Health Services Research & Development

Veterans Crisis Line Badge
Go to the ORD website
Go to the QUERI website

IIR 08-075 – HSR&D Study

New | Current | Completed | DRA | DRE | Portfolios/Projects | Centers | Career Development Projects

IIR 08-075
Improving MRSA Control through Simulation and Surveillance
Michael Adam Rubin MD PhD MS
VA Salt Lake City Health Care System, Salt Lake City, UT
Salt Lake City, UT
Funding Period: July 2009 - June 2012

BACKGROUND/RATIONALE:
The infection control community is highly divided on the most appropriate strategy to control transmission of methicilllin-resistant Staphylococcus aureus (MRSA) in hospitals. Our previous work suggests that an approach using targeted surveillance of patients classified as high risk for MRSA carriage using electronic data may be a practical consideration for an approach to MRSA surveillance. Given that large studies of infection control strategies are often prohibitively difficult and expensive, however, an approach using detailed computer simulations might provide important insight and direction for future research efforts.

OBJECTIVE(S):
Control of MRSA transmission is dependent on the interactions of innumerable factors and processes. With this in mind, our objectives are to (a) develop an electronic classification algorithm to identify VA patients at high risk of MRSA carriage at the time of hospital admission; (b) adapt, calibrate, and validate an agent-based computer simulation of MRSA transmission to the VA inpatient setting; (c) use this computer simulation to gain a deeper understanding of the factors that influence MRSA transmission and control in hospitals; and (d) evaluate and compare alternative policies for MRSA control in VA hospitals, including targeted surveillance, with a particular focus on assessing the cost-benefit of these different strategies.

METHODS:
First we created and validated an electronic classification rule to estimate risk of MRSA carriage in hospitalized veterans at the time of admission. Predictors of MRSA carriage in this population were determined through a retrospective cohort study using electronic data. We then adapted our simulation to the VA setting and created a base-case simulation scenario. We then assessed the various strategies and factors that impact MRSA transmission through simulation experiments. A targeted active surveillance strategy was included by incorporating the performance of our classification rule into the simulation. Traditional quantitative analytic methods were used, with MRSA prevalence and transmission rates as primary outcomes. Cost-benefit analyses were also performed using simulated cost figures.

FINDINGS/RESULTS:
First, we established a reference standard for proof of concept to use nationwide microbiology data from Patient Care Services (PCS). Organism and susceptibility information from text-based microbiology reports from most VA stations nationwide was extracted, and data pertaining to methicillin-resistant Staphylococcus aureus (MRSA) was validated. This resulted in the compilation of a structured data set of all positive MRSA microbiology and surveillance tests from across the nation. This structured database was used to adapt our simulation model to the VA setting and create an accurate base-case simulation scenario. Submodels were also developed to assess the factors that impact organism transmission including: patient flow, hospital room contamination and decontamination, patient-health care worker contact and organism transmission, infection control interventions, hand hygiene, diagnostic testing, and antibiotic prescribing.

Next, we successfully collaborated on the development of a large, longitudinal dataset from PCS that contained the data from most VA facilities nationwide required for our analysis. These data were used to create the electronic classification algorithm to identify VA patients at high risk of MRSA carriage at the time of hospital admission. Included in the data set were over 327,000 hospital admissions and over 270 different prediction variables. Through logistic regression, we were able to generate a model for predicting MRSA carriage with a C-statistic of 0.79, which is favorable in comparison with similar, published work in the field. Using the algorithm in clinical practice, we would be able to reduce MRSA testing by 36% while only missing approximately 10% of MRSA days in the hospital.

The performance of the electronic prediction rule was then incorporated into the computer simulation, and used to compare alternative policies for MRSA control in VA hospitals, including targeted surveillance. This work demonstrated that the rate of MRSA acquisition declined significantly with the use of a targeted surveillance system based on our electronic algorithm. Adding an MRSA decolonization policy on top of this further reduced the mean rate of MRSA acquisition by up to 50%.

IMPACT:
Our work has provided valuable information for the VA and Veteran health on a number of fronts. First, the creation of a structured data set with all positive MRSA microbiology and surveillance tests from across the nation is proving to be an extremely valuable and needed tool for VA researchers beyond our group. The electronic classification algorithm has also provided unique information about the various risk factors for MRSA colonization among VA patients, and has attracted the attention of the VHA National MRSA Prevention Initiative, who would now like to pursue clinical testing of the algorithm in VA facilities as a demonstration project. The simulation work also has the potential to direct future infection control investigations that would normally be infeasible or excessively costly to perform. Overall, the findings from this project have the potential to impact infection control practice and Veteran safety throughout the VA health care system.

PUBLICATIONS:

Journal Articles

  1. Jones M, DuVall SL, Spuhl J, Samore MH, Nielson C, Rubin M. Identification of methicillin-resistant Staphylococcus aureus within the nation's Veterans Affairs medical centers using natural language processing. BMC medical informatics and decision making. 2012 Jul 11; 12(1):34.
  2. LaFleur J, Nelson RE, Sauer BC, Nebeker JR. Overestimation of the effects of adherence on outcomes: a case study in healthy user bias and hypertension. Heart (British Cardiac Society). 2011 Nov 1; 97(22):1862-9.
Book Chapters

  1. Rubin MA. Pharyngitis, sinusitis, otitis, and other upper respiratory tract infections. In: Harrison's Principles of Internal Medicine. 18th ed. New York: McGraw-Hill; 2011. Chapter 31.
Center Products

  1. Rubin MA. High fidelity agent-based model of spread of resistant bacteria within hospitals. 2017 Aug 15.
Conference Presentations

  1. Nelson RE, Jones MM, Leecaster M, Ray W, Huttner B, Huttner A, Khader K, Campo J, Gerding D, Samore MH, Rubin MA. Using an Agent-Based Simulation Model to Conduct an Economic Analysis of Strategies to Control Clostridium difficile Transmission and infection. Poster session presented at: ID Week: A Joint Meeting of IDSA, SHEA, HIVMA, and PIDS; 2013 Oct 2; San Francisco, CA.
  2. Rubin M, Garvin JH, Doebbeling B, Merchant M, Martinello RA, Mutalik P, Goldstein MK, Luther S, Samore M, South B, Gullans S. An Informatics Approach to Methicillin Resistant Staphylococcus Aureus Surveillance in the Department of Veterans Affairs. Poster session presented at: American Medical Informatics Association Annual Symposium; 2011 Oct 25; Washington, DC.


DRA: Health Systems, Infectious Diseases
DRE: Research Infrastructure, Epidemiology, Prevention, Technology Development and Assessment
Keywords: Computational Modeling, Computer Simulations, Decision support, Healthcare Algorithms, Infectious disease, Informatics, Predictive Modeling, Statistical Methods, Surveillance
MeSH Terms: none