Social Network Modeling Can Control Hospital Acquired Infections
Computer models of interactions between patients and health care workers could be utilized to control deadly hospital acquired infections, according to recent findings by Smith School researchers Sean Barnes, assistant professor of operations management, and Bruce Golden, the France-Merrick Chair in Management Science.
According to the Department of Health and Human Services (HHS), these transmissions strike one out of every 20 inpatients, drain billions of dollars from the national health care system and cause tens of thousands of deaths annually.
Collaborating with Edward Wasil of American University's Kogod School of Business, the Smith researchers utilized computer models to simulate patient-health care worker interactions to determine whether this network connectivity is a source for spreading multi-drug resistant organisms (MDROs). The findings correlate a “sparse, social network structure” with low infection transmission rates.
The research has come in advance of HHS’ launch and enforcement of a new initiative in 2015 that penalizes hospitals at an estimated average rate of $208.642 for violating specific requirements for infection control.
In response, the authors have introduced a conceptual framework for hospitals to model their social networks to predict and minimize the spread of bacterial infections that often are resistant to antibiotic treatments.
The authors manipulated and tracked the dynamics of the social network in a mid-Atlantic hospital’s intensive care unit. They focused on interactions between patients and health care workers – primarily nurses – and the multiple competing factors that can affect transmission.
The study demonstrated “a starkly lower” transmission rate in a “sparse network” made up of 45 connections in comparison to a “dense network” of 190 connections. Both of these networks represented the operations of a 20-patient ICU, but the former unit had 10 nurses whereas the latter unit had two nurses.
“The basic reality is that healthcare workers frequently cover for one another due to meetings, breaks and sick leave,” said Barnes. “These factors, along with the operating health care-worker-to-patient ratios and patient lengths of stay, can significantly affect transmission in an ICU… But they also can be better controlled.”
The next step is to enable hospitals to adapt this framework, which is based on maximizing staff-to-patient ratio to ensure fewer nurses and physicians come in contact with each patient, especially high-risk patients.
In these illustrations of dense (left) and sparse (right) patient ICU social networks, patients that share a nurse are connected by a link, while patients that share a physician have the same color.
"The health care industry's electronic records movement could soon generate data that captures the structure of patient-healthcare worker interaction in addition to multiple competing, related factors that can affect MDRO transmission,” said Barnes.
“Such a working model would provide near-immediate feedback for infection control practitioners who historically rely on empirical studies and infrequent clinical trials to evaluate intervention strategies,” Barnes said. “You may not be able to change the amount of patient-sharing, but you can structure it in such a way as to prevent additional transmission.”
Barnes said the findings, in the meantime, should prompt managers to closely monitor nurse-patient interaction and strategize for fewer staff to come into contact with high-risk patients. A simple measure would be to employ more nurses. But a more feasible approach is to devise paired or revolving sharing strategies for nurses covering for one another – strategies that minimize each worker's patient connections.
“Exploring the Effects of Network Structure and Healthcare Worker Behavior on the Transmission of Hospital-Acquired Infections,” appears in IIE Transactions on Healthcare Systems Engineering.
|Table of Contents||Next Article|