Objetivo: Reduzir a transmissão de Enterobactérias resistentes a carbapenêmicos (ERC) em uma unidade de terapia intensiva com intervenções baseadas em simulações por um modelo matemático desenvolvido. Desenho: Ensaio antes e depois com um período basal de 44 semanas e um período de intervenção de 24 semanas. Métodos: O estudo foi conduzido em uma Unidade de Terapia Intensiva Médica (UTI) de um hospital de ensino terciário. Desenvolvemos um modelo de transmissão de ERC em uma UTI e medimos todos os parâmetros necessários para a entrada do modelo. Foram estabelecidas metas de adesão à higienização das mãos e às precauções de contato com base nas simulações e conduziu-se uma intervenção focada em atingir essas métricas como metas, com auditoria semanal da unidade e reuniões com a equipe para dar feedback sobre o desempenho. Resultados: As metas para a adesão à higienização das mãos e às precauções de contato foram alcançadas na terceira semana do período de intervenção. Durante o período basal, o R0 calculado foi de 11, a prevalência mediana de pacientes colonizados por ERC na unidade foi de 33% e ultrapassou 50% por três vezes. No período de intervenção, a prevalência média de pacientes colonizados por ERC passou para 21%, com uma mediana semanal do Rn de 0,42 (intervalo: 0 a 2,1). Conclusões: As simulações ajudaram a estabelecer e atingir metas específicas para controlar as altas taxas de prevalência de ERC e reduzir a transmissão de ERC dentro da unidade. O modelo foi capaz de prever os resultados observados. Este é o primeiro estudo em controle de infecção que mede a maioria das variáveis de um modelo matemático na vida real e o aplica como uma ferramenta de apoio à decisão para a intervenção / Objective: To reduce the transmission of carbapenem-resistant Enterobacteriaceae (CRE) in an intensive care unit with interventions based on simulations by a mathematical model. Design: Before-after trial with a 44-week baseline period and a 24 week intervention period. Setting: Medical Intensive Care unit (ICU) of a tertiary care teaching hospital. Participants: All patients admitted to the unit. Methods: We developed a model of transmission of CRE in an ICU and measured all necessary parameters for the model input. Goals of compliance with hand hygiene and with isolation precautions were established based on simulations. An intervention focused on achieving these goals was conducted with weekly auditing and feedback. Results: The goals for compliance with hand hygiene and contact precautions were reached on the third week of the intervention period. During baseline period, the calculated R0 was 11, median prevalence of patients colonized by CRE in the unit was 33% and exceeded 50% on three occasions. In the intervention period, median prevalence of colonized CRE patients was 21%, with a median weekly Rn of 0.42 (range: 0 to 2.1). Conclusions: The simulations helped to establish and to achieve specific goals in order to control the high prevalence rates of CRE and reduce CRE transmission within the unit. The model was able to predict the observed outcomes. This is the first study in infection control to measure most variables of a model in real life and to apply the model as a decision support tool for intervention

The University Hospital of Liège has 955 beds in 8 intensive care units, 15 medical wards, 10 surgical wards and 1 paediatric ward. Approximately 36,000 patients are admitted each year, giving a total of 265,000 patient-days hospitalization. Extended-spectrum beta-lactamase-producing Enterobacteriaceae (E-ESBL) constitute, along with methicillin-resistant Staphylococcus aureus (MRSA), the main multi-resistant bacteria recovered in our hospital. The aims of the present study were to: - evaluate the epidemiology of E-ESBL - evaluate the impact of an infection control programme to reduce the spread of E-ESBL in the University Hospital of Liège. In order to do this, several studies were carried out between 2001 and 2007: 1. Determination of the high risk units in the CHU (2001) The high risk units were determined by comparing the incidence rates of each type of unit. Two types of high risk unit were identified in this way: the Intensive Care Units (ICUs) and the Onco-Haematology Unit. 2. Epidemiology of E-ESBL in the Onco-Haematology unit (2002-2003 and 2005-2006) Digestive tract colonization by E-ESBL was found to be relatively high (7.3%) and this explains the high incidence of E-ESBL in Onco-Haematology in comparison with the rest of the hospital. However, the clinical gravity associated with exposure to the risk factor (digestive tract colonization by E-ESBL) was found to be relatively weak. 3. Importance of digestive tract colonization by E-ESBL in General ICUs (2002-2003) Digestive tract colonization by E-ESBL was found to be relatively common (8.8%) and faecal carriage of E-ESBL was found to be a good marker for infection with E-ESBL at another body site. Even though the number of infected patients was found to be low, the risk of infection due to E-ESBL was multiplied by 14.7 in a group of digestive carriers of E-ESBL with regard to a group of non carriers. Our data also showed that Enterobacter aerogenes is the most frequent species producing extended-spectrum beta-lactamase (ESBL) and that TEM-24 is the most prevalent ESBL produced by E-ESBL species in our ICUs. No CTX-M-type genes were identified. With regard to antibiotic susceptibility, meropenem and cefepime appeared to be the most active agents against the majority of isolates. 4. Impact of an infection control programme to reduce the spread of E-ESBL (2006-2007) A surveillance programme was carried out to evaluate the implementation of infection control procedures including surveillance of ESBL-producing strains, utilization of computer alerts for E-ESBL positive patients and the application of contact precautions for colonized or infected patients. Infection control compliance observations were performed by trained referring nurses. During the 2 years of application, one or more E-ESBL were identified in 500 patients. A total of 2268 internal messages regarding the identification of E-ESBL were sent within the hospital, among which 91.84 % were received (at least 1 for every patient). An alert was associated with 406 patients, who were always hospitalized as the identification of the E-ESBL by the laboratory was obtained. A total of 257 registration forms were filled in by the referring nurses, resulting in a survey compliance of 63%. This survey showed that door signs identifying positive patients, hydro-alcoholic solution and gloves were present in 90% of the cases, but that gowns were only present in 59%. The overall incidence of nosocomial acquisition of E-ESBL between 2006 and 2007 was 0.92/1000 patient-days, more or less the same as in 2002. In relation to this research, several questions remain: - Even though the rates of digestive tract colonization with E-ESBL in the 2 types of high risk unit were found to be more or less the same (7.3 and 8.8%), the impact on infections due to E-ESBL was very different. - Are the infections due either to E-ESBL endogenous infections (owing notably to the use of broad spectrum antibiotics) or to secondary infections (resulting from cross-transmission) or to both? The implementation of an infection control programme to limit the spread of E-ESBL has been based on the limitation of the cross-transmission of these micro-organisms. An enhanced barrier precautions policy has been in place in our institution for 2 years, and we have seen no erosion in compliance. We should not however lose sight of the fact that, whatever the institutional policy for the management of multi-resistant bacteria, the correct application of standard precautions for all patients is the first measure to limit the cross-transmission of all micro-organisms.