Drivers of Variation in Contact Networks in Long Term Care Facilities

University Of Utah, Salt Lake City UT

Investigators

Abstract

PROJECT SUMMARY Our overarching goal is to improve understanding of transmission of resistant bacteria by characterizing contact networks in a national sample of approximately 25 long-term care facilities (LTCFs). Each participating LTCF will be visited twice during the course of this study for primary data collection. Healthcare professional (HCP) ? resident contact networks will be ascertained in several different ways. We define contact networks as interactions between individuals that have the potential to mediate transmission. Our primary technique for estimating who contacted whom will be through self-report by health care professionals in the participating LTCFs. This approach will be compared to use of information about staff assignments, unit sub-divisions, and staff-resident ratios to make inferences about contact patterns. In addition, we will use direct observation to collect information about room entry rates and behaviors, such as hand hygiene, that influence whether an interaction results in acquisition. Facility and resident assessments will also be performed through chart review and survey. Resident level data will include presence of devices, infection status, antibiotic use, wound care, and comorbidities. In one to two of the participating facilities, we will use two new technologies to capture more detailed data on contacts within LTCFs. First, we will use a badge-based-sensor network to construct networks on the basis of personal proximity. This approach will allow us to examine how healthcare professionals and residents move within units of LTCFs and to analyze sequences of close contacts. However, this technology may not be sufficient to determine if the healthcare professional touched the patient. Thus, in a smaller number of rooms, we will use a second technological strategy, computer vision and depth sensing. This approach makes it feasible to collect highly granular data on behaviors such as use of gowns, gloves, and masks. Our analytical models will focus on dependencies between contact network structure, facility characteristics, and individual resident and HCP attributes. We will determine whether the number of unique contacts per individual are statistically dissimilar for different types of residents and HCPs. Alternative ways to derive contact networks will be compared with respect to overlap and completeness. Temporal models will also be used to examine dynamic changes in networks over time. Our hypothesis is that the structure of contact networks will vary according to the clinical services offered by LTCFs and that these structural properties will have significant implications for transmission and control.