Articles Selective digestive tract decontamination and selective oropharyngeal decontamination and antibiotic resistance in patients in intensive-care units: an open-label, clustered group-randomised, crossover study Anne Marie G A de Smet, Jan A J W Kluytmans, Hetty E M Blok, Ellen M Mascini, Robin F J Benus, Alexandra T Bernards, Ed J Kuijper, Maurine A Leverstein-van Hall, Arjan R Jansz, Bartelt M de Jongh, Gerard J van Asselt, Ine H M E Frenay, Steven F T Thijsen, Simon N M Conijn, Jan A Kaan, Jan P Arends, Patrick D J Sturm, Martin C J Bootsma, Marc J M BontenSummary Background Previously, we assessed selective digestive tract decontamination (SDD) and selective oropharyngeal Published Online decontamination (SOD) on survival and prevention of bacteraemia in patients in intensive-care units. In this analysis, DOI:10.1016/S1473- we aimed to assess eﬀ ectiveness of these interventions for prevention of respiratory tract colonisation and bacteraemia 3099(11)70035-4 with highly resistant microorganisms acquired in intensive-care units.
See Online/Comment DOI:10.1016/S1473- Methods We did an open-label, clustered group-randomised, crossover study in 13 intensive-care units in the 3099(11)70067-6 Netherlands between May, 2004, and July, 2006. Participants admitted to intensive-care units with an expected Division of Perioperative and Emergency Care duration of mechanical ventilation of more than 48 h or an expected stay of more than 72 h received SOD (topical
(A M G A de Smet MD), tobramycin, colistin, and amphotericin B in the oropharynx), SDD (SOD antibiotics in the oropharynx and stomach Department of Medical plus 4 days’ intravenous cefotaxime), or standard care. The computer-randomised order of study regimens was applied Microbiology (H E M Blok MSc, by an independent clinical pharmacist who was masked to intensive-care-unit identity. We calculated crude odds M A Leverstein-van Hall MD, ratios (95% CI) for rates of bacteraemia or respiratory tract colonisation with highly resistant microorganisms in
Prof M J M Bonten MD), and patients who stayed in intensive-care units for more than 3 days (ie, acquired infection). This trial is registered at Julius Centre for Health Sciences http://isrctn.org, number ISRCTN35176830. and Primary Care (M J M Bonten), University Medical Centre Utrecht, Utrecht, Netherlands; Findings Data were available for 5927 (>99%) of 5939 patients, of whom 5463 (92%) were in intensive-care units Department of Intensive Care, for more than 3 days. 239 (13%) of 1837 patients in standard care acquired bacteraemia after 3 days, compared Onze Lieve Vrouwe Gasthuis, with 158 (9%) of 1758 in SOD (odds ratio 0·66, 95% CI 0·53–0·82), and 124 (7%) of 1868 in SDD (0·48, 0·38–0·60). Amsterdam, Netherlands Eight patients acquired bacteraemia with highly resistant microorganisms during SDD, compared (A M G A de Smet); Department of Medical Microbiology, with 18 patients (with 19 episodes) during standard care (0·41, 0·18–0·94; rate reduction [RR] 59%, absolute risk Amphia Hospital, Breda, reduction [ARR] 0·6%) and 20 during SOD (0·37, 0·16–0·85; RR 63%, ARR 0·7%). Of the patients staying in Netherlands intensive-care units for more than 3 days, we obtained endotracheal aspirate cultures for 881 (49%) patients (Prof J A J W Kluytmans MD); receiving standard care, 886 (50%) receiving SOD, and 828 (44%) receiving SDD. 128 (15%) patients acquired Department of Medical Microbiology and Infectious respiratory tract colonisation with highly resistant microorganisms during standard care, compared with 74 (8%) Diseases, Vrije Universiteit during SDD (0·58, 0·43–0·78; RR 38%, ARR 5·5%) and 88 (10%) during SOD (0·65, 0·49–0·87; RR 32%, Medical Centre, Amsterdam ARR 4·6%). Acquired respiratory tract colonisation with Gram-negative bacteria or cefotaxime-resistant and (J A J W Kluytmans); Laboratory colistin-resistant pathogens was lowest during SDD. for Medical Microbiology and Immunology, Rijnstate Hospital, Arnhem, Netherlands Interpretation Widespread use of SDD and SOD in intensive-care units with low levels of antibiotic resistance is (E M Mascini MD); Department justiﬁ ed. of Medical Microbiology, University Medical Centre, Groningen, Netherlands Funding None.
(R F J Benus MD, J P Arends MD); Department of Medical Introduction
ant pathogens, which is the main drawback of such Microbiology, Leiden University Medical Centre, Leiden,
Respiratory tract infections acquired in intensive-care interventions. SDD aims to prevent colonisation by
Netherlands (A T Bernards MD,
units have been associated with high rates of morbidity Gram-negative bacteria, Staphylococcus aureus, and yeasts E J Kuijper MD); Laboratories for
and mortality and increased health-care costs, especially in patients in intensive care through application of non- Pathology and Medical
when they are caused by highly resistant microorganisms
absorbable antimicrobial treatments in the oropharynx Microbiology, Catharina Hospital, Eindhoven,
(table 1).1–3 Reductions in the rates of respiratory tract and gastrointestinal tract. This intervention also pre-
Netherlands (A R Jansz MD);
infections have been achieved through prophylactic emptively treats possible infections with commensal Department of Medical
antibiotic regimens such as selective digestive tract respiratory tract bacteria through systemic administration Microbiology and Immunology,
decontamination (SDD)4,5 and selective oropharyngeal of cephalosporins during the ﬁ rst 4 days in intensive- St Antonius Hospital, Nieuwegein, Netherlands
decontamination (SOD).6,7 However, prophylactic use of care units, and maintains anaerobic intestinal ﬂ ora
(B M de Jongh MD);
anti biotics might increase selection for antibiotic-resist-
through selective use of antibiotics (both topically and
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles Department of Medical
systemically).8,9 SOD is oropharyngeal decontamination individualised, randomised design would have led to
Microbiology, Medical Centre
only, without any speciﬁ c recommendations for systemic
interference between treated and untreated patients.
Haaglanden, The Hague,
Therefore, cluster randomisation was needed and all
Netherlands (G J van Asselt MD); Regional Laboratory for Medical
We previously reported primary and secondary clinical three study regimens (SDD, SOD, and standard care)
Microbiology and Infectious
outcomes, including rates of survival, bacteraemia, were to be applied to all eligible patients during 6 months
antibiotic use, length of stay, and duration of mechanical with the order of regimens randomly assigned. The
and Albert Schweitzer Hospital,
ventilation, from a multicentre study10 comparing SOD study periods were preceded by a wash-in and/or wash-
(I H M E Frenay MD);
and SDD with standard care. Both interventions were out month during which newly admitted patients were
Department of Medical
associated with improved day-28 survival and reduced treated according to the centre-speciﬁ c allocation
incidence of acquired bacteraemia in 13 Dutch intensive-
treatment for the following period, but actual inclusion
care units with low rates of antibiotic resistance. In the started after the wash-in period (see supplementary
(S F T Thijsen MD); Department of Clinical Microbiology,
paper10 and a subsequent analysis11 we reported the eco-
online appendix of de Smet and colleagues10). Patients
logical eﬀ ects of SDD and SOD on the overall intensive-
admitted to intensive-care units with an expected
Amsterdam (S N M Conijn);
care-unit population including short-staying patients who duration of mechanical ventilation of more than 48 h or
Department of Medical
were not enrolled in the study. Compared with standard an expected stay of more than 72 h were eligible for
Microbiology and Immunology, Mesos Medical Centre, Utrecht
care, SDD and SOD were associated with decreased inclusion. Expected stay was assessed by doctors
(J A Kaan MD); and Department
mortality and reduced need for systemic antibiotics.10
responsible for care of patients in each unit.10
of Medical Microbiology,
Moreover, SDD was also associated with a reduction in
The study protocol was approved by institutional
Radboud University, University
occurrence of candidaemia and bacteraemia with Gram-
review boards at all participating hospitals. Because
Medical Centre Nijmegen, Nijmegen, Netherlands
negative microorganisms as compared with SOD.
SDD and SOD were regarded as common practice in
However, neither publication reported the eﬀ ects of the Netherlands and not harmful to patients, and
SDD and SOD on bacteraemia and respiratory tract because only a unit-wide approach would allow reliable
colonisation with highly resistant microorganisms or estimation of the true eﬀ ect of the interventions, the
bacteria resistant to the SOD and SDD antibiotics that institutional review boards waived, after review, the
were acquired while patients were in intensive-care units.
need for informed consent. We obtained permission to
We aimed to assess such outcomes in this study.
use patient-speciﬁ c medical data for analysis from
firstname.lastname@example.org Methods Study sites and patients Randomisation and masking
Our group did a pragmatic, open-label, clustered group-
A clinical pharmacist, who was not involved in care of
randomised, controlled, crossover study in 13 intensive-
patients at any of the participating units, did the
care units between May, 2004, and July, 2006. The randomisation and was masked to intensive-care unit participating intensive-care units diﬀ ered in size and identity. The order of the three study regimens provided teaching status, covering all levels of intensive care in to patients was randomly generated by computer the Netherlands. Because the interventions included software and allocated to the wards in consecutive ecological changes in the intensive-care units, an order of study start. Quinolones Aminoglycosides Carbapenems Co-trimoxazole Ceftazidime Piperacillin Penicillins Glycopeptides Oxacillin Meticillin Enterobacteriaceae Glucose non-fermenting Gram-negative rods Gram-positive bacteria
Resistance to one antibacterial agent in category A, to two or more in category B, or to three or more in category C is required to deﬁ ne the microorganism as highly resistant. ESBL=extended-spectrum β-lactamase. *During the study, determination of ESBL was not standardised and multiple diﬀ erent methods were used in participating laboratories; therefore, resistance to any third-generation cephalosporins (eg, cefotaxime, ceftazidime, and ceftriaxone) was used as proxy for presence of ESBL in Escherichia coli,Klebsiella spp, and Proteus spp. Table 1: Deﬁ nitions of highly resistant microorganisms
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles Procedures
non-fermenters, and Gram-positive bacteria. Resistance
The SDD regimen was 4 days of intravenous cefotaxime to any third-generation cephalosporin (eg, cefotaxime, and topical application of tobramycin, colistin, and ceftazidime, and ceftriaxone) was used as proxy for the amphotericin B in the oropharynx and stomach. Use of presence of extended spectrum β-lactamase in antibiotics that impair colonisation resistance, such as Escherichia coli, Klebsiella spp, and Proteus spp. amoxicillin, penicillin, amoxicillin-clavulanic acid, and carbapenems was discouraged during SDD. Surveillance Statistical analysis cultures of endotracheal aspirates and oropharyngeal and
We calculated crude odds ratios (ORs) and 95% CI for
rectal swabs were done on admission and then twice rates of acquisition of bacteraemia and respiratory tract every week. The SOD regimen was oropharyngeal colonisation in intensive care units between the three application of the same paste as used for SDD, with groups, as well as rate reductions, absolute risk surveillance cultures of endotracheal aspirates and reductions, and numbers needed to treat. We did a time-oropharyngeal swabs done on admission and twice every to-ﬁ rst-event analysis Kaplan-Meier survival analysis week, without restrictions in doctors’ choices of systemic with follow-up until 40 days (or discharge from the antibiotic therapy. During standard care, participating intensive-care unit if this was before 40 days). We tested intensive-care units were free to follow their own diﬀ erences between groups with the log-rank test guidelines and surveillance cultures were not done in all (signiﬁ cance deﬁ ned as p<0·05). units. There were no restrictions on the choice of systemic
Patients who had the same species of highly resistant
antibiotic therapy. By contrast with the protocols for microorganism isolated during the ﬁ rst 3 days of stay endotracheal aspirate surveillance cultures, there were no
and after the third day were not included as an event in
diﬀ erences between the three groups in terms of the analysis. The same rule was applied for patients indication for taking blood cultures.
with cefotaxime-resistant and tobramycin-resistant
In this analysis, we used all microbiological culture Gram-negative rods. All analyses were done with SPSS
results from blood and endotracheal aspirate samples version 15.0. obtained from patients in the study. We use the term
This trial is registered at http://isrctn.org, number
colonisation to mean bacterial growth in respiratory tract ISRCTN35176830. samples. For patients with more than one endotracheal aspirate culture in 1 day, we interpreted the results as one
Role of the funding source
culture. We regarded bacterial growth in endotracheal There was no funding source for this study. All authors aspirate samples obtained on the day of admission or the
had full access to all the data in the study and had ﬁ nal
2 days after as colonisation on admission. We regarded responsibility for the decision to submit for publication. bacterial growth in samples obtained after the third day, and with documented absence in previous cultures, as Results acquired in the intensive-care unit. Microbiological 5939 patients were enrolled and data for 5927 patients procedures were identical for respiratory tract samples were available for analysis (table 2). Baseline characteristics that were obtained for clinical reasons (ie, as in standard diﬀ
ered between treatment groups (webappendix).10 See Online for webappendix
care, but also potentially SOD and SDD) as for surveillance
Patients in the SDD and SOD groups had higher acute
In all three study periods, we obtained blood cultures on
Standard care SOD (n=1904) SDD (n=2034)
clinical indication, as assessed by the treating doctors. For
patients with more than one blood culture in 1 day, results
were pooled as results for one blood culture. We regarded
bacterial growth in blood cultures obtained on the day of
admission or the 2 days after as bacteraemia on admission.
We deﬁ ned growth from blood cultures collected after the third day or later, with documented absence in cultures
during the ﬁ rst 3 days, as bacteraemia acquired in the
intensive-care unit. We excluded coagulase-negative
Patients with ≥1 EAC in ﬁ rst 3 days in ICU
Susceptibility testing was done according to the national
guidelines.12 Only the ﬁ rst culture results with an acquired
EAC-days per patient-day in ﬁ rst 3 days in ICU
microorganism were analysed; further positive culture
results with the same microorganism were regarded as
EAC-days per patient-day after day 3 in ICU
Data are n (%) or n (patient-days), unless otherwise stated. SOD=selective oropharyngeal decontamination.
We deﬁ ned highly resistant microorganisms according
SDD=selective digestive tract decontamination. EAC=endotracheal aspirate culture. ICU=intensive-care unit.
to the national guidelines (table 1),13 and established
Table 2: Population of patients and microbiological sampling
three main groups: Enterobacteriaceae, Gram-negative
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles Standard care Crude odds ratio (95% CI) (n=1837) (n=1758) (n=1868)
Data are n (%), unless otherwise stated. SOD=selective oropharyngeal decontamination. SDD=selective digestive tract decontamination. ARR=absolute risk reduction. NNT=number needed to treat. HRMO=highly resistant microorganism. *One case of Saccharomyces cerevisiae in the standard-care group. †One patient in the control group had two episodes of bacteraemia with HRMOs (one episode on day 9 with Enterobacter cloacae and Escherichia coli and one on day 30 with Acinetobacter baumanii). Table 3: Patients with bacteraemia and candidaemia acquired in intensive-care units
standard care, 0·13 in SOD, and 0·11 in SDD (table 2).
Standard care SOD
Bacteraemia occurred in 305 patients in the ﬁ rst 2 days
(n=1837) (n=1758) (n=1868)
after admission to intensive-care units. Bacteraemia
Bacteraemia acquired in an intensive-care unit was
more common in standard care than it was in SDD or
SOD, and was more common in SOD than SDD (table 3).
Candidaemia acquired in an intensive-care unit was more
common in standard care and SOD than it was in SDD
(table 3). 47 episodes of acquired bacteraemia were caused
by highly resistant micro organisms (tables 3 and 4). One
patient had more than one episode (table 3). Development
of bacteraemia acquired in intensive-care units caused by
highly resistant microorganisms was 59% less frequent
Respiratory tract colonisation
with SDD than with standard care and 63% less frequent
for SDD than with SOD (tables 3 and 4).
We did 19 404 microbiological cultures from endo-
tracheal aspirates. When divided between the time of
Other glucose non-fermenting Gram-negative organisms
analysis, frequencies of microbiological cultures from
endotracheal aspirates were lowest during standard care
(table 2). Frequency of respiratory tract cultures was
about 30% lower during standard care than it was in SOD
We obtained endotracheal aspirate cultures on admission
for 688 (35%) of 1989 patients in the standard-care group, 1044 (55%) of 1904 patients in the SOD group, and
1025 (50%) of 2034 in the SDD group (table 2). Distribution
of species isolated was much the same between groups.
Prevalence of highly resistant microorganisms at the time
SOD=selective oropharyngeal decontamination. SDD=selective digestive tract decontamination.
of admission to intensive-care units was low, ranging from
Table 4: Acquired highly-resistant microorganisms
2·4% during SDD to 3·9% during SOD, and did not diﬀ er between study groups. Endotracheal aspirate cultures were obtained from 2595 (48%) of 5463 patients who
physiology and chronic health evaluation (APACHE) II stayed in an intensive-care unit for more than 3 days: scores than did patients in standard care, and were more cultures were obtained for 881 (49%) of 1837 patients often mechanically ventilated and admitted to intensive-
in standard care, 886 (50%) of 1758 in SOD, and
care units for non-surgical reasons. Patients in the SDD 828 (44%) of 1868 in SDD. group were older than were patients in the standard care
After day 3 in the intensive-care unit, colonisation of all
groups of Gram-negative bacteria that were acquired in
5463 (92%) of 5927 patients remained in an intensive-
intensive-care units was highest in patients receiving
care unit for more than 3 days (table 2). Blood cultures standard care. Acquisition rates of Acinetobacter spp and were obtained from 3421 (58%) of 5927 patients, at an Stenotrophomonas maltophilia (gathered as one group) average frequency per patient-day of 0·11 cultures in and of other glucose non-fermenting Gram-negative rods
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles Patients with length of stay >3 days and EACs Crude odds ratio (95% CI) Acinetobacter sppand Stenotrophomonas maltophilia
Other glucose non-fermenting Gram-negative rods*
Escherichia coli and Klebsiella spp
Data are n (%), unless otherwise stated. EAC=endotracheal aspirate culture. SOD=selective oropharyngeal decontamination. SDD=selective digestive tract decontamination. ARR=absolute risk reduction. NNT=number needed to treat. HRMO=highly resistant microorganism. *Such as Pseudomonas aeruginosa. †Haemophilus spp, Neisseria spp, and Pasteurella spp. ‡Vancomycin-resistant enterococci were not cultured. §Two patients in the SOD group and two patients in the SDD group acquired meticillin-resistant Staphylococcus aureus.Table 5: Patients with respiratory tract colonisation acquired in intensive-care units
(predominantly Pseudomonas aeruginosa) were much the and 30 (3%) of 1025 during SDD. We analysed the presence same in patients receiving both SDD and SOD. However,
of cefotaxime resistance only for Enterobacteriaceae;
the rate of acquired colonisation with Enterobacteriaceae
prevalence at admission was nine (1%) during standard
was lower with SDD than with SOD (table 5).
care, 20 (2%) during SOD, and 30 (3%) during SDD.
For Gram-positive bacteria, acquired colonisation with
Colonisation with Gram-negative bacteria acquired in
enterococci occurred more frequently in patients the intensive-care unit and resistant to tobramycin receiving SDD than standard care and SOD (table 5). occurred with much the same frequency and time to Acquired colonisation with yeasts and fungi was more acquisition during all three interventions (ﬁ gure). common with SDD than with standard care (table 5).
Although acquisition with tobramycin-resistantEntero-
After day 3, we noted acquired colonisation with highly bacteriaceaewas lowest during SDD, this reduced
resistant microorganisms in 290 patients. Such colon-
isation was most common during standard care with acquisition rate of tobramycin-resistant non-fermenting 128 (15%) patients, compared with 88 (10%) during SOD Gram-negative rods (table 6). Acquired colonisation and 74 (9%) during SDD (tables 4 and 5). Compared with with cefotaxime-resistant Enterobacteriaceae occurred standard care, SDD reduced the rate of acquired highly less frequently and later during SDD than it did in the resistant microorganisms by 38% and SOD by 32% other treatment groups (ﬁ gure). SDD was associated (tables 4 and 5). Gram-negative bacteria accounted for 98%
with a 62% reduction in acquisition rate of cefotaxime-
of all such microorganisms (tables 4 and 5). Four patients resistant Enterobacteriaceae compared with SOD and acquired respiratory tract colonisation with meticillin-
resistant Staphylococcus aureus in two centres; two patients
Incidence of intensive-care-unit acquired respiratory
were receiving SOD and two were receiving SDD. No tract colonisation with Gram-negative bacteria that were patients acquired vancomycin-resistant entero
cocci intrinsically resistant to colistin (ie, Proteus spp and
Serratia spp) was lower during SDD than it was in
Tobramycin and colistin were used as topical antibiotics
standard care and SOD, with rate reductions of 55% for
in all patients during SOD (in oropharyngeal paste) and SDD compared with standard care and 48% for SDD SDD (in oropharyngeal paste and enteral suspension). compared with SOD (table 6). Diﬀ erences between the Cefotaxime was given intravenously during the ﬁ rst 4 days
three interventions were not caused by clustering of
to all patients receiving SDD. Gram-negative bacteria patients in a minority of the participating intensive-care resistant to either of these antibiotics were rarely found units (webappendix). during the ﬁ rst 3 days of admission to an intensive-care
We also investigated the eﬀ ects of less frequent
unit. During standard care, 13 (2%) of 688 patients in an respiratory tract culture sampling during standard care intensive-care unit with one or more culture in the ﬁ rst by comparing ORs for acquired colonisation with Entero-3 days after admission were colonised with a Gram-
bacteriaceae, non-fermenter species, and Candida
negative bacterium that was resistant to tobramycin at species for the units with comparable culture frequencies admission, compared with 42 (4%) of 1044 during SOD in all three study periods and the units with lower
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles Discussion
In intensive-care units with low rates of antibiotic
resistance, the use of SDD was associated with a
reduction in acquired bacteraemia and respiratory tract
colonisation caused by highly resistant microorganisms compared with standard care. SOD reduced rates of
acquired respiratory tract colonisation with highly
resistant microorganisms compared with standard care,
SDD vs standard care p=0·1421SOD vs standard care p=0·4974
but rates of bacteraemia from highly resistant
microorganisms did not diﬀ er. Daily use of topical tobramycin and colistin was not associated with higher
acquisition rates of resistant Gram-negative bacteria. Intravenous administra tion of cefotaxime for 4 days as
part of SDD, which is generally regarded as a risk factor
Number at risk
for increasing resistance, was associated with a large
reduction in the acquisition rate of cefotaxime-resistant
Enterobacteriaceae in the respiratory tract compared with standard care and SOD.
Our study was the largest prospective assessment of
topical antimicrobial prophylaxis in patients in intensive care, with an estimated overall inclusion rate of 89% in
13 Dutch centres from a cluster-randomised design
(panel). Nevertheless, there were slight diﬀ erences in the baseline characteristics between the study groups,
suggesting that patients receiving SDD or SOD were
more severely ill than were those receiving standard care.10
As a result, reductions in day-28 mortality were apparent only after adjustment for these baseline diﬀ erences. In
the present analysis of microbiological outcomes, no
adjustments were made for these baseline characteristics
and the reported diﬀ erences in resistance between
Number at risk
standard care and both intervention periods should,
therefore, be considered as conservative estimates.
Surveillance of respiratory tract colonisation was done
in all study groups, whereas surveillance of digestive tract
Figure: Kaplan-Meier analysis of time to event of acquisition of tobramycin-resistant Gram-negative rods (A)
colonisation was done only in the SDD group. Therefore
and cefotaxime-resistant Enterobacteriaceae (B)
comparison of colonisation of the digestive tract with
SDD=selective digestive tract decontamination. SOD=selective oropharyngeal decontamination.
resistant microorganisms between the three groups was not possible. However, respiratory colonisation usually
frequen cies during standard care. The average respiratory
reﬂ ects colonisation of the patient.
tract culture frequencies were 0·30 per day during
Another imbalance between the standard-care
standard care, 0·31 during SOD, and 0·32 during SDD population and both intervention groups was the in centres with equal culture frequencies, and 0·18, culture frequency of respiratory tract samples. 0·30, and 0·30, respectively, in those with less frequent Respiratory tract samples were obtained twice every culture sampling during standard care. Compared with week during SDD and SOD, as per protocol, but this standard care, ORs for acquisition of Enterobacteriaceae
was not included in the protocol for standard care. As a
were 0·19 (95% CI 0·12–0·30) during SDD for the three
result, culture frequency was 30% lower during
centres with equal culture frequencies in all study standard care, which might have introduced a negative periods and 0·17 (0·12–0·24) for ten centres with lower detection bias for intensive-care-unit acquired culture frequencies during standard care, and 0·52 colonisation. Potentially, endo
(0·36–0·75) and 0·64 (0·50–0·82), respectively, during done most frequently in the more severely ill patients SOD. For SDD compared with SOD, odds ratios were in standard care, which could have introduced a positive 0·36 (0·22–0·60) in those centres with equal culture selection bias. However, when we compared intensive-frequencies and 0·26 (0·18–0·39) in those with lower care units that had the same routine and frequencies of culture frequencies during standard care. ORs were also
respiratory tract sampling for all three groups with
much the same for the other groups of pathogens (data those units in which sampling occurred less frequently not shown).
for standard care, we noted equivalent ORs for
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles Patients with length of stay >3 days and EACs Crude odds ratio (95% CI) Tobramycin resistance Escherichia coli and Klebsiella spp
Acinetobacter sppand Stenotrophomonas maltophilia
Other glucose non-fermenting Gram-negative rods*
Cefotaxime resistance Colistin resistance Proteus spp and Serratia spp
Data are n (%) unless otherwise stated. EAC=endotracheal aspirate culture. SOD=selective oropharyngeal decontamination. SDD=selective digestive tract decontamination. ARR=absolute risk reduction. NNT=number needed to treat. *Such as Pseudomonas aeruginosa.Table 6: Respiratory tract colonisation with Gram-negative bacteria resistant to tobramycin, cefotaxime, or colistin acquired in intensive-care-units
acquisition of pathogens. Furthermore, microbiological
However, we calculated these numbers needed to treat on
procedures were identical for samples obtained for the basis of patients with a minimum length of stay in an clinical reasons or as part of surveillance, suggesting intensive-care unit of more than 3 days. In an intention-the eﬀ ect of detection bias was small.
to-treat analysis, including the 8% of the patients with
Baseline characteristics and culture frequencies did not
less than 3 days’ stay (equally distributed over the three
diﬀ er between patients receiving SDD or SOD. Our data,
study groups), the number needed to treat would be
therefore, allow investigation of the eﬀ ects of enteral higher. The attributable eﬀ ects of such episodes on decontamination in combination with intravenous survival and length of stay will establish whether these administration of cefotaxime during the ﬁ rst 4 days on ﬁ gures justify a preference for SDD over SOD. In our acquisition of respiratory tract colonisation. SDD was study, with 1904 patients receiving SOD and 2034 receiving associated with a substantial reduction in acquired SDD, diﬀ erences in bacteraemia rates did not produce respiratory tract colonisation with Enterobacteriaceae, signiﬁ cant diﬀ erences in outcomes.10but not with that of non-fermenting Gram-negative rods,
By comparison with SOD, the number needed to treat
such as Acinetobacter species, S maltophilia, and with SDD to prevent one episode of acquired respira -P aeruginosa. As most non-fermenting bacteria are tory col
intrinsically resistant to cefotaxime, intestinal colonisation
Enterobacteriaceae and 17 for intrinsically colistin-
seems to be a relevant source for Enterobacteriaceae, but
resistant Enterobacteriaceae. However, this beneﬁ cial
eﬀ ect on the acquisition with cefotaxime-resistant
As previously reported,10 clinical outcomes (eg, day-28 Enterobacteriaceae is, at least partly, balanced by higher
mortality) were comparable between patients receiving acquisition rates of non-fermenting bacteria that are SDD and SOD, and therefore a preference for either intrinsically resistant to this antibiotic. Overall, rates of regimen was diﬃ
cult to make. Another study14 showed acquired respiratory tract colonisation with highly
that respiratory tract decolonisation was associated with a
resistant microorganisms did not diﬀ er signiﬁ cantly
33% reduction in occurrence of bacteraemia with Gram-
negative bacteria acquired in intensive-care units;
Selection for antibiotic-resistance is an important
intestinal tract decolonisation was associated with a 45% threat associated with SDD and SOD.15 The results of reduction in this study. This eﬀ ect can explain why we our study suggest the opposite. SDD was associated reported large and signiﬁ cant relative reductions in with a 48% reduction in the rate of bacteraemia with bacteraemia in both interventions. The clinical highly resistant microorganisms, and both interventions signiﬁ cance of these ﬁ ndings can best be expressed in yielded reductions in acquired respiratory tract numbers needed to treat. Compared with SOD, 43 patients
colonisation with highly resistant microorganisms of
would have to be treated with SDD to prevent one episode
32% for SOD and 38% for SDD. Moreover, compared
of bacteraemia, 127 patients to prevent one episode of with standard care, SOD was not associated with higher candidaemia, and 145 patients to prevent one episode rates of acquired colonisation with tobramycin-resistant with bacteraemia with highly resistant microorganisms. or colistin-resistant Gram-negative bacteria and SDD
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles
The ecological eﬀ ects of SOD and SDD need to be
Panel: Research in context
established in more detailed and longer studies.
Results from the Extended Prevalence of Infection in
Intensive Care (EPIC) II study conﬁ rmed a strong
We searched the PubMed database without language
relation between infection rate and mortality, and 62% of
restrictions for papers published up to Feb 15, 2011 with the
participating patients had infections with Gram-negative
search terms “ICU” AND “antibiotic resistance” AND “SOD” OR
bacteria and 17% with Candida species; 71% of patients
“selective oropharyngeal decontamination” AND “SDD” OR
received antibiotics as prophylaxis or treatment.3 New
“selective digestive decontamination”. Of 639 search results,
approaches for control of infections and antimicrobial
no studies compared eﬀ ects of selective digestive
resistance in intensive-care units are needed.
decontamination (SDD) with selective oropharyngeal
The beneﬁ ts of prophylactic antibiotic use need to be
decontamination (SOD), or SOD with standard care with
balanced against the inevitable risks of selection of
regard to emergence of antibiotic resistance in patients in
antibiotic-resistant bacteria. As such, widespread use of
intensive-care units. One study5 compared the eﬀ ects of SDD
topical antimicrobial prophylaxis in patients in intensive-
with standard care on incidence of antibiotic resistance, and
care units has been the subject of debate for years. The
had an equivalent study design to our analysis (unit-wide
beneﬁ cial eﬀ ects of SDD and SOD on outcomes, together
with the favourable results for infection and colonisation
with antibiotic-resistant pathogens reported in this study,
We report a novel study quantifying the eﬀ ects of SDD, SOD,
justify the extended use of these interventions in settings
and standard care on occurrence of bacteraemia and
with low rates of antibiotic resistance. However, the
respiratory tract colonisation with highly resistant
long-term eﬀ ects on development of resistance, especially
microorganisms acquired in intensive-care units. SDD was
with presently reported increase in infections caused by
associated with a 59% rate reduction of highly resistant
extended-spectrum β-lactamase-producing bacteria and
bacteraemia compared with standard care in this setting, and
carba penamase-producing bacteria, should be carefully
a 63% rate reduction compared with SOD. SOD was
monitored. Our results warrant further studies in
associated with a rate reduction in acquired respiratory tract
settings with higher baseline rates of resistance than in
colonisation with highly resistant microorganisms of 32%
compared with standard care, and SDD was associated with a
38% reduction compared with standard care.
AMGAdS, JAJWK, and MJMB designed and managed the study. MCJB, EMM, ATB, EJK, MALvH, BMdJ, SFTT, and JAK critically reviewed the study design for intellectual content. AMGAdS, JAJWK, EMM, RFJB, ATB, EJK, MALvH, ARJ, BMdJ, GJvA, IHMEF, SFTT,
was even associated with reduced acquisition rates of SC, JAK, JPA, and PDJS collected data. AMGAdS, JAJWK, MCJB, bacteria resistant to cefotaxime and colistin. These HEMB, and MJMB did the data analysis. AMGAdS, JAJWK, and ﬁ ndings conﬁ rm the results from a single-centre study5
MJMB wrote the ﬁ rst draft of the report. All authors reviewed and revised the report.
in the Netherlands, in which SDD was also associated with smaller proportions of patients colonised with Conﬂ icts of interest
JAWJK has been a consultant to 3M, Destiny Pharma, Cepheid, and
resistant pathogens in the respiratory tract.
Phico Therapeutics and has received grants from Pﬁ zer and Cepheid.
Our patient-based analysis supports the results of our MJMB has participated in advisory boards of 3M, Aventis, and Novartis
previous ecological analysis,11 in which we reported unit-
and has received lecture fees from 3M, Novartis, Pﬁ zer, and
wide reductions in prevalence of antibiotic-resistant Kimberly Clark. All other authors declare no conﬂ icts of interest. Gram-negative bacteria in the respiratory tract during References treatment with SDD of SOD. Nevertheless, in that 1
Vincent J-L. Nosocomial infections in adult intensive-care units. Lancet 2003; 361: 2068–77.
analysis we also reported a unit-wide increase in 2 Roberts RR, Hota B, Ahmad I. Hospital and societal costs of ceftazidime resistance after SOD and SDD were
antimicrobial-resistant infections in a Chicago teaching hospital:
discontinued, which seems to contradict our present
implications of antibiotic stewardship. Clin Infect Dis 2009; 49: 1175–84.
ﬁ ndings. However, there are two important diﬀ erences 3 Vincent J-L, Rello J, Marshall J, et al. International study of the
between the analyses. In the ecological analysis,11 we used
prevalence and outcomes of infection in intensive care units.
point-prevalence surveys in which all patients in the unit
JAMA 2009; 302: 2323–29.
D’Amico R, Piﬀ eri S, Leonetti C, Torri V, Tinazzi A, Liberati A.
(whether or not enrolled in the trial) were included and
Eﬀ ectiveness of antibiotic prophylaxis in critically ill adult
resistance trends were analysed over time. Many of these
patients: systemic review of randomised controlled trials. BMJ
patients were in the intensive-care unit for a very short
1998; 316: 1275–85.
amount of time (≤3 days), and did, therefore, not receive
de Jonge E, Schultz M, Spanjaard L, et al. Eﬀ ects of selective decontamination of the digestive tract on mortality and
SDD or SOD and were not included in the present
acquisition of resistant bacteria in intensive care: a randomised
patient-based analysis. Furthermore, as we did not have
controlled trial. Lancet 2003; 362: 1011–16.
control units without SDD and SOD, we could not 6 Pugin J, Auckenthaler R, Lew DP, Suter PM. Oropharyngeal
decontamination decreases incidence of ventilator- associated
establish to what extent the reported increase of resistance
pneumonia: a randomized, placebo-controlled, double-blind
in time developed independent of either intervention.
clinical trial. JAMA 1991; 265: 2704–10.
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4 Articles
Bergmans DC, Bonten MJ, Gaillard CA, et al. Prevention of
12 CLSI. Performance standards for antimicrobial susceptibility
ventilator-associated pneumonia by oral decontamination: a
testing. Wayne, PA, USA: Clinical and Laboratory Standards
prospective, randomized, double-blind, placebo-controlled study.
Am J Respir Crit Care Med 2001; 164: 382–88.
13 Kluytmans-VandenBergh MFQ, Kluytmans JAJW, Voss A.
van der Waaij D, Berghuis-de Vries JM, Lekkerkerk-van der Wees JEC.
Dutch guideline for preventing nosocomial transmission
Colonization resistance of the digestive tract in conventional and
of highly-resistant micro-organisms. Infection 2005; 33: 309–13.
antibiotic-treated mice. J Hygiene 1971; 69: 405–11.
14 Oostdijk EA, de Smet AM, Kesecioglu J, et al. The role of intestinal
Stoutenbeek CP, van Saene HKF, Miranda DR, Zandstra DF.
colonization with Gram-negative bacteria as a source for intensive
The eﬀ ect of selective decontamination of the digestive tract
care unit-acquired bacteremia. Crit CareMed 2011; published online
on colonization and infection rate in multiple trauma patients.
Jan 28. DOI:10.1097/CCM.0b013e318208ee26. Intensive Care Med 1984; 10: 185–92.
15 Bonten MJ, Brun-Buisson C, Weinstein RA. Selective
10 de Smet AM, Kluytmans JA, Cooper BS, et al. Decontamination
decontamination of the digestive tract: to stimulate or to stiﬂ e?
of the digestive tract and oropharynx in ICU patients. N Engl J MedIntensive Care Med 2003; 29: 672–76.
2009; 360: 20–31.
11 Oostdijk EA, de Smet AM, Blok HE, et al. Ecological eﬀ ects
of selective decontamination on resistant Gram-negative bacterial colonization. Am J Respir Crit Care Med 2010; 181: 452–57.
www.thelancet.com/infectionPublished online March 21, 2011 DOI:10.1016/S1473-3099(11)70035-4
Scientific Reference Guide Depression in Adults What are the best treatments? Description of Condition Depression is a common medical condition with a lifetime prevalence in the United States of 15% among adults. Symptoms include feelings of sadness, hopelessness, or guilt; diminished interest or pleasure in usual activities; decreased ability to think or concentrate; and disruptions in sl
EXAMPLES OF ELIGIBLE EXPENSES Abortions, Insurance companies for eligible Obstetrical OVER-THE COUNTER DRUGS Over-the counter medicines may be reimbursable under your Health Reimbursement Account, depending upon whether the item treats a specific medical condition or is used for a person’s general good health. If the item is obviously used to treat a medical condition (such