Journal of Hospital Infection (1996) 34, 279-289

Secondary carriage with multi-resistant Acinetobacter baumannii and Klebsiella pneumoniae in an adult ICU population:

relationship with nosocomial infections and mortality

M. Garrouste-Orgeas *, 0. Marie-f, M. Rouveaus, S. Villierst, G. Arlet$ and B. Schlemmer*

*Service de Re’animation Medicale, -J-Dipartement d’Anesthdsie-Reanimation, and $Service de Microbiologic, H6pital Saint Louis et Universite’ Paris VII,

1 avenue Claude Vellefaux, 75010 Paris, France

Received 1 May 1996; revised manuscript accepted 10 July 1996

Summary: A one year prospective, observational survey was performed to evaluate the abnormal carriage of multi-resistant Klebsiella pneumoniae and/ or Acinetobacter baumannii, to determine associated risk factors for carriage, and to correlate the abnormal carriage with infectious morbidity and mortality in the intensive care unit (ICU) of a University Hospital. Two hundred and ninety-eight patients who stayed in the ICU >48h, and were not neutropenic, were studied. Salivary and rectal samples were obtained on admission and weekly until discharge. Out of 265 evaluable patients, 88 (33%) developed oropharyngeal and/or rectal carriage within a median of nine days. Three factors were significantly associated with abnormal carriage: higher ‘severity of illness’ score on admission, a threefold increase in ICU stay, and the need for mechanical ventilation. K. pneumoniae or A. baumannii accounted for 57/158 (36%) of all ICU-acquired infections (in 46 patients). They were considered as secondary endogenous infections (SEI) in 42 patients who were previously colonized with the same strains, and developed infection within a median of three days (range O-68 days). Prolonged stay in ICU was the only factor associated with SE1 in the carrier population. Mortality was significantly greater in the carrier group (43 vs 25%, P=O.O006). Post hoc stratification suggested that abnormal carriage only influenced mortality in patients showing a low severity of illness score on admission to ICU. Abnormal carriage was found in the most severely ill patients, predisposed to secondary nosocomial infections, and could influence mortality in the less severely ill.

Keywords: Nosocomial infections; secondary nosocomial endogenous in- fections; colonization; digestive bacterial carriage; Klebsiella pneumoniae; Acinetobacter baumannii; intensive care.

Correspondence to: Benoit Schlemmer, Service de Reanimation Medicale, Hopital Saint Louis, 1 avenue Claude Vellefaux, 75010 Paris, France. This work has been presented in part as oral communication at the XXXIst Interscience Conference on Antimicrobial Agents and Chemotherapy, 1991, Chicago.

0195-6701/96/120279+11 512.00/O

279

0 1996 The Hospital Infection Society

280 M. Garrouste-Orgeas et al.

Introduction

About 25 years ago, Johanson et al.’ showed that oropharyngeal carriage of aerobic Gram-negative bacilli (AGNB) could reflect severity of disease. Abnormal carriage and infection with multi-resistant AGNB has been described in numerous studies in the last few years, specially with Acine- tobacter baumanniizA and Enterobacteriaceae.‘” There is a general consensus that carriage is the first stage in development of endogenous infection and a qualitative relationship for this has been established.5-6 However, the risk factors and clinical impact of the abnormal carrier state on mortality have not been previously investigated in an intensive care unit (ICU) population.

During the past decade, in our unit, as in other ICUs in Paris, France, we have detected a dramatic increase in hospital-acquired infection caused by A. baumannii (Acinetobacter calcoaceticus var. anitratus) and Klebsiella pneumoniae with extended spectrum p-lactamase. This disturbing situation prompted us to investigate risk factors for the acquisition of such multi- resistant strains in order to develop preventive measures. We describe the results of an epidemiological study to identify risk factors for oropharyngeal and rectal carriage of K. pneumoniae and A. baumannii in a mixed ICU population. Correlation between the abnormal carrier state with the two marker strains and infectious morbidity and mortality was the secondary goal of the study.

Methods

Patients The study was performed in the general ICU of Saint-Louis Hospital which consists of two areas: an acute cubicle of 12 beds (six isolation rooms, two triple bedded rooms), and an intermediate area of eight beds (two isolation rooms, three twin bedded rooms). Patients were selected on admission if their ICU stay was likely to exceed 48 h and if they were not neutropenic (absolute neutrophil count ~500 cells/mm3). The Hospital Institutional Review Board waived the need for informed consent, because of the epidemiologic nature of the study and absence of invasive procedures.

From February 1990 to February 1991, 484 patients were hospitalized in the unit. Having excluded 145 patients who stayed less than two days, and 41 with neutropenia we studied 298 medical and surgical patients prospectively. We recorded age, sex, severity of condition on admission (using SAPS, the simplified acute physiological score7), severity of the underlying disease using MacCabe’s score,* length of ICU stay, need of mechanical ventilation and ICU outcome. All patients were examined daily for the occurrence of hospital-acquired infections. Indwelling urinary catheters, nasogastric tubes, central venous lines and arterial catheters were used when necessary. Indwelling urinary catheters were attached to a closed drainage bag. Central venous lines were dressed every 48 h. Intravenous

Multi-resistant bacteria in ICU 281

catheters remained in place as long as necessary or until signs of infection were detected. The same ventilator circuits, with heat and moisture ex- changers changed daily, were used throughout the stay of each patient.

Antibiotic policy There was no strict antibiotic policy. The antibiotic choice depended on the clinical conditions, microbiological results and the physician’s discretion. The most commonly used antibiotics were co-amoxiclav, piperacillin, ce- fotaxime, imipenem, gentamicin, amikacin, and ciprofloxacin. Selective gut decontamination was not used. The oropharynx was disinfected six times daily, with hexitidine solution.

Sampling Surveillance samples were obtained from the throat and rectum on ad- mission and weekly thereafter until discharge from the ICU. Diagnostic samples (bronchial protected specimen brush, blood) were taken on clinical indication only, except for urine which was cultured routinely once a week.

Bacteriology Saliva samples (100 yL) and rectal swabs were inoculated on a selective medium similar to MacConkey’s medium (Drigalski agar, Sanofi Dia- gnostics Pasteur SA, Marnes la Coquette, France) for Gram-negative bacilli to detect K. pneumoniae producing extended spectrum (3-lactamase and A. baumannii resistant to third generation cephalosporins and aztreonam. Two disks (ceftazidime and co-amoxiclav) were put 30 mm apart into the first third of the streaked area. Protected bronchial brush specimens were quantitatively cultured on Columbia sheep blood agar plates and chocolate with Polyvitex agar plates (BioMerieux SA, Marcy l’Etoile, France). The plates were incubated overnight at 37”C, and any colonies emerging near the disks were identified with the API 20 E or 20 NE systems (BioMerieux SA).

Susceptibility testing was performed on Mueller-Hinton agar by the disk diffusion technique (Sanofi Diagnostics Pasteur SA) and extended spectrum p-lactamases were detected using the double-disk synergy test.’ Strains were identified and compared by a combination of biotyping and anti- biotyping using 32 agents (amoxicillin, amoxicillin-clavulanate, ticarcillin, ticarcillin-clavulanate, mezlocillin, piperacillin, cephalothin, cefamandole, cerfuroxime, cefoxitin, cefoperazone, cefotaxime, ceftazidime, moxalactam, aztreonam, imipenem, kanamycin, gentamicin, tobramycin, netilmicin, ami- kacin, nalidixic acid, norfloxacin, pefloxacin, ciprofloxacin, tetracycline, chloramphenicol, trimethoprim, cotrimoxazole, rifampicin, colistin and fosfomycin).

Dejkitions Abnormal carriage was defined as having at least one surveillance sample (oropharynx and/or rectum) yielding the multi-resistant K. pneumoniue

282 M. Garrouste-Orgeas et al.

and/or A. baumannii strains. Infections were considered as ICU-acquired when signs and symptoms of infection developed 48 h or more after admission. Infections caused by strains that had already been identified in the oropharynx and/or gut were considered to be ‘secondary endogenous infections’. Exogenous infections were infections caused by K. pneumoniae and/or A. baumannii without previously detected carriage. Infections were defined as follows: lower respiratory tract infections by clinical and radio- logical criteria together with a positive quantitative culture of the protected specimen brush.” Pneumonia was distinguished from distal bronchitis through microbiological data, using the cut-off of 103cfu/mL on the pro- tected specimen brush, applied regardless of prior antibiotic therapy. Pneu- monia was diagnosed by the presence of at least one organism at a concentration 2 1 03cfu/mL, whilst at least one organism isolated at <l 03cfu/ mL defined distal bronchitis. Catheter-related bacteraemia was defined using simultaneous quantitative blood cultures (Du Pont Isolator 1.5 mL). When the colony count in the central blood culture was at least threefold higher than that in the peripheral blood culture, the septicaemia was considered to be catheter-related.” Bacteraemia was interpreted to be of gut origin when there was no evidence of a primary septic focus or a catheter-related infection. Urinary tract infection was defined by growth 2 lO’cfu/mL from urine. In patients with an indwelling catheter, the same strain had to be isolated from two consecutive samples or be present in a blood culture as well as the urine. Sinusitis was defined as the occurrence of purulent sinus discharge with positive sinus culture and opacification of the sinus on cranial CT scan.

Statistical analysis Outcome was assessed by ICU survival. Descriptive statistics were ex- pressed as mean &SD or medians. Groups were compared for demographic data by one way analysis of variance (ANOVA) by Scheffe test, and for categorical data by Chi-square analysis. RO.05 was considered as significant.

Results

Two hundred and ninety-eight patients (210 medical, 88 surgical) were enrolled in the study. Forty-eight surgical patients underwent unscheduled surgery. The most frequent medical diagnoses were: acute respiratory failure (46*2%), neurological conditions (25.7%), shock (16*2%), acute renal failure (5.2%), metabolic disorders (5.2%) and miscellaneous (1.5%). Eighty- three patients (28%) were immunocompromised; due to AIDS (9%), malig- nant haematologic disease (ll%), cancer (7%) or conditions such as pro- longed steroid or immunosuppressive therapy (8%). Thirty-three patients were excluded from analysis: 23 had no samples taken, and 10 were already hospitalized and carried the multi-resistant strains on admission to the

Multi-resistant bacteria in ICU 283

Table I. Demographic and clinical characteristics of the 298 baseline patients and the 265 evaluable patients

Characteristics Baseline (N= 298)

Evaluable (N= 265)

Age (year) SAPS MacCabe’s score (%)

1 2 3

Immunodepression (%) Days on ICU Mechanical ventilation (%) Mortality (%)

55+17 57+17 12f5 12f5

(64)

1:;; ‘,;:i (13)

(28) (28) 17-Jl8 lSk19 (67) (30)

Plus-minus are means f SD. Values in parentheses are percentages. SAPS denotes simplified acute physiological sc~re.~ MacCabe’s score8 1: no underlying disease, MacCabe’s score 2: ultimately fatal disease (five years), MacCabe’s score 3: rapidly fatal disease (one year).

Table II. Factors associated with abnormal carriage

Evaluable Carriers Non-carriers

Number (%) Age (years) Severity (SAPS) Days on ICU Immunosuppresion (%) MacCabe’s score (%)

1 3

Mechanical ventilation (%) Mortality (%)

265 57+17 12f5 18+19 (28)

(64) (23)

I:;{ (31)

88 (33) 60f15 13f4.6 31+22 (30)

(61) (25)

Ii:; (43)

177 (67) 5.5&19 11.3*5* 11+12-f (27)

Plus-minus value are means k SD. Values in parentheses are percentages. *P=O.O002, tP=O.OOOl, $P=O.O006 (between carriers and non-carriers).

ICU. Seven of the latter had abnormal carriage and three were infected on admission. No significant difference was observed between the baseline patients and the remaining 265 evaluable for either demographic char- acteristics or outcome (Table I).

Carriage Of 1647 surveillance samples, 247 from 88 patients (33%) were positive for one or both marker strains. The median time from admission to the first positive surveillance sample was nine days (range 2-58). Thirty-two patients (36.3%) carried K. pneumoniae alone, 34 patients (38.6%) carried A. bau- mannii alone, and 22(25%) carried both strains. Abnormal carriage was detected first in the oropharynx of 34 (38.6%) patients.

The characteristics of the 88 carriers and the 177 non-carriers are listed in Table II. A higher severity of illness on admission, a threefold increase

284 M. Garrouste-Orgeas et al.

Table III. Factors associated with secondary endogenous infections in carriers patients

Carriers Infected Non-infected

Number 88 (33) Age, years

42 (47.7) 46 (52.3) 60+15 59*13 60+15

Severity SAPS 135-4 1424 13+4 Days on ICU 31 k22 36+24 27f22* Immunosuppression (%) (30) MacCabe’s score (%)

(31) (28)

1 2 [E{

(60) (63)

3 I::; Mechanical ventilation (%) I;:{

I::{

Mortality (%) (43) gp’

Values in parentheses are percentages. Plus-minus are means *SD. *P=O.OS.

in duration of ICU stay, and the need for mechanical ventilation were significantly associated with carriage.

Infections One hundred and fifty-eight hospital-acquired infections occurred in 80 patients, 114 (72.2%) due to Gram-negative bacilli and 44 (27.8%) to Gram- positive cocci. K. pneumoniae or A. baumannii accounted for 57 (36%) of these and were found in 46 patients. Fifty-two (91%) of the 57 infections (in 42 patients) were considered as secondary endogenous infections. The median interval between admission and infection was 14 days (range 2-83) and between detection of abnormal carriage and infection was three days (range O-68). A. b aumanni caused 34 infections, whilst K. pneumoniae was involved in 23. Secondary endogenous infections with A. baumannii were mainly of the upper and lower respiratory tract including: pneumonia (N= 9 patients), distal bronchitis (N= 9), sinusitis (N= 5). There were only four bacteraemias without any primary focus and two urinary tract infections. However, there was no statistically significant relationship between oro- pharyngeal carriage of A. baumannii and respiratory tract infection. Sec- ondary endogenous infections with K. pneumoniae included: bacteraemia without any primary focus (A/=8), urinary tract infection (N=6)1 distal bronchitis (N=4), sinusitis (N= 3), catheter-related infections (N= 2). There were five ICU-acquired infections among non-carriers: three cases of bacteraemia without evidence of a septic focus and without evidence of catheter-related infection, one distal bronchitis, and one pneumonia due to A. baumannii.

Prolonged stay in the ICU was the only factor associated with secondary endogenous infection in the carrier population (Table III). When analysis was restricted to survivors only, the length of hospitalization was always significantly higher in infected patients (43 + 20 days vs 28 f 19 days, P= 0.009). However, in this group of survivors, the severity of illness on

Multi-resistant bacteria in ICU

Table IV. Mortality versus carriage and severity on admission, (SAPS)

285

SAPS Carriers Non-carriers

1-15 23158 (39.6) 28/146 (19.1)* >15 15/30 (50) 16/31 (51.6)

The first value is the number of patients. Values in parentheses are percentages of mortality. *P<O.Ol.

admission, as assessed by the SAPS, was not significantly higher in infected patients (14.6 f 3.1 vs 12.5 + 4.9, P=O*O8).

Mortality Of the 88 ICU patients who carried one or both marker strains 38 (43%) died, whilst in the non-carrier group the mortality was only 25% (P= 0.0006) (Table II). Wh en severity on admission was stratified into two SAPS groups and then plotted against mortality, abnormal carriage influenced mortality only in the group of ICU patients with a SAPS score ~16, mortality was 39.6% in the carrier group versus 19.2% in non-carriers (P<O.Ol) (Table IV). When the analysis was focused on carriers, secondary endogenous infections influenced mortality. In patients with a SAPS score ~16, mortality was 54.5% in the infected group, versus 26.9% in non- infected patients (P<O*OOl). On the contrary, no difference in mortality was observed when carrier patients were severely ill on admission. There was no impact of underlying disease on mortality among infected versus non-infected carrier patients when plotting mortality against three groups of MacCabe’s scores (patients separated without a fatal disease, patients with a fatal disease in the next five years and patients with a fatal disease in the next 12 months).

Discussion

Because the digestive tract in critically ill patients is now known to play an important role in the pathophysiology of hospital-acquired infections, the major aim of this study was to explore the risk factors and the consequences of an abnormal carrier state in an adult ICU population. Our study differs from previous reports”‘h’6 in the following ways: first, we followed only two multi-resistant marker strains responsible for several nosocomial infections in our unit. Second, we defined underlying disease’ and severity on admission’ with appropriate ICU scores.

One third of 265 patients acquired the multi-resistant marker strains K. pneumoniae and/or A. baumannii after a median exposure of nine days in the ICU. Three factors were associated with abnormal carriage: the severity of the acute illness, the need for mechanical ventilation, and the length of ICU stay. Several studies have focused on risk factors of abnormal carriage in ICU patients; severity of underlying disease,“” length of hospitalization

286 M. Garrouste-Orgeas et al.

and the use of broad-spectrum prophylactic antibiotics13,14 were identified as main risk factors for the emergence of colonization. Recently, Lortholary et aZ.’ have shown that the severity of illness on admission was an in- dependent risk factor for nosocomial acquisition of multi-resistant A. baumannii.

The secondary endogenous infection rate was 57/158 (36%). The only risk factor associated with infections was the length of ICU stay. To focus on the relationship between length of hospitalization and infection, we restricted the analysis to survivors only, because some infected patients had potential length of hospitalization decreased by early death and so the average time spent in ICU would not reflect the negative impact of infection. In infected patients, the length of stay was increased by 15 days. This result must be interpreted cautiously, because proper analysis, i.e. a matched cohort study, was not performed to address this issue. However, this study is in accordance with recent studies demonstrating that ‘excess of days spent in the ICU’ could be attributable to infections,” particularly to nosocomial infections such as A. baumannii or Pseudomonas aeruginosa pneumonia. l6 There is a general consensus that abnormal carriage is the first stage of secondary endogenous infections. Our study showing a 91% incidence of secondary endogenous infections with the marker strains is in accordance with this opinion. The exact sources and routes for the development of nosocomial pneumonia are still under debate.17 However, the endogenous route of colonization has been assumed to be of the most importance. In our study, although surveillance tracheal samples were not simultaneously obtained and analysed in relation to the results of oropharyngeal or gastric samples, all nosocomial pneumonias were due to A. baumannii and, although statistically nonsignificant, a predominant carriage of A. baumannii was found in oropharyngeal surveillance samples.

Four factors are known to influence ICU outcome: age, previous health status, severity of disease and diagnosis. Diagnosis plays a key role when assessing prognosis and mortality. However, the link between infection and mortality in the ICU remains unclear. Whether the underlying disease or infection could contribute to mortality has been intensively debated. The difficulty of resolving this problem is in part elucidated in recent studies in which ‘excess mortality’ attributable to infection was analysed.

In patients with bloodstream infections, Smith et al.” used the APACHE I I score to predict mortality. In a group of 34 ICU patients with bloodstream infections, the observed mortality (82.4%) significantly exceeded the pre- dicted value (54.1%, P= O-025), whilst in the subgroup of control patients without bloodstream infections the predicted and observed death rates were similar (53.1% vs 52.9%). In a case-control study of 48 ventilated patients where the severity of acute illness and the underlying disease were carefully matched, Fagon et al.” found a 54.2% mortality in patients with nosocomial pneumonia as compared with 27.1% in controls. The ‘excess mortality’ of nosocomial pneumonia was 27.1%. Therefore well-designed studies suggest

Multi-resistant bacteria in ICU 287

an independent effect of nosocomial infection on mortality. In a case- control study of 100 dead patients and 100 surviving patients, Gross et ~1.‘~ focused on the relative roles of admission health status and nosocomial infection. A poor clinical status on admission was more significant in hospital-related mortality than were nosocomial infections. However, noso- comial infection might have affected hospital mortality in patients whose conditions were not terminal on admission.

In our study, mortality did not appear to be clearly influenced by nosocomial infections; however infected patients showed a trend towards an increased mortality as compared with non-infected patients (50% in the infected group vs 37% in the non-infected group). The small size of the sample is more likely to account for not reaching statistical significance.

The relationship between abnormal carriage and mortality deserves further comment: abnormal carriage could influence mortality. The mor- tality among carrier patients was 43% vs 25% in the non-carriers. Post hoc stratification into two groups of SAPS showed that abnormal carriage only influenced mortality in the less severely ill patients with a SAPS ~16. These results are in accordance with two recents studies. Lortholary et aL2 found in a matched cohort study that attributable mortality for A. baumannii colonization was 25.9% and the estimated risk ratio was 2.16. Recently, Bueno-Cavanillas et aZ.,15 using a logistic regression analysis adjusted for several confounding factors, found more impact of nosocomial infection in young patients with APACHE II scores of 11-30 and therapeutic intensity scores ~20, i.e. in patients with an a priori better prognosis and with a lower intrinsic risk of death.

Regardless of the particular epidemiological conditions in the ICU during the study period, prevention of abnormal carriage and subsequent secondary endogenous infections might decrease infectious morbidity and mortality. This is the rationale for selective gut decontamination. However, despite a major impact on the rate of ICU acquired infection, the overall clinical benefit of gut decontamination has been less clear.” A recent meta-analysis21 of randomized controlled trials of selective gut decontamination suggested that the most promising results came from trials in which the combination of topical and systemic treatment was used, with a 20% significant reduction in mortality. Better preventive results have been observed in trauma patients** than in a mixed population.23p24 This discrepancy could be ex- plained by the fact that, as discussed above, nosocomial infections are not a major event in determining mortality in all critically ill patients, but remain important in patients without critical underlying illness. Due to the close link between abnormal carriage and secondary endogenous infections, and its influence on mortality in the subgroup of the less severe patients, the acquisition and cross-transmission of colonizing multi-resistant patho- gens has to be controlled through a strict hygiene policy.

The early identification of patients at high risk for acquisition of a potentially pathogenic multi-resistant organism has to be emphasized,

288 M. Garrouste-Orgeas et al.

together with the need to develop preventive measures which limit cross- transmission and facilitate the eradication of such strains in the ICU setting.

We are indebted to the nursing staff for their generous co-operation, and to Armelle Morin for assistance in the preparation of the manuscript.

References

I.

2.

3.

Johanson WG, Pierce AK, Sanford JP. Changing pharyngeal flora in hospitalized patients. N EngZJ Med 1969; 281: 1137-1140. Lortholary 0, Fagon JY, Buu Hoi A, Slama MA, Pierre J, Giral P, Rosenzweig R, Gutmann L, Safar M, Acar J. Nosocomial acquisition of multi-resistant Acinetobacter baumannii: risk factors and prognosis. Clin Infect Dis 1995; 20: 790-796. Scerpella EG, Wanger AR, Armitige L, Anderlini P, Ericsson CD. Nosocomial outbreak caused by a multi-resistant clone of Acinetobacter baumannii: results of a case control and molecular epidemiologic investigations. Infect Control Hosp Epidemiol 1995; 16: 92-97.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

Constant J, Kemeny E, Oxley C, Perry E, Garber G. Investigation of an outbreak of Acinetobacter calcoaceticus var. anitratus infections in an adult intensive care unit. Am J Infect Control 1990; 18: 288-291. Kerver AJH, Rommes JH, Mevissen-Verhage EAE, Hulstaert PF, Vos A, et al. Colonization and infection in surgical intensive care patients-a prospective study. Intensive Care Med 1987; 13: 347-351. Van Uffelen R, Van Saene HKF, Fiddler V, Lowenberg A. Oropharyngeal flora as a source of bacteria colonizing the lower airways in patients on artificial ventilation. Intensive Care Med 1984; 10: 233-237. Le Gall JR, Loirat P, Alperovitch A. A simplified acute physiology score for ICU patients. Crit Cure Med 1984; 12: 957-977. McCabe WR, Jackson GG. Gram-negative bacteremia: etiology and ecology. Arch Intern Med 1969; 110: 847-855. Legrand P, Fournier G, Bure A, Jarlier V, Nicolas MH. Deere D. et al. Detection of extended broad-spectrum p-lactamase in Enterobacteriaceae in four French hospitals. Eur 7 Clin Microbial Infect Dis 1989: 8: 527-529. Chagtre J, Viau F, Brun P, Pierre J, Dauge MC, Bouchama A, et al. Prospective evaluation of the protected specimen brush for the diagnosis of pulmonary infections in ventilated patients. Am Rev Respir Dis 1984; 130: 924-929. Douard MC, Arlet G, Leverger G, Paulien P, Waintrop C, Clementi E, et al. Quantitative blood cultures for diagnosis and management of catheter-related sepsis in pediatric hematology and oncology patients. Intensive Cure Med 1991; 17: 30-35. Valenti WM, Trudell RG, Bentley DW. Factors predisposing to oropharyngeal col- onization with Gram-negative bacilli in the aged. N Engl J Med 1978; 298: 1108-1111. Olson B, Weinstein RA, Nathan C, Chambertin W, Kabins SA. Epidemiology of endemic Pseudomonas aeruginosa: why infection control efforts have failed. J Infect Dis 1984; 150: 808-816. Flynn DM, Weinstein RA, Nathan C, Gaston MA, Kabins SA. Patients’ endogenous flora as the source of ‘nosocomial’ Enterobacter in cardiac surgery. J Infect Dis 1987; 156: 363-368.

15.

16.

17.

18.

Bueno-Cavanillas A, Delgado-Rodriguez M, Lopez-Luque A, Schaffino-Cano $3, Galvez-Vargas R. Influence of nosocomial infection on mortality rate in an intensive care unit. Crit Care Med 1994; 22: 55-60. Fagon JY, Chastre J, Hance AJ, Montravers P, Novara A, Gilbert C. Nosocomial pneumonia in ventilated patients: a cohort study evaluating attributable mortality and hospital stay. Am J Med 1993; 94: 281-288. De Latorre FJ, Pont T, Ferrer A, Rossello J, Palomar M, Planas M. Pattern of tracheal colonization during mechanical ventilation. Am 7 ResiGr Crit Care Med 1995: 152: 1023-1033. -

- _

Gross PA, Antwerpen CV. Nosocomial infections and hospital deaths. A case-control study. AmJ Med 1983; 75: 658-662.

Multi-resistant bacteria in ICU 289

19. Smith RL, Meixler SM, Simberkoff MS. Excess mortality in critically ill patients with nosocomial bloodstream infections. Chest 1991; 100: 164-l 67.

20. Selective digestive decontamination in intensive care unit patients. First European consensus conference in intensive care and emergency medicine. Intensive Care Med 1992; 18: 182-188.

21. Selective Decontamination of the Digestive Tract Trialists’ Collaborative Group. Meta- analysis of randomised controlled trials of selective decontamination of the digestive tract. BMJ 1993; 307: 525-532.

22. Ledingham IM, Alcock SR, Eastaway AT, McDonald JC, McKay IC, Ramsay G. Triple regimens of selective decontamination of the digestive tract, systemic cefotaxime, and microbiological surveillance for prevention of acquired infection in intensive care. Lancet 1988; 16: 1087-1093.

23. Brun-Buisson C, Legrand P, Rauss A, Montravers C, Besbes M, et al. Intestinal decontamination for control of nosocomial multi-resistant Gram-negative bacilli. Ann Intern Med 1989; 110: 873-881.

24. Gastinne H, Wolff M, Delatour F, Faurisson F, Chevret S. A controlled trial in intensive care units of selective decontamination of the digestive tract with nonabsorbable antibiotics. N Engl J Med 1992; 326: 594-599.