Modeling the transmission of vancomycin-resistant Enterococcus in hospitals: A case study

Nosocomial, or hospital-acquired infections, the fourth cause of death in the US, are evidence that hospitals provide not only medical care but also harbor pathogens that pose serious, often fatal, risks of infection, particularly to the young, the elderly, and immune-compromised individuals. Infection-control measures aimed at reducing their impact are being implemented with various degrees of efficiency at US hospitals. Data from general and oncology hospital units on Vancomycin-resistant Enterococcus (VRE), one of the most prevalent and dangerous pathogens involved, are used to highlight the importance of modeling nosocomial infection dynamics as a prelude to the testing and evaluation of control measures.;New mathematical models of the transmission dynamics of VRE in hospitals are introduced in order to identify and quantitatively assess the time evolution of nosocomial infections. Ordinary differential equation (ODE) and discrete delay differential (DDD) models in conjunction with statistical methods are used to estimate key population-level nosocomial transmission parameters. This framework is tested using unpublished surveillance data from two types of hospital units. The population is divided into uncolonized, VRE colonized, and VRE colonized-in-isolation categories and the use of constant and variable rates of isolation admitted with VRE or VRE-colonized during hospital stays is evaluated in models including health care workers hand-hygiene compliance. The process of model calibration detected irregularities in the available surveillance data; these irregularities that are most likely the result of the data recording-process. Efforts to fit data within our highly flexible dynamic-modeling framework suggest that clinical-trial level surveillance data is needed.;The usefulness of our new flexible modeling framework for the transmission dynamics of nosocomial infections like VRE was first evaluated using synthetic noisy data and then tested against the available data. Parameters whose estimates are required for the testing and evaluation of competing or integrated intervention/control measures via mathematical models could not be accurately estimated from the available data. It would be extremely difficult to take advantage of transmission dynamic models in the fight against nosocomial infections unless systemic and careful efforts to collect data are put in place at hospitals.