Controlling the spread of healthcare associated infections is a priority to reduce morbidity, mortality, and cost. Interventions and control strategies abound but have had inconsistent impacts. Intervention decisions are often based on model or simulation outputs that depend on input data and generally focus on one facility. Poor hand hygiene adherence contributes to the spread of disease and attempts to improve hand hygiene adherence have limited documented effect. The adherence measurements are based on observation and therefore error- prone, so actual adherence rates before and after interventions are uncertain in most studies. Transmission of disease occurs during contact between providers and between providers and patients. A recent study linking transmission to an observed contact network has clearly revealed the need for more data on contact. Limited small scale studies have used the electronic health record to create contact networks and wireless sensors to measure hand hygiene adherence and capture contact networks. Healthcare provider behaviors (contact and hand hygiene) vary among individuals, wards, and facilities so larger scale studies are needed in order to understand behavior across different facilities. Models to evaluate interventions must also consider multiple facilities and take these differences into account. We propose to extend the use of electronic health record and wireless sensor methods to multiple facilities in the Veterans Affairs Medical Centers (VAMC's) and expand model simulations to estimate and incorporate differences among facilities to evaluate control strategies for healthcare associated infections. Using the electronic health record we will link patients to providers who write notes in their record, thus creating a contact network. We will do this for 3 years of data from 78 VAMC's representing complexity and regions. We will use wireless sensors in three of these VAMC's to collect contact network and hand hygiene data. Wireless sensors detect proximity between people, occupancy of a room, and use of alcohol-based hand sanitizer and soap dispensers and sinks. There will be two two-week deployments at each site. An existing one-facility agent based model simulation will be expanded to input contact network and hand hygiene adherence data and be calibrated to multiple facilities. The data will be used to estimate differences in contact network characteristics among wards and facilities. Explanations for differences in contact networks may be due to different staff mixes, use of consults, types of wards, or even regional social norms. Hand hygiene rates will be assessed for differences among wards, health care provider roles, self-reported beliefs, and care situations. We hypothesize that differences will shed light on when, where, and by who hand hygiene adherence breaks down and therefore can lead to improvements. The expanded agent based model simulations will be used to evaluate interventions at multiple facilities that aim to interrupt transmission. Interventions targeting specific hand hygiene opportunities and behaviors as well as interventions on cohorting will be simulated. The data and model outputs from this project represent improvements in methodology aimed to improve health care quality for Veterans.
