Economic impact of an outbreak of carbapenemase producing-Enterobacteriaceae in a surgical intensive care unit.

BACKGROUND: A 15-month outbreak of carbapenemase-producing Enterobacterales (CPE) occurred in the surgical intensive care unit (SICU) of our institution. We aimed to estimate the financial impact of this outbreak from the perspective of the French public health insurance system. METHODS: The characteristics of the colonised/infected CPE patients and outbreak management according to French national guidelines were prospectively collected. Loss of productivity was assessed in terms of the reduction in total number of admissions (TNA) and discharges and in ICU length of stay (LoS). The additional financial burden associated with this outbreak was estimated by the accounting department of the hospital, including the impact of the extended LoS and restricted admissions. RESULTS: Sixteen CPE patients (19 stays) were hospitalised in the SICU (10/2016-01/2018). The median ICU LoS for the CPE cases was 17 [8-36] days versus 6.5 and 6.1 days in 2016 and 2017, respectively, for the whole SICU population. The total number of lost bed days during the outbreak was 452. The TNA dropped dramatically in 2017 (decrease of 20.6%). The estimated costs were 768,386 EUR for bed days lost; 297,176 EUR and 63,675 EUR for the extended LoS for the CPE cases and the patients on contact precautions, respectively; 34,045 EUR for staff reinforcements; 85,764 EUR for bacteriological screening tests; and 42,857 EUR for antimicrobial treatment. The total financial burden of the outbreak was 1,291,903 EUR. CONCLUSION: Management of a CPE outbreak in the SICU is associated with a huge financial burden for the unit and for the institution.

Simultaneous Hospital Outbreaks of New Delhi Metallo-ß-Lactamase-Producing Enterobacterales Unraveled Using Whole-Genome Sequencing.

Multidrug-resistant Enterobacterales, including carbapenemase producers, are currently spreading in health care facilities and the community. The Bichat Claude Bernard hospital in Paris faced a prolonged NDM-producing Enterobacterales (NDM-CPE) outbreak. Whole-genome sequencing (WGS) was performed on all isolated NDM-CPE to evaluate its benefits for outbreak surveillance and comprehension. All NDM-CPE isolates collected during the outbreak period (August 2016 to January 2018) were sequenced using the Illumina NextSeq platform. Gene content and core genomes were compared. Genomics results underwent epidemiological analysis which classified NDM-CPE cases as imported (positive sample during the 48 h after admission), hospital acquired, or uncertain. Over the epidemic period, 61 patients were colonized or infected with 81 distinct NDM-CPE isolates. Klebsiella pneumoniae was the most common species (n = 52, 64%), followed by Escherichia coli (13.5%) and other species (22.5%). In all, 43/52 (83%) K. pneumoniae isolates were clonal (≤18 single nucleotide polymorphisms [SNPs] except for three isolates) and belonged to ST307. The IncFIIK [K2:A-/B-] plasmid carrying blaNDM-1 present in all ST307 K. pneumoniae isolates was also detected in 18 other NDM-CPE isolates. Additionally, eight clonal ST144 Klebsiella oxytoca (≤18 SNPs) isolates lacking the epidemic plasmid were observed. The WGS analyses confirmed the acquired and imported cases except for two patients and resolved uncertain cases, which all turned out to be hospital acquisitions. WGS coupled with epidemiological analysis unraveled three epidemic phenomena: mainly the spread of a clonal ST307 K. pneumoniae strain and its conjugative plasmid carrying blaNDM-1 but also the unexpected clonal spread of an ST144 K. oxytoca strain. IMPORTANCE Carbapenemase-producing Enterobacterales (CPE) can spread and cause outbreaks in health care facilities, resulting in increased lengths of stay and morbidity. Control of outbreaks requires epidemiological surveillance, usually based on microbiological screening and patient follow-up. These data are sometimes insufficient to identify the routes of dissemination. There is therefore a need for more accurate tools such as whole-genome sequencing (WGS), which allows comparison of isolates but also plasmids carrying resistance with a high definition. In this work, we retrospectively sequenced the genomes of all NDM-producing Enterobacterales isolated during a prolonged NDM outbreak in our hospital. We demonstrated the value of combining WGS with epidemiological data that unveiled the multiple mechanisms of dissemination involved in the outbreak and confirmed transmission cases. This work reinforces the potential of WGS in outbreak surveillance and suggests that it could improve outbreak control if used in real time by confirming transmission cases more rapidly.

Costs associated with implementation of a strict policy for controlling spread of highly resistant microorganisms in France.

OBJECTIVE: To assess costs associated with implementation of a strict 'search and isolate' strategy for controlling highly drug-resistant organisms (HDRO). DESIGN: Review of data from 2-year prospective surveillance (01/2012 to 12/2013) of HDRO. SETTING: Three university hospitals located in northern Paris. METHODS: Episodes were defined as single cases or outbreaks of glycopeptide-resistant enterococci (GRE) or carbapenemase-producing Enterobacteriacae (CPE) colonisation. Costs were related to staff reinforcement, costs of screening cultures, contact precautions and interruption of new admissions. Univariate analysis, along with simple and multiple linear regression analyses, was conducted to determine variables associated with cost of HDRO management. RESULTS: Overall, 41 consecutive episodes were included, 28 single cases and 13 outbreaks. The cost (mean ± SD) associated with management of a single case identified within and/or 48 h after admission was €4443 ± 11,552 and €11,445 ± 15,743, respectively (p<0.01). In an outbreak, the total cost varied from €14,864 ± 17,734 for an episode with one secondary case (€7432 ± 8867 per case) to €136,525 ± 151,231 (€12,845 ± 5129 per case) when more than one secondary case occurred. In episodes of single cases, contact precautions and microbiological analyses represented 51% and 30% of overall cost, respectively. In outbreaks, cost related to interruption of new admissions represented 77-94% of total costs, and had the greatest financial impact (R(2)=0.98, p<0.01). CONCLUSIONS: In HDRO episodes occurring at three university hospitals, interruption of new admissions constituted the most costly measure in an outbreak situation.

Ventilator-associated pneumonia due to carbapenem-resistant Gram-negative bacilli in an intensive care unit without carbapenemase-producing Enterobacteriaceae or epidemic Acinetobacter baumannii.

Risk factors for ventilator-associated pneumonia (VAP) caused by carbapenem-resistant Gram-negative bacilli (CR-GNB) have rarely been evaluated in intensive care units (ICU) without epidemic carbapenemase-producing Acinetobacter baumannii or Enterobacteriaceae. We addressed this issue in a cohort of 141 patients (previous antimicrobial exposure, n = 131) with a first episode of VAP in a medico-surgical ICU. Twenty-six VAP (18.4%) involved a CR-GNB (Pseudomonas aeruginosa, n = 14, Stenotrophomonas maltophilia, n = 11, and A. baumannii, n = 1), without previous carbapenem exposure in 12 (46.1%) cases. GNB resistant to all ß-lactams except carbapenems were equally isolated in CR-GNB VAP (co-infections, 23%) and other episodes (30%). Previous exposure to aminoglycosides (odds ratio (OR) 1.14 per day, 95% confidence interval (CI) 1.02-1.30, p = 0.02) and the number of antimicrobial classes used before VAP (OR 1.38 per class, 95% CI 1.10-1.73, p = 0.006) were the only independent predictors of CR-GNB. These results suggest that the empirical use of a carbapenem-colistin combination should be evaluated in late-onset VAP following broad-spectrum antimicrobial exposure.

Rapid detection of glycopeptide-resistant enterococci: impact on decision-making and costs.

BACKGROUND: According to French national recommendations, the detection of a patient colonized with glycopeptide-resistant enterococci (GRE) leads to interruption of new admissions and transfer of contact patients (CPs) to another unit or healthcare facility, with weekly screening of CPs. FINDINGS: We evaluated the medical and economic impact of a pragmatic adaptation of national guidelines associated with a real-time PCR (RTP) (Cepheid Xpert™ vanA/vanB) as part of the strategy for controlling GRE spread in two medical wards. Screening was previously performed using chromogenic selective medium (CSM). Turn around time (TAT), costs of tests and cost of missed patient days were prospectively collected. In February 2012, the identification of GRE in one patient in the diabetology ward led to the screening of 31 CPs using CSM; one secondary case was identified in a CP already transferred to the Nephrology ward. Awaiting the results of SCM (median TAT, 70.5 h), 41 potential patient days were missed, due to interruption of admissions. The overall cost (screening tests + missing patient.days) was estimated at 14, 302.20 €. The secondary case led to screening of 22 CPs in the Nephrology ward using RTP. Because of a short median TAT of 4.6 h, we did not interrupt admissions and patients' transfers. Among 22 CPs, 19 (86%) were negative for vanA, 2 were positive for vanB and 3 had invalid results needing CSM. The overall cost of the strategy was estimated at 870.40 € (cost of screening tests only), without missing patient days. CONCLUSION: The rapid PCR test for vanA-positive GRE detection both allowed rapid decision about the best infection control strategy and prevented loss of income due to discontinuation of patient transfers and admissions.

Duration of colonization by extended-spectrum ß-lactamase-producing Enterobacteriaceae after hospital discharge.

BACKGROUND: The duration of gastrointestinal colonization with extended-spectrum ß-lactamase-producing Enterobacteriaceae (ESBL-E) may play a major role in the spread of these organisms. We evaluated the time to, and factors associated with, ESBL-E clearance after hospital discharge. METHODS: We retrospectively reviewed prospective surveillance results obtained over 14 years in a 1,000-bed hospital. The surveillance collected demographic, hospital stay, microbiologic, and outcome data. An automatic alert system identified readmitted patients with prior ESBL-E carriage. ESBL-E clearance was defined as a negative rectal screening sample at readmission with no new positive clinical sample during the stay. Variables associated with ESBL-E clearance were identified using a Cox model. RESULTS: We included 1,884 patients with 2,734 admissions. Four hundred forty-eight patients with readmission screening formed the basis for the study. Of 448 patients with 1 to 16 readmissions, 180 (40%) were persistent carriers. The median time to ESBL-E clearance was 6.6 months. Variables independently associated with clearance was having the first positive culture in a screening sample only (adjusted hazard ratio, 1.31; 95% confidence interval, 1.02-1.69; P = .04) and period 2005-2010 (hazard ratio, 1.88; 95% confidence interval, 1.33-2.67; P < .01). CONCLUSION: We found a long duration of ESBL-E carriage after hospital discharge. An automatic alert system was useful for identifying, screening, and isolating previous ESBL-E carriers.

Is universal HBV vaccination of healthcare workers a relevant strategy in developing endemic countries? The case of a university hospital in Niger.

BACKGROUND: Exposure to hepatitis B virus (HBV) remains a serious risk to healthcare workers (HCWs) in endemic developing countries owing to the strong prevalence of HBV in the general and hospital populations, and to the high rate of occupational blood exposure. Routine HBV vaccination programs targeted to high-risk groups and especially to HCWs are generally considered as a key element of prevention strategies. However, the high rate of natural immunization among adults in such countries where most infections occur perinatally or during early childhood must be taken into account. METHODOLOGY/PRINCIPAL FINDINGS: We conducted a cross sectional study in 207 personnel of 4 occupational groups (medical, paramedical, cleaning staff, and administrative) in Niamey's National Hospital, Niger, in order to assess the prevalence of HBV markers, to evaluate susceptibility to HBV infection, and to identify personnel who might benefit from vaccination. The proportion of those who declared a history of occupational blood exposure ranged from 18.9% in the administrative staff to 46.9% in paramedical staff. Only 7.2% had a history of vaccination against HBV with at least 3 injections. Ninety two percent were anti-HBc positive. When we focused on170 HCWs, only 12 (7.1%) showed no biological HBV contact. Twenty six were HBsAg positive (15,3%; 95% confidence interval: 9.9%-20.7%) of whom 8 (32%) had a viral load >2000 IU/ml. CONCLUSIONS/SIGNIFICANCE: The very small proportion of HCWs susceptible to HBV infection in our study and other studies suggests that in a global approach to prevent occupational infection by bloodborne pathogens, a universal hepatitis B vaccination of HCWs is not priority in these settings. The greatest impact on the risk will most likely be achieved by focusing efforts on primary prevention strategies to reduce occupational blood exposure. HBV screening in HCWs and treatment of those with chronic HBV infection should be however considered.

Diversity of extended-spectrum ß-lactamase-producing Enterobacteriaceae on hospital admission.

Extended-spectrum ß-lactamase (ESBL)-producing Enterobacteriaceae have become prevalent in both the hospital and the community. We describe the epidemiology of ESBL-producing isolates and patient characteristics at hospital admission. Data on clinical properties, medical history, previous hospitalizations, and previous antibiotic treatments were collected. ESBL genes (bla(CTX-M), bla(TEM), and bla(SHV)) were identified by polymerase chain reaction. One hundred and sixteen patients carried 122 ESBL-producing Enterobacteriaceae: 79 Escherichia coli, 26 Klebsiella pneumoniae, 16 Enterobacter spp., and 1 Citrobacter koseri. ESBL-producing E. coli were associated with admission from home (odds ratio (OR) 3.04, p = 0.02) and a history of recent urinary tract infection (OR 3.38, p = 0.04), and exhibited a lower rate of antimicrobial resistance to aminoglycosides (p ≤ 0.005) and co-trimoxazole (p = 0.003), whereas other ESBL-producing Enterobacteriaceae tended to be associated with a recent surgery (OR 0.42, p = 0.057). Although the CTX-M enzymes were more frequently found in E. coli (76%), they were also identified in other Enterobacteriaceae (45%), suggesting penetration of CTX-M-type enzymes into both community- and hospital-acquired enterobacterial species.

Development of a phenotypic method for detection of fecal carriage of OXA-48-producing enterobacteriaceae after incidental detection from clinical specimen.

We report incidental isolation of an OXA-48-producing Escherichia coli strain in urine of a 62-year-old woman recently returning from a 2-month vacation in Morocco. Commercially available extended-spectrum beta-lactamase (ESBL)-targeting medium failed to detect it in the patient's stools, although a locally developed and easy-to-implement method using ertapenem-supplemented brain heart infusion (BHI) broths could.

Long-term impact of a multifaceted prevention program on ventilator-associated pneumonia in a medical intensive care unit.

BACKGROUND: Ventilator-associated pneumonia (VAP), the most common hospital-acquired infection in intensive care units, increases mortality and health care costs. We describe the long-term impact of a multifaceted program for decreasing VAP rates that markedly improved compliance with 8 targeted preventive measures. METHODS: We compared VAP rates during a 45-month baseline period and a 30-month intervention period in a cohort of patients who received mechanical ventilation for > 48 h. VAP was diagnosed on the basis of quantitative cultures of distal specimens. VAP incidence density rates were expressed as total VAP episodes over total mechanical ventilation duration and as first VAP episodes over mechanical ventilation duration at VAP or hospital discharge. We used segmented regression analysis and a Cox proportional hazard model to assess the impact of the program on first VAP occurrence. RESULTS: Baseline and intervention VAP rates were 22.6 and 13.1 total VAP episodes over total mechanical ventilation duration per 1000 ventilation-days, respectively, and 26.1 and 14.9 first VAP episodes over mechanical ventilation duration at VAP or hospital discharge per 1000 procedure-days, respectively (P < .001). VAP rates decreased by 43% in both statistical analyses and remained significant after adjustment for confounders (Cox adjusted hazard ratio, 0.58; 95% confidence interval, 0.46-0.72; P < .001). Daily VAP hazard rates on ventilation days 5, 10, and 15 were 2.6%, 3.5%, and 3.4%, respectively, during the baseline period and 1.4%, 2.3%, and 2%, respectively, during the intervention period. CONCLUSION: Our preventive program produced sustained VAP rate decreases in the long term. However, VAP rates remained substantial despite high compliance with preventive measures, suggesting that eliminating VAP in the intensive care unit may be an unrealistic goal.

Changes in knowledge, beliefs, and perceptions throughout a multifaceted behavioral program aimed at preventing ventilator-associated pneumonia.

PURPOSE: Prevention of ventilator-associated pneumonia (VAP) requires a complex approach that should include factors affecting healthcare workers' (HCWs) behavior. This study attempted to assess change of individual factors throughout a multifaceted program focusing on VAP prevention. METHODS: The prevention program involved all HCWs in a 20-bed medical intensive care unit (ICU) and included a multidisciplinary task force, an educational session, direct observations and performance feedback, technical improvements, and reminders. Knowledge, beliefs, and perceptions (cognitive factors) were assessed with a test and a self-reporting questionnaire based on social-cognitive theories. They were completed before and 1 and 12 months after the educational session. RESULTS: Of the 100 HCWs initially evaluated, 84 were present 1 year later. Overall, individual factors (knowledge and cognitive factors) changed positively and significantly, immediately after the educational session. Five cognitive factors were significantly associated with knowledge: perceived susceptibility, seriousness, knowledge, benefits, and self-efficacy (P < 0.05). The other factors, i.e., perceived barriers, subjective and behavioral norm, intention to perform action, and motivation, were not. The positive cognitive change was significantly reinforced at 1 year. Three distinct cognitive profiles derived from answers to the baseline questionnaire were individualized. The positive impact of our behavioral approach was highest for the HCW group with the lowest baseline cognitive profiles. CONCLUSIONS: Behavior changed gradually throughout the program and was especially pronounced for HCWs with the lowest baseline cognitive profiles.

Needlestick injury rates according to different types of safety-engineered devices: results of a French multicenter study.

OBJECTIVES: To evaluate the incidence of needlestick injuries (NSIs) among different models of safety-engineered devices (SEDs) (automatic, semiautomatic, and manually activated safety) in healthcare settings. DESIGN: This multicenter survey, conducted from January 2005 through December 2006, examined all prospectively documented SED-related NSIs reported by healthcare workers to their occupational medicine departments. Participating hospitals were asked retrospectively to report the types, brands, and number of SEDs purchased, in order to estimate SED-specific rates of NSI. Setting. Sixty-one hospitals in France. RESULTS: More than 22 million SEDs were purchased during the study period, and a total of 453 SED-related NSIs were documented. The mean overall frequency of NSIs was 2.05 injuries per 100,000 SEDs purchased. Device-specific NSI rates were compared using Poisson approximation. The 95% confidence interval was used to define statistical significance. Passive (fully automatic) devices were associated with the lowest NSI incidence rate. Among active devices, those with a semiautomatic safety feature were significantly more effective than those with a manually activated toppling shield, which in turn were significantly more effective than those with a manually activated sliding shield (P < .001, chi(2) test). The same gradient of SED efficacy was observed when the type of healthcare procedure was taken into account. CONCLUSIONS: Passive SEDs are most effective for NSI prevention. Further studies are needed to determine whether their higher cost may be offset by savings related to fewer NSIs and to a reduced need for user training.

A multifaceted program to prevent ventilator-associated pneumonia: impact on compliance with preventive measures.

OBJECTIVE: To determine the effect of a 2-yr multifaceted program aimed at preventing ventilator-acquired pneumonia on compliance with eight targeted preventive measures. DESIGN: Pre- and postintervention observational study. SETTING: A 20-bed medical intensive care unit in a teaching hospital. PATIENTS: A total of 1649 ventilator-days were observed. INTERVENTIONS: The program involved all healthcare workers and included a multidisciplinary task force, an educational session, direct observations with performance feedback, technical improvements, and reminders. It focused on eight targeted measures based on well-recognized published guidelines, easily and precisely defined acts, and directly concerned healthcare workers' bedside behavior. Compliance assessment consisted of five 4-wk periods (before the intervention and 1 month, 6 months, 12 months, and 24 months thereafter). MEASUREMENTS AND MAIN RESULTS: Hand-hygiene and glove-and-gown use compliances were initially high (68% and 80%) and remained stable over time. Compliance with all other preventive measures was initially low and increased steadily over time (before 2-yr level, p < .0001): backrest elevation (5% to 58%) and tracheal cuff pressure maintenance (40% to 89%), which improved after simple technical equipment implementation; orogastric tube use (52% to 96%); gastric overdistension avoidance (20% to 68%); good oral hygiene (47% to 90%); and nonessential tracheal suction elimination (41% to 92%). To assess overall performance of the last six preventive measures, using ventilator-days as the unit of analysis, a composite score for preventive measures applied (range, 0-6) was developed. The median (interquartile range) composite scores for the five successive assessments were 2 (1-3), 4 (3-5), 4 (4-5), 5 (4-6), and 5 (4-6) points; they increased significantly over time (p < .0001). Ventilator-acquired pneumonia prevalence rate decreased by 51% after intervention (p < .0001). CONCLUSIONS: Our active, long-lasting program for preventing ventilator-acquired pneumonia successfully increased compliance with preventive measures directly dependent on healthcare workers' bedside performance. The multidimensional framework was critical for this marked, progressive, and sustained change.

[Effect of a long-term quality improvement program on the risk of infection related to peripheral venous catheters]. / Impact d'un programme prolongé d'amélioration continue de la qualité sur le risque infectieux lié aux cathéters veineux périphériques.

INTRODUCTION: Although peripheral venous catheter (PVC) placement is one of the most common invasive procedures used in hospitals, data about the infectious risk associated with it are sparse. Nurses and physicians often underestimate this risk. We describe here a 10-year continuous quality improvement program in a large university hospital. METHODS: The prevention program included: 1/ an observational audit of practices of PVC insertion and maintenance (1996), 2/ two studies of incidence and risk factors for PVC-related infection (1996 and 1999), 3/ 8 annual (1996 to 2006) cross-sectional studies that collected information about duration of PVC placement and other process indicators, 4/ surveillance of nosocomial bacteremia from 2002 through 2006, and 5/ continued education sessions for healthcare workers, based on local surveillance data. RESULTS: Rates of colonization and infection (0.7 cases of bacteremia/1000 PVC days) were similar to data from other studies. PVC that remained in place for more than three days was associated with a higher risk of catheter-related bacteremia in both incidence studies. The annual prevalence studies showed that 15 to 20% of PVCs remained in place for more than three days from 1996 through 2002, but this rate decreased thereafter; it was 6.7% in 2005 and 10.7% in 2006. Practices audits indicated room for improvement in prevention measures. The absolute number of PVC-related bacteremia decreased by more than 50% from 2002 (n=23) to 2006 (n=10). CONCLUSION: This prolonged multifaceted program, including practice audits, continued medical education and monitoring of PVC-related infection, has led to a substantial decrease over time in the infectious risk related to PVC placement.

Epidemiology of multidrug-resistant bacteria in patients with long hospital stays.

OBJECTIVE: To determine rates of colonization with multidrug-resistant (MDR) bacteria (ie, methicillin-resistant Staphylococcus aureus [MRSA], vancomycin-resistant Enterococcus [VRE], extended-spectrum beta -lactamase [ESBL]-producing Enterobacteriaceae, and Acinetobacter baumannii) after prolonged hospitalization and to assess the yield of surveillance cultures and variables associated with colonization with MDR bacteria. DESIGN: Prospective observational cohort study conducted from February 6 to May 26, 2006. METHODS: All patients who spent more than 30 days in our university hospital (Paris, France) were included. Rectal and nasal swab samples obtained during day 30 screening were examined for MRSA, VRE, ESBL-producing Enterobacteriaceae, and A. baumannii. RESULTS: Of 470 eligible patients, 439 had surveillance culture samples available for analysis, including 51 patients (11.6%) with a history of colonization or infection due to 1 or more types of MDR bacteria (MRSA, recovered from 35 patients; ESBL-producing Enterobacteriaceae, from 16 patients; A. baumannii, from 6 patients; and VRE, from 0 patients) and 37 patients (9.5% of the 388 patients not known to have any of the 4 MDR bacteria before day 30 screening) newly identified as colonized by 1 or more MDR bacteria (MRSA, recovered from 20 patients; ESBL-producing Enterobacteriaceae, from 16 patients; A. baumannii, from 1 patient; and VRE, from 0 patients). A total of 87 (19.8%) of 439 patients were identified as colonized or infected with MDR bacteria at day 30. Factors that differed between patients with and without MRSA colonization included age, McCabe score, comorbidity score, receipt of surgery, and receipt of fluoroquinolone treatment. Patients with ESBL-producing Enterobacteriaceae colonization were younger than patients with MRSA colonization. CONCLUSIONS: Differences in the variables associated with MRSA colonization and ESBL-producing Enterobacteriaceae colonization suggest differences in the epidemiology of these 2 organisms. Day 30 screening resulted in a 72.5% increase in the number of patients identified as colonized with at least 1 type of MDR bacteria.

Successful long-term program for controlling methicillin-resistant Staphylococcus aureus in intensive care units.

OBJECTIVE: To evaluate the effectiveness of screening strategy and contact precautions for patients with methicillin-resistant Staphylococcus aureus (MRSA). DESIGN AND SETTING: Prospective observational cohort from 1 February 1995 to 31 December 2001 in three intensive care units (45 beds) in a French teaching hospital. PATIENTS: 8,548 patients admitted to the three ICUs had nasal screening on ICU admission and weekly thereafter. Contact precautions were used in MRSA-positive patients. The following variables were collected: age, gender, severity score, length of stay, workload, and colonization pressure (percentage of patient-days with an MRSA to the number of patient-days in the unit). Alcohol-based handrub solution was introduced in July 2000. We compared the period before this (P1) with that thereafter (P2). RESULTS: Of the 8,548 admitted patients 554 (6.5%) had MRSA at ICU admission, and 456 of the 7,515 (6.1%) exposed patients acquired MRSA. Acquisition incidence decreased from 7.0% in P1 to 2.8% in P2. Independent variables associated with MRSA acquisition were: age (adjusted odds ratio 1.013), severity score (1.047), length of ICU stay (1.015), colonization pressure (1.019), medical ICU (1.58), and P2 (0.49). CONCLUSIONS: MRSA control in these ICUs characterized by a high prevalence of MRSA at admission was achieved via multiple factors, including screening, contact precautions, and use of alcoholic handrub solution. Our results after adjustment of risk factors for MRSA acquisition and the steady improvement in MRSA over several years strengthen these findings. MRSA spreading can be successfully controlled in ICUs with high colonization pressure.