Philipsen, Kirsten Riber3; Bootsma, M. C. J.4; Leverstein-van Hall, M.A.5; Cohen Stuart, J.5; Bonten, M.J.M.5
1 Mathematical Statistics, Department of Informatics and Mathematical Modeling, Technical University of Denmark2 Department of Informatics and Mathematical Modeling, Technical University of Denmark3 National Food Institute, Technical University of Denmark4 Utrecht University5 University Medical Centre Utrecht
In this study a mathematical model for the spread of ESBL resistant E.coli among patients in a hospital and the surrounding catchment population has been introduced and used to described prevalence data from the Netherlands. Several statistical methods have been applied to estimate the model parameters. The patient flow data was studied by survival analysis. This enabled us to get an estimate for the time to readmission when discharged from either the low-risk or high-risk hospital wards. It is hypothesized that readmission plays a role for the spread of resistant bacteria. The high prevalence of EC+ and EC++ colonized patients in the core group with patients discharged from the high-risk ward, indicates that especially patients readmitted to the high-risk wards contribute to the increasing prevalence. It could be interesting in a future simulation study to further investigate the importance of readmission and the effect of different interventions on the prevalence. There are several theories with regards to the spread of ESBL resistant bacteria, but the actual prevalence data is very sparse. Based on the available data we have developed an adequate model that can explain the increase in prevalence from year 2000. It has not been possible to separate the e®ect from conjugation and cross-transfer on the ESBL prevalence of type CTX-M, as several combination of the cross-transfer rate and conjugation rate give a good ¯t to data. However, the model predicts that a minimum of 57% of the acquisition of EC++ colonization is due to cross transfer. The transfer rates for each hospital and community compartments are all related by a ratio fixed in the model. Due to the high prevalence of resistant bacteria in the high-risk ward, this ward is a central element for estimation of the model parameters. It could therefore be interesting to look at the transfer going on inside the high-risk wards alone. A surveillance study, where the colonization status of all patient in one or two hospital high-risk hospital wards are followed over a couple of month, could be an idea for a better understanding of the transfer mechanisms. The mean duration of colonization with EC+, EC++ and EB++ after discharge from the hospital has in this study been fixed to 141 days. Whether patients readmitted to the hospital are colonized with resistance bacteria is among other things dependent on the duration of colonization. It would therefore be interesting to investigate the effect of increased or decreased length of colonization by simulation studies. Furthermore the model could be improved, if data from colonization studies of each of the colonization states EC+, EC++ and EB++ were available.
Main Research Area:
Technical University of Denmark, DTU Informatics, Building 321, 2009