© 2019 SourceMedia. All rights reserved.

Hunting Down Healthcare-Associated Infections with Data Analysis

Healthcare-associated infections have received heightened attention from providers, patients and payers since the Center for Medicare and Medicaid Services added HAIs to the list of “never events” — conditions that can be prevented by reasonable methods, and would no longer be paid for by CMS if they occurred — the agency established in October 2008. 

Since then, 38 states have enacted laws requiring hospitals to report various infections, 29 of which are required to report their results publicly. In July 2011, the Joint Commission issued 13 hospital national patient safety goals, which include four goals targeting infection prevention. 

At the same time, the Office of the Actuary at CMS reported U.S. health care spending on track to increase 5.8 percent annually through 2020 when total spending could reach $4.64 trillion, accounting for 19.8 percent of the 2010 gross domestic product. HAIs drive up the total health care spend by contributing an added cost burden of up to $45 billion annually, according to a 2009 Center for Disease Control report. The CDC estimates that preventing just 20 percent of HAI could save the health care system up to $6.8 billion annually, while prevention of 70 percent could equal a savings of $31.5 billion.

As such, innovation around prevention of costly HAIs such as methicillan-resistant Staphylococcus aureus (MRSA), Clostridium difficile (C. dif) and vancomycin-resistant enterococci (VRE) is trying to harness technology to identify patients potentially at risk for infection in real time, before illness takes hold and additional patients become infected. 

Tapping Data

Ari Robicsek, M.D., is a hospital epidemiologist and Harvard-trained infectious disease physician focused on utilizing technology to track and defeat these bugs. His work at NorthShore University HealthSystem in the Chicagoland area centers on creating cost effective, real-time infection prevention strategies by tapping into the health system’s own custom-made data warehouse.

“Back in 2005, we were one of the first hospitals to institute universal surveillance for MRSA upon admission,” says Robicsek, Northshore’s associate chief medical information officer. “Once this program was fully instituted we saw a 70 percent reduction in MRSA infections.” Results of NorthShore’s universal MRSA surveillance initiative were published in the March 2008 Annals of Internal Medicine.

The initial challenges were achieving buy-in by all nursing units to perform the universal testing, which was a time-eating, cumbersome process, as was the random chart audits required to verify testing was being performed. 

A partial solution was a large-scale data analysis of patients who did or did not receive testing. NorthShore then focused on eliminating the “human factors,” such as workflow bottlenecks and other issues that were getting in the way of testing. That yielded wider acceptance of testing, and the health system currently is at 95 percent compliance with the program, Robicsek says.

Pushing the Protocol

NorthShore now has the ability to isolate appropriate patients in a timely manner. The clinical decision support tool now in place not only reminds providers to perform the screening test, but also initiates the protocol for an isolation cart to arrive at an infected patient’s room, including a requirement that gloves and gown be worn by all staff when caring for the patient.

“The delay in our old process took at least a couple hours,” says Robicsek. “The testing, reporting, results, ordering of the isolation cart was time consuming. The process now allows the lab worker to add the diagnosis to the patient’s chart, order the patient to be isolated and send the isolation cart. This has cut the time involved down to almost nothing.” 

Molly Opela-Greenwood, nurse manager at the Cardiac Care Center at NorthShore Evanston Hospital, appreciates how care at the bedside has been improved upon by access to real-time data concerning her patients.

One of the most effective features of NorthShore’s EHR, according to Opela-Greenwood, is the Banner system, a color-coded electronic alert that appears at the top of the patients’ EHR and stays in the EHR from encounter to encounter until cleared by a physician. For example, any patient who found to be MRSA positive would have an orange banner across the top of their EHR until cleared at their next visit. This allows each service to be acutely aware of previous infections, even if they were not the service that initially made the diagnosis.

“Continuity and consistency across the board are the best things resulting from these programs,” says Opela-Greenwood.

While NorthShore has been successful in identifying the majority of MRSA-positive patients before they pose a problem, testing everyone becomes yet another cost to the health care system. 

Robicsek shares that NorthShore tests between 50,000 and 60,000 patients per year, and only 3 percent to 4 percent are positive. False positives are also frequently encountered.  As a result, the question then became how to reduce the number of patients tested while still capturing similar numbers of MRSA-positive patients. Robicsek had the answer, as well as the support from NorthShore’s President and CEO, Mark Neaman, needed to pursue it. 

Continuous improvements to NorthShore’s custom data warehouse , which includes patient information from the enterprise EHR (from Epic System’s Corp.), billing, admissions, laboratory, microbiology and pharmacy databases, enabled Robicsek and his team to develop more complex prediction rules than used during past efforts at infection control. A research study was designed with the goal of developing prediction rules that could be run by an automated decision support system able to calculate a patient’s risk score and provide feedback within 24 hours of a patient’s admission on whether or not they needed to be tested. Results of the study were published in January in Infection Control and Hospital Epidemiology.

Isolating Variables

“We performed multi-variable modeling to determine which factors would best predict if someone was a MRSA carrier and we were able to isolate 18 variables that were independent predictors of carriers,” says Robicsek. “The result is an equation that predicts the likelihood of a patient being a carrier with pretty good accuracy.”

Robicsek and his team have successfully built this equation into NorthShore’s EHR and integrated it with a clinical decision support tool alerting the provider if the patient is high risk. NorthShore plans to go live with this new approach in the fall which will allow the health system to test 50 percent of patients and still capture 85 percent of those that are MRSA positive.

“We will be able to cut costs in half, with an estimated $500,000 savings,” says Robicsek. “We will cut false positives in half and substantially reduce the burden of MRSA without losing the ability to capture the majority of MRSA infections.”  

For reprint and licensing requests for this article, click here.