Introduction

Multidrug resistant bacteria (MDRB) are common in critically ill patients, and are frequently reported to be responsible for ICU-acquired infections. Based on the results of the large multinational EPIC II study, 35% of infectious episodes diagnosed in 7,087 patients were related to MDRB [1]. Further, the EUROBACT multinational study showed that MDRB were responsible for 48% of hospital-acquired bacteremia episodes [2]. A more recent large multinational observational study, aiming at assessing the incidence of ventilator-associated lower respiratory tract infections, reported that MDRB were identified as causative pathogens in 61% of patients with ventilator-associated pneumonia (VAP) or ventilator-associated tracheobronchitis [3].

Infections related to MDRB are associated with higher mortality rates, longer duration of mechanical ventilation, and ICU stay [4, 5]. Potential explanations for the high mortality rate in patients with infections related to MDRB include inappropriate initial antibiotic treatment, drug toxicity, and subsequent resistance [6].

The main risk factors for resistance are prior exposure to antibiotics, prolonged hospital and ICU length of stay, invasive devices, comorbidities and local epidemiology [7–10]. Identifying risk factors for ICU-acquired MDRB might be helpful to improve preventive strategies and outcome of critically ill patients. Colonization pressure, defined as the ratio of patients colonized with MDRB relative to all patients, has been reported to be a risk factor for ICU-acquired methicillin-resistant Staphylococus aureus (MRSA) [11], vancomycin resistant enterococcus (VRE) [12], Clostridium difficile [13], Acinetobacter baumannii [14], and Pseudomonas aeruginosa [15]. However, none of these studies evaluated the impact of colonization pressure related to all MDRB on the risk for acquiring these bacteria in the ICU. Therefore, we hypothesized that higher colonization pressure related to MDRB would be associated with increased risk for ICU-acquired MDRB.

The primary aim of this prospective observational study is to determine if colonization pressure for all MDRB is a risk factor for ICU-acquired colonization or infection related to MDRB.

Material and methods

Study design

This prospective observational study was performed during a 13-month period (from January 2007 to January 2008), in three 10-bed medical and surgical ICUs at the University Hospital of Lille, France.

Study Population

All adult patients admitted to the ICU for >48 h were included. Exclusion criteria were length of stay ≤ 48 h, age less than 18 years, and ICU readmission.

Infection control policy included hospitalization in single-bed rooms, adequate hand hygiene, achieved by using an alcohol-based hand rub formulation before and after each patient contact, routine screening for MDRB, written antibiotic treatment protocols, continuous surveillance of nosocomial infections and adequate cleaning of ICU rooms.

Routine screening for MDRB was performed for all patients at ICU admission and weekly thereafter. This screening included nasal and anal swabs. In addition, tracheal aspirates were performed in intubated or tracheotomized patients. Microbiological cultures of other specimens were performed according to clinical status.

In all patients, isolation techniques were used at ICU admission until receipt of screening results. Thereafter, these techniques were performed in all patients with infection or colonization due to MDRB. Preventive isolation techniques were applied in all immunosuppressed patients, during the whole ICU stay. These techniques included protective gowns, gloves, and mask usage.

Nurse to patient ratio was 1:3 in the three participating units.

Data collection and definitions

All data were prospectively collected. MDRB were defined based on our institution’s definition, as MRSA, ceftazidime or imipenem resistant Pseudomonas aeruginosa, ESBL-producing GNB, and all strains of Acinetobacter baumannii, and Stenotrophomonas maltophilia [16].

Daily occupancy rate was defined as the ratio of hospitalized patients to the total number of operational beds per ICU. In patients with MDRB, occupancy rate was defined as the mean of daily occupation rate, from ICU admission until the last MDRB acquisition. In patients with no MDRB, occupancy rate was defined as the mean of daily occupation rate during the whole ICU stay.

Daily colonization index was defined as the ratio of patients with MDRB (infected and/or colonized) to the total number of patients. In patients with MDRB, colonization pressure was determined as the mean of colonization index, from ICU admission until the last MDRB acquisition. In patients with no MDRB, colonization pressure was determined as the mean of colonization index during the whole ICU stay.

In patients with MDRB, percentage of days with antibiotics was defined as the ratio of days with antibiotics, from ICU admission until the last MDRB acquisition. In patients with no MDRB, percentage of days with antibiotics was defined as the ratio of days with antibiotics during the whole ICU stay.

VAP was defined by the presence of new or progressive pulmonary infiltrate, associated with two of the following findings: temperature ≥38.5 °C or <36 °C; leukocyte count ≥10 000/μL or <1500/μL; and purulent sputum or tracheal aspirate. Microbiological confirmation of pneumonia was required, and defined as bronchoalveolar lavage or tracheal aspirate with ≥104 CFU/mL and ≥106 CFU/mL, respectively [17]. Only first episodes of VAP diagnosed > 48 h of mechanical ventilation were taken into account. Other infections were defined according to the modified Centers for Disease Control criteria [18].

The primary objective was to determine the impact of MDRB colonization pressure on the risk for ICU-acquired MDRB colonization or infection. Secondary objective was the impact of ICU-acquired MDRB on outcome.

Statistical analysis

SPSS 11.5 software (SPSS, Chicago, IL, USA) was used for data analysis. Distribution of quantitative variables was tested. Normally distributed and skewed quantitative variables are presented as mean ± SD, and median (interquartile range), respectively. Results of qualitative variables are presented as numbers (percentage). All p values are two-tailed. The statistical significance was set at p < 0.05.

Univariate analysis was used to determine variables associated with acquisition of MDRB during ICU stay. Categorical variables were compared using Pearson chi-square, or Fischer exact test. Quantitative variables were compared using Student’s t-test or Mann–Whitney U test, as appropriate. Exposure to all risk factors for MDRB was taken into account until the last acquisition of MDRB, or ICU discharge, whichever happened first. Multivariate analysis was used to determine variables independently associated with acquisition of MDRB. All predictors showing an association at p <0.1 with infection or colonization caused by MDRB in univariate analysis were included in the multivariate logistic regression analysis. Potential interactions were tested, and goodness of fit was assessed by the Hosmer-Lemeshow test. Cox proportional hazards univariate and multivariate models were also used to determine risk factors for ICU-acquired MDRB.

To determine the impact of ICU-acquired MDRB on outcome, duration of mechanical ventilation, ICU length of stay, and ICU mortality were compared between patients with MDRB, and patients with no MDRB. Risk factors for ICU mortality were determined using univariate analysis. All predictors showing an association at p <0.1 with ICU mortality in univariate analysis were included in the multivariate logistic regression analysis. Potential interactions were tested, and goodness of fit was assessed by the Hosmer-Lemeshow test.

Results

Five hundred ninety-three consecutive patients were included during the study period (Fig. 1), of which 204 (34%) acquired MDRB (colonization and/or infection) during their ICU stay. The most common MDRB were P. aeruginosa (38%), A. baumannii (22%), ESBL-GNB (18%), MRSA (15%), and S. maltophilia (7%). ICU-acquired infection related to MDRB occurred in 111 patients (19%). VAP and ICU-acquired bacteremia were the most common ICU-acquired infections related to MDRB. Prior colonization related to MDRB was significantly higher in patients with ICU-acquired infections related to MDRB compared with those who had infections related to other bacteria (86 of 111 (77%) versus 44 of 108 (41%), p < 0.001, OR (95% CI) 5 (2.8-9)). No outbreak occurred during the study period, but only endemic transmission of MDRB. No significant difference was found in rate of ICU-acquired MDRB between the three study ICUs (89 of 226 (39%), 63 of 181 (35%), and 73 of 186 (39%), p = 0.58). Patient characteristics are presented in Tables 1 and 2.

Fig. 1
figure 1

Study flowchart

Full size image
Table 1 Patient characteristics at ICU admission
Full size table
Table 2 Patient characteristics during ICU stay
Full size table

Risk factors for ICU-acquired MDRB by univariate analysis

At ICU admission, age, SAPS II, LOD score, transfer from other wards, prior antibiotic treatment, surgery, and infection at admission were identified as risk factors for MDRB (Table 1).

During ICU stay, colonization pressure, use and duration of intravascular and urinary catheters, sedation, mechanical ventilation, and ICU length of stay were identified as risk factors for MDRB by univariate analysis (Table 2). Percentage of days with antimicrobials, including pipercillin-tazobactm, fourth generation cephalosporins, carbapenem, fluoroquinolones, aminoglycoside, and glycopeptides was significantly higher in patients with ICU-acquired MDRB, compared with those with no MDRB (Table 3).

Table 3 Exposure to antimicrobials during ICU stay
Full size table

Colonization pressure for different ICU-acquired MDRB was also significantly higher in patients with ICU-acquired MDRB compared with those with no MDRB (Fig. 2).

Fig. 2
figure 2

Colonization pressure according to different multidrug resistant bacteria. P < 0.05 for all comparisons

Full size image

Risk factors for ICU-acquired MDRB by multivariate analysis

Colonization pressure, need for mechanical ventilation and the presence of an arterial catheter were independently associated with ICU-acquired MDRB (Table 4).

Table 4 Risk factors for ICU-acquired MDRB by multivariate analysis
Full size table

Risk factors for ICU-acquired MDRB by univariate and multivariate Cox proportional hazards models

Colonization pressure was also identified as a risk factor for ICU-acquired MDRB by univariate and multivariate Cox proportional hazards models (Table 5).

Table 5 Risk factors for ICU-acquired MDRB by univariate and multivariate Cox proportional hazards models
Full size table

Impact of ICU-acquired MDRB on outcome

Total duration of mechanical ventilation, length of ICU stay, and ICU mortality rate were significantly higher in patients with ICU-acquired MDRB, compared with those with no ICU-acquired MDRB (Table 2). However, ICU-acquired MDRB was not independently associated with ICU-mortality (Table 6).

Table 6 Risk factors for ICU-mortality
Full size table

Discussion

Our results suggest that colonization pressure is independently associated with ICU-acquired MDRB. Mechanical ventilation and arterial catheter use were also identified as independent risk factors for ICU-acquired MDRB. Further, acquisition of MDRB in the ICU was associated with increased duration of mechanical ventilation, ICU length of stay, and mortality rate. However, ICU-acquired MDRB was not independently associated with ICU mortality.

The strengths of our study are the large number of included patients, the daily calculation of colonization pressure in all participating units and patients, and the fact that this study is the first to evaluate the relationship between colonization pressure and all ICU-acquired MDRB. Previous studies identified colonization pressure as an independent risk factor for specific MDRB. However, none of these studies evaluated the relationship between colonization pressure and all ICU-acquired MDRB. Colonization pressure was previously identified as an independent risk factor for VRE acquisition in a medical ICU with non-individual rooms [12], and in an ICU setting with individualized rooms [19]. Single center and multicenter studies have shown MRSA-related colonization pressure to be independently associated with MRSA acquisition, not only in the ICU setting but also in medicine wards [20, 21]. As far as Gram-negative bacilli in ICU are concerned, two recent studies established colonization pressure as a risk factor for carbapenem-resistant A. baumannii [14] and multiresistant P. aeruginosa [15].

Our results could be explained by the higher risk for cross-transmission of MDRB in units where colonization pressure is high. However, molecular typing was not performed to confirm this hypothesis. Cross-transmission of MDRB might have occurred directly between two patients via health-workers or indirectly via environment. Previous studies clearly showed that the risk for ICU-acquired MDRB was significantly higher when prior room occupant was colonized or infected with MDRB [22–24].

Clinical implications of our findings include the potential usefulness of cohorting patients with MDRB, and the importance of screening for MDRB at ICU admission and during ICU stay. Based on our results and those of previous studies, one could argue that cohorting of staff or of ICU patients with MDRB might be beneficial in reducing transmission of ICU-acquired MDRB. Whilst no interventional study has shown beneficial effect of such an intervention, European Society of Clinical Microbiology and Infectious Diseases guidelines recommend patient cohorting during outbreaks only [25], and European Centers for Disease Control recommends staff cohorting in all settings, and patient cohorting in outbreaks only [26]. However, additional measures such as hand hygiene, environment cleaning, isolation measures, and antibiotic stewardship should be used to reduce cross-transmission of MDRB. Another potential implication of our findings is that preventive strategies aiming at reducing cross-transmission of MDRB should be enhanced in ICUs with high colonization pressure.

The present study clearly argues for a systematic screening of MDRB among ICU patients, thus isolation contact measures could be performed in these patients to stop the spread of MDRB. Recent studies reported conflicting results regarding the efficiency of contact isolation measures [27, 28]. However, recent recommendations clearly encourage physicians to isolate patients with MDRB [25].

Mechanical ventilation and arterial catheter use were identified as independent risk factors for ICU-acquired MDRB. Further, ICU-acquired MDRB are significantly associated with negative impact on outcome. These results are in accordance with other reports [5, 29, 30].

In addition to the above-discussed limitations, this study was performed in a single center, which precludes generalization of its results to other centers. Second, actual workload, compliance with hand hygiene, and room cleansing protocols were not evaluated. Third, our definition of MDRB included all A. baumannii, and S. maltophilia strains. However, if another definition, taking into account only resistant strains of these bacteria, had been used, different study results might have been obtained. Finally, our study was focused on all MDRB, with different virulence, risk factors and prevalence. However, analysis of specific MDRB showed similar results regarding the relationship between colonization pressure and ICU-acquired MDRB.

Conclusion

Colonization pressure is independently associated with ICU-acquired MDRB. Our results suggest that cohorting staff or critically ill patients with MDRB and enhancement of preventive measures in ICUs with high colonization pressure might be beneficial to reduce cross-transmission of MDRB. Further large interventional multicenter studies are required to confirm our findings.