Online supplement
Letter LDB 050517
Selective Decontamination of the Digestive tract halves the prevalence of Ventilator–associated Pneumonia compared to Selective Oral Decontamination
Lieuwe D. Bos1,2, Cheryl Stips1, Laura R Schouten1, Lonneke van Vught5, Maryse A. Wiewel4,5, Luuk Wieske1, Roosmarijn van Hooijdonk1, Marleen Straat1, Friso M. de Beer1,2, Gerie Glas1, Caroline E. Visser4, Evert de Jonge6, Nicole P. Juffermans1,2,Janneke Horn1, Marcus J. Schultz1,2
Academic Medical Center, Amsterdam, the Netherlands:
1Department of Intensive Care
2Laboratory for Experimental Intensive Care & Anesthesiology (L·E·I·C·A)
4Department of Medical Microbiology
5Center for Experimental and Molecular Medicine 7Division of Infectious Diseases
Leiden University Medical Center, Leiden, the Netherlands:
6Department of Intensive Care
Online supplement
Supplemental methods
Design
This was a prospective observational cohort study on the epidemiology of sepsis and sepsis-related adverse events. SDD was standard of care in the ICU of the Academic Medical Center (AMC), Amsterdam, the Netherlands, the center of which the data were used for the present analysis. TheInstitutional review board granted the request to use the so–called opt–out consent method in that project. The prospective data collection on VAP coincided with the randomization of our intensive care unit to a period of SOD for a study in which the effectof SOD and SDD on antibiotic resistance and patient outcome was tested [2]. For this study we had data on VAP occurrence during 6 months of SDD, 12 months of SOD and 6 months of SDD, consecutively. Patients with an ICU length of stay of less than 2 days were excluded, as these patients neither received selective decontamination nor were at risk for VAP. Patients with VAP on admission (e.g.,patients who were transferred from other ICU where they were already ventilated) were also excluded.Patients who received SDD in the two separate periods were grouped and analyzed as one group. Data from patients in the 1–month wash–in and the 1–month wash–out between the two decontamination regimens were not used in the analysis, as described previously[2].
SOD and SDD protocols
SOD consisted of a paste containing colistin, tobramycin, and amphotericin B, each in a 2% concentration, which wasapplied every 6 hours. With SDD, in addition to the application of this paste, a 10–mL suspension containing 100 mg colistin, 80 mg tobramycin, and 500 mg amphotericin B was administered via the nasogastric tube to decontaminate the gut. Also with SDD, 1 g cefotaxime was administered intravenously every 4 hours during the first 4 days in the ICU. Patients with clinically suspected or documented infection were treated according to standard clinical practice. The use of amoxicillin, penicillin, amoxicillin–clavulanic acid, and carbapenems was discouraged during the SDD period. Surveillance cultures were obtained to monitor the effectiveness of the regimen and consisted of endotracheal aspirates and oropharyngeal swabs every week.
VAP registration
The occurrence of hospital–acquired infections, including sepsis and VAP, was prospectively scored by trained MARS researchers on a daily basis[3]. The likelihood of VAP was scored,partly based on the Clinical Pulmonary Infection Score (CPIS)[5]. A CPIS belowseven resulted in a likelihood of ‘none’; a CPIS of seven or higher combined with purulent tracheal aspirate, positive semi–quantitative cultures or positive gram–stain and an abnormal chest radiograph resulted in VAP likelihood of ‘possible’; bacteremia, or more than 104 colony forming units per ml bronchoscopic aspirate, or detection of a pathogen in pleura fluid, or histopathological evidence of pneumonia but negative cultures resulted in a VAP likelihood of ‘probable’;a diagnosis of VAP was ‘definite’ when there was radiological evidence of lung abscesses or empyema with positive cultures or histopathological evidence of pneumonia with positive cultures. Notably, there was no diagnostic protocol for microbiological confirmation of a clinical suspicion of VAP. Consequently broncho-alveolar lavage fluid for microbiologic testing was not frequently obtained.
A cleaning algorithm that alarmed the researchers when there was inconsistency with other clinical variables (e.g., hospital–acquired pneumonia and community–acquired pneumonia are more likely in the first days of ICU admission) was used to check all data regarding VAP and to increase the inter–observer agreement from 69% to 85%[3].
Outcomes
The primary outcome used in this study was the composite of‘possible’, ‘probable’ and ‘proven’VAP. We performed a sensitivity analysis that included all patients that were treated for VAP, irrespective of the likelihood of VAP, as this was frequently used in literature on SDD or SOD [6, 7]. Second, we analyzed only patients that developed VAP after four days in the ICU, as this marked the end of systemic antibiotic therapy in the SDD regimen. Third, ICU related nosocomial pneumonia was analysed, thus including pneumonia not related to mechanical ventilation.
Structured risk analysis
The following risk factors for VAP [8]were investigated: age, sex, neurosurgery, thoracic surgery, COPD as described in the patients history, highest summed sequential organ failure assessment (SOFA) score[9], coma at admission defined by the Glasgow Coma Scale[10], and a diagnosis of acute respiratory distress syndrome (ARDS)[11]. All had to be present before the start of VAP. The protocol for feeding and patient positioning did not change during the study period. All mechanically ventilated patients received aerosol treatment with salbutamol and N–acetyl cysteine four times daily. Cuff pressures were checked only once a day. The ventilator circuit was changed weekly.
Data analysis
Data reporting was performed according to STROBE guidelines for reporting observational studies[12].Data was summarized using the mean and standard deviation for normally distributed continuous variables and with a median and 25–75th percentile for the rest. Categorical variables were summarized with count and percentage. Hazard ratios were presented with 95% confidence intervals. As this was an analysis of data collected during two independent studies there was no formal sample size calculation. Differences between the groups were compared using the Mann–Whitney U or Student–T test for continuous variables and chi–square for categorical variables. All analyses were performed in R statistics using R studio [13]. P–values below 0.05 were considered significant.
The prevalence of VAP was visualized per month during the three consecutive periods. As VAP can only develop before the occurrence of death and while being admitted to the ICU for ventilation and the decontamination regimens have an effect on mortality and/or length of ICU stay a competing risk Cox–regression model was used to correct for this. Risk factors for VAP could also have change over time and therefore a multivariate approach was used as well. The direct effects of decontamination strategy on VAP prevalence was analyzed by multivariate competing risk Cox–regression as described before[14]. Cause specific hazard–ratios (CSHR) were estimated using the ‘CSC’ function and sub distribution hazard–ratios (SDHR) were estimated with Fine–Gray regression using the ‘FGR’ function, both available via the package ‘riskRegression’ for R. The former gives a better estimation of the hazard-ratio per cause for each of the variables of interest while the latter more accurately estimates the cumulative incidence[15]. Multiple imputation chained equations were used to limit the influence of missing data (EMV scores were missing in 4 patients) [16]. Data were imputed 5 times, with 5 alterations, and the fitted models were averaged. A sensitivity analysis was performed by repeating above described steps for (1) patients that were suspected of VAP irrespective of the likelihood; (2) patients that developed VAP after day 4; (3) patients that developed any nosocomial pneumonia.
Supplemental results
Three hundred fifty–one patients were included in the first period of SDD, 807from the SOD period and 484 from the second SDD period, resulting in a total of 1,639 patients. Table S1 shows the baseline characteristics.
One hundred thirty seven(8.3%) patients were suspected of VAP during their stay in ICU. Of them, 69 patients (4% of included population; 50% of suspected patients) had a posterior likelihood of VAP of possible (N = 46, 67%), probable (N = 20, 29%) or proven (N = 3, 4%), which were combined in the primary outcome in this study. The median time to VAP was 6 [3 – 11] days after ICU admission. The most prevalent causative organism was Staphylococcus aureus (17%) and a wide variety of gram–negative bacteria was isolated from airway material and blood cultures (Table S2).
Table S3 shows the univariate distribution of risk factors between patients with and without VAP. Patients with VAP had undergone neurosurgery more often (P0.001), had higher SOFA scores (P0.001), had lower Glasgow Coma Scale scores (P=0.007) and were more frequently diagnosed with ARDS (P =0.002). The known risk factors for VAP, age, gender, preceding thoracic surgery and a history of COPD were not different between patients who did and did not develop VAP in the univariate analysis. Patients with VAP were more likely to die during their stay in ICU (P = 0.003) and had fewer days free of MV and alive at day 28 (P < 0.001). More patients who developed VAP had received SOD then SDD regimen (N = 46 vs. N = 23, P=0.002) (Figure S2). In univariate competing risk analysis SOD treatment corresponded to an unadjusted cause specific hazard ratio of 2.0 (95%–CI: 1.2 – 3.4) and a subdistribution hazard ratio 2.1 (95%–CI 1.3 – 3.5) of for VAP development when compared to SDD treatment.
SupplementalReferences
1. Oostdijk EAN, Kesecioglu J, Schultz MJ, et al. (2017) Notice of Retraction and Replacement: Oostdijk et al. Effects of Decontamination of the Oropharynx and Intestinal Tract on Antibiotic Resistance in ICUs: A Randomized Clinical Trial. JAMA . 2014;312(14):1429-1437. JAMA 317:1583. doi: 10.1001/jama.2017.1282
2. Oostdijk EAN, Kesecioglu J, Schultz MJ, et al. (2014) Effects of Decontamination of the Oropharynx and Intestinal Tract on Antibiotic Resistance in ICUs. JAMA. doi: 10.1001/jama.2014.7247
3. Klouwenberg PMCK, Ong DSY, Bos LDJ, et al. (2013) Interobserver Agreement of Centers for Disease Control and Prevention Criteria for Classifying Infections in Critically Ill Patients. Crit Care Med 41:2373–2378. doi: 10.1097/CCM.0b013e3182923712.
4. de Smet AM, Kluytmans JA, Cooper BS, et al. (2009) Decontamination of the digestive tract and oropharynx in ICU patients. N Engl J Med 360:20–31. doi: 360/1/20 [pii]10.1056/NEJMoa0800394
5. Pugin J, Auckenthaler R, Mili N, et al. (1991) Diagnosis of ventilator-associated pneumonia by bacteriologic analysis of bronchoscopic and nonbronchoscopic “blind” bronchoalveolar lavage fluid. Am Rev Respir Dis 143:1121–1129.
6. Schultz MJ, de Jonge E, Kesecioglu J (2003) Selective decontamination of the digestive tract reduces mortality in critically ill patients. Crit Care 7:107–10.
7. Schultz M, Haas L (2011) Antibiotics or probiotics as preventive measures against ventilator-associated pneumonia: a literature review. Crit Care 15:R18.
8. Weinstein RA, Bonten MJM, Kollef MH, Hall JB (2004) Risk Factors for Ventilator-Associated Pneumonia: From Epidemiology to Patient Management. Clin Infect Dis 38:1141–1149. doi: 10.1086/383039
9. Vincent JL, Moreno R, Takala J, et al. (1996) The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 22:707–710.
10. Teasdale G, Jennett B (1974) Assessment of coma and impaired consciousness. A practical scale. Lancet 2:81–84. doi: S0140-6736(74)91639-0 [pii]
11. Bernard GR, Artigas A, Brigham KL, et al. (1994) The American-European Consensus Conference on ARDS. Definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 149:818–824.
12. Field N, Cohen T, Struelens MJ, et al. (2014) Strengthening the Reporting of Molecular Epidemiology for Infectious Diseases (STROME-ID): an extension of the STROBE statement. Lancet Infect Dis 14:341–52. doi: 10.1016/S1473-3099(13)70324-4
13. R Development Core Team. R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing. Book R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing. 2010. (Editor ed.^eds.). City.
14. Melsen WG, Rovers MM, Groenwold RHH, et al. (2013) Attributable mortality of ventilator-associated pneumonia: a meta-analysis of individual patient data from randomised prevention studies. Lancet Infect Dis 13:665–671. doi:
15. Andersen PK, Geskus RB, de Witte T, Putter H (2012) Competing risks in epidemiology: possibilities and pitfalls. Int J Epidemiol 41:861–70. doi: 10.1093/ije/dyr213
16. Buuren S van, Groothuis-Oudshoorn K (2011) mice: Multivariate Imputation by Chained Equations in R. J. Stat. Softw.
SupplementalTables
Table S1: Patient characteristics
SODN=807 / SDD N=835 / P-value
Age Median (IQR) / 62 (50-71) / 62 (50-71) / 0.87
MaleN (%) / 487 (60) / 511 (61) / 0.77
APACHE IV scoreMedian (IQR) / 70 (53-89) / 71 (53-92) / 0.40
Highest SOFA sum scoreMedian (IQR) / 9 (6-11) / 8 (6-11) / 0.23
Specialty N (%) / Surgery / 180 (22) / 201 (24) / 0.30
Cardiothoracic surgery / 111 (14) / 120 (14)
Neurosurgery / 107 (13) / 90 (11)
Neurology / 48 (6) / 50 (6)
Internal medicine / 173 (21) / 159 (19)
Cardiology / 125 (15) / 131 (16)
Pulmonology / 26 (3) / 45 (5)
Other / 36 (5) / 39 (5)
Admission Type
N (%) / Medical / 486 (60) / 514 (62) / 0.60
Planned surgical / 153 (19) / 152 (18)
Emergency surgical / 166 (21) / 169 (20)
Admission Source N (%) / Emergency dep. / 264 (33) / 273 (33) / 0.77
Nursing ward / 210 (26) / 198 (24)
Operating theatre / 168 (21) / 201 (24)
ICU other hospital / 62 (8) / 63 (8)
Coronary care facility / 41 (5) / 44 (5)
Medium care facility / 43 (5) / 43 (5)
Other / 16 (2) / 10 (1)
Comorbidities / COPD / 70 (9) / 91 (11) / 0.14
Diabetes mellitus / 133 (16) / 108 (13) / 0.07
Chronic RRT / 11 (1) / 21 (3) / 0.10
Chronic renal insufficiency / 67 (8) / 92 (11) / 0.07
Metastatic malignancy / 16 (2) / 16 (2) / 1.0
Liver cirrhosis / 15 (2) / 22 (3) / 0.31
Immune deficiency / 98 (12) / 92 (11) / 0.50
Chronic cardiovascular insufficiency / 9 (1) / 7 (1) / 0.63
Antibiotics
(mean days per patient) / Total / 8.1 / 8.6 / 0.51
Carbapenems / 0.2 / 0.2 / 0.77
Cephalosporines / 2.7 / 3.1 / 0.02
Lincosamides / 0.7 / 0.4 / 0.01
Penicillines / 1.8 / 1.4 / 0.079
Quinolones / 0 / 0 / 0.55
Other / 4.8 / 4.2 / 0.35
ICU mortality N (%) / 148 (18) / 125 (15) / 0.08
Data is shown as number with percentage for categorical data or as median with inter-quartile range for continuous data.
Table S2:
VAP N (%) / 69 (4)Time to VAP days median (25-75th percentile) / 6 (3-11)
Likelihood
N (%) / Possible / 46 (67)
Probable / 20 (29)
Proven / 3 (4)
Isolated pathogen
N (%) / Staphylococcus aureus / 12 (17)
Streptococcus species / 1 (1)
Haemophilus influenzae / 5 (7)
Pseudomonas species / 4 (6)
Klebsiella species / 5 (7)
Enterococcus species / 2 (3)
Citrobacter species / 2 (3)
Enterobacter species / 3 (4)
Stenotrophomonas maltophilia / 3 (4)
Proteus mirabilis / 2 (3)
Unknown / 27 (39)
Table S3: Univariate analysis
No VAPN=1,573 / VAP
N=69 / P-value
Age, Median (IQR) / 62 (50-71) / 59 (47-70) / 0.18
Male, N (%) / 951 (60.5) / 48 (69.6) / 0.13
Neurosurgery, N (%) / 179 (11.4) / 18 (26.1) / 0.001
Thoracic surgery, N (%) / 219 (13.9) / 11 (15.9) / 0.72
Highest SOFA sum,Median (IQR) / 8 (6-11) / 11 (9-14) / <0.001
Glasgow Coma Scale,Median (IQR) / 14 (9-15) / 13 (4-15) / 0.14
ARDS, N (%) / 371 (23.6) / 28 (40.6) / 0.002
COPD,N (%) / 154 (9.8) / 8 (11.6) / 0.67
Regimen,N (%) / SDD / 814 (51.7) / 23 (33.3) / 0.002
SOD / 759 (48.3) / 46 (66.7)
ICU–free days and alive at day 28, N (IQR) / 23 (15-25) / 1 (0-16) / <0.001
Death in ICU, N (%) / 251 (16) / 21 (30) / 0.003
Data is shown as number with percentage for categorical data or as median with inter-quartile range for continuous data. Abbreviations: VAP = ventilator-associated pneumonia; SOFA = sequential organ failure assessment; ARDS = acute respiratory distress syndrome; COPD = chronic obstructive pulmonary disease; SDD = selective decontamination of the digestive tract; SOD = selective oropharyngeal decontamination; ICU = intensive care unit.
Table S4: Antibiotic exposure per 1000 ICU days.
Antibiotic group / SOD / SDD / Difference / P-valueCarbapenems / 23 / 30 / +7 / 0.70
Cephalosporines / 296 / 388 / +92 / <0.001
Lincosamides / 10 / 12 / +2 / 0.77
Penicillines / 194 / 179 / -15 / 0.23
Quinolones / 68 / 38 / -30 / 0.006
Other / 445 / 435 / -10 / 0.89
Others mainly include vancomycin, gentamycin and co-trimoxazole.
Supplemental figures
Figure S1:Screening and inclusion of patients.
All patients with an expected stay on the ICU of more than 24 hours were screened. Exclusion criteria were a shorted ICU-stay than 2 days or ventilator-associated pneumonia at admission. Patients were included during two years and grouped based on the decontamination strategy that was used in the ICU at that time. Patients that were admitted during the wash in or wash out were not used for statistical analyses.
Figure S2: Prevalence of VAP per month
The prevalence of VAP per month of inclusion. ). Left panel: suspected VAP (sensitivity analysis). Right panel: confirmed VAP (primary endpoint. The grey bars give the prevalence in months that were used for inclusion. The white bars give the prevalence during wash in and wash out months, which are also indicated by the dashed vertical lines. The horizontal solid black lines indicate the prevalence of VAP during the three periods of SDD, SOD and SDD, respectively.
Figure S3:Hazard ratios from competing risk analysis for multiple risk factors and all patients with possible, probable and proven VAP that developed after day 4.
The figures display the cause-specific and sub-distribution hazard ratios (HR) for VAP of multiple risk factors. The mean hazard ratio (rectangle) with confidence interval (horizontal bar) is displayed. Statistically significant associations do not cross the vertical dashed line. All values above 1.0, on the right side of the dashed line indicate an increased risk for VAP associated with the risk factor.
Figure S4: Hazard ratios from competing risk analysis for multiple risk factors and all patients with nosocomial pneumonia on the ICU, irrespective of mechanical ventilation.
The figures display the cause-specific and sub-distribution hazard ratios (HR) for ICU acquired pneumonia (ICUAP) of multiple risk factors. The mean hazard ratio (rectangle) with confidence interval (horizontal bar) is displayed. Statistically significant associations do not cross the vertical dashed line. All values above 1.0, on the right side of the dashed line indicate an increased risk for ICUAP associated with the risk factor.
1