CD, O Grady N, Farr BM. Prevention and control of antimicro- microbial pharyngeal nonabsorbable paste. Crit Care Med 1990; 18: bial-resistant infections in intensive care patients. Crit Care Med 2005; 1239–42. 33:2373–82. 37. Sánchez García M, Cambronero Galache JA, López Diaz J, et al. Effective- 57. Naiemi Al N. Emergence of multidrug-resistant gram-negative bacteria ness and cost of selective decontamination of the digestive tract in critical- during selective decontamination of the digestive tract on an intensive ly ill intubated patients: a randomized, double-blind, placebo-controlled, care unit. J Antimicrob Chemother 2006; 58:853–6. multicenter trial. Am J Respir Crit Care Med 1998; 158:908–16. 58. Daneman N, Sarwar S, Fowler RA, Cuthbertson BH; SuDDICU Cana- 38. Silvestri L, van Saene HK, Milanese M, et al. Prevention of MRSA pneu- dian Study Group. Effect of selective decontamination on antimicrobial monia by oral vancomycin decontamination: a randomised trial. Eur resistance in intensive care units: a systematic review and meta-analysis. Respir J 2004; 23:921–6. Lancet infect dis 2013; 13:328–41.

74 • CID 2015:60 (1 January) • Roquilly et al 59. Melsen WG, de Smet A, Kluytmans JAJW, Bonten MJM; Dutch SOD- 62. Morris A, Beck JM, Schloss PD, et al. Comparison of the respiratory mi- SDD Trialists’ Group. Selective decontamination of the oral and diges- crobiome in healthy non-smokers and smokers. Am J Respir Crit Care tive tract in surgical versus non-surgical patients in intensive care in a Med 2013; 187:1067–75. cluster-randomized trial. Br J Surg 2012; 99:232–7. 63. Melsen WG, Rovers MM, Groenwold RH, et al. Attributable mortality 60. Silvestri L, van Saene HKF, Milanese M, Gregori D, Gullo A. Selective of ventilator-associated pneumonia: a meta-analysis of individual pa- decontamination of the digestive tract reduces bacterial bloodstream in- tient data from randomised prevention studies. Lancet Infect Dis fection and mortality in critically ill patients. Systematic review of ran- 2013; 13:665–71. domized, controlled trials. J Hosp Infect 2007; 65:187–203. 64. Esperatti M, Ferrer M, Theessen A, et al. Nosocomial pneumonia in 61. Charlson ES, Bittinger K, Haas AR, et al. Topographical continuity of the intensive care unit acquired by mechanically ventilated versus bacterial populations in the healthy human respiratory tract. Am J Re- nonventilated patients. Am J Respir Crit Care Med 2010; 182: spir Crit Care Med 2011; 184:957–63. 1533–9.

Pneumonia Prevention • CID 2015:60 (1 January) • 75