Home Analysis State-of-the-art data trip over the bottom line

State-of-the-art data trip over the bottom line

November 2008

Published in the November 2008 issue of Today’s Hospitalist

Across the country, physicians are hotly debating what kind of surveillance efforts are needed to identify asymptomatic methicillin-resistant Staphylococcus aureus (MRSA) carriers in the hospital. Now a research team based out of Intermountain Health Care in Salt Lake City has figured out a way to help bring that surveillance crisis under better control.

The team included R. Scott Evans, PhD, MS, a senior medical informaticist at Intermountain. Researchers harnessed the tremendous power of the health system’s unique electronic medical record (EMR), which captures data on 60% of the health care delivered in Utah and southeastern Idaho. The team used data-mining techniques to identify which patients among admissions should be screened for MRSA.

While the results validated the use of information technology to help target screening efforts, the news wasn’t all good. That’s because researchers also found that bottom-line realities may limit the technology’s use, even when it’s available to physicians and hospitals.

Mining data
The star of the study, which was published in the July/August 2008 issue of the Journal of the American Medical Informatics Association, was Intermountain’s EMR, which covers more than 20 hospitals, 70 health centers, five dialysis centers and a home health agency. The EMR “which has a greater reach than any other hospital- based system in the U.S. “offers this tantalizing snapshot of just how fluid health care delivery really is: Patients being discharged from LDS Hospital, which until last November was the network’s flagship hospital in Salt Lake City, have been treated at more than 60 Intermountain facilities over the course of five years. Researchers using the system had access to all those data.

Conducted in February and March 2007, the study was limited to LDS Hospital. Every night, data-mining software combed through information on new admissions to find patients at high risk of carrying MRSA. The software zeroed in on several risk factors, including patients who had contracted MRSA within the past five years; patients who were in the hospital for more than 10 days; patients receiving antibiotics; patients on hemodialysis; and individuals 75 or older who had been in an acute care hospital in the previous six months.

The system first sent out an alert to staff members to screen specific patients who were found to be at high risk. The system was then used to immediately flag positive results, giving infection-control nurses the opportunity to decide whether or not to isolate those patients.

According to findings, 20% of high-risk patients tested positive, compared to 4.1% of low-risk patients, a more than five-fold increase.

“The key is who you test,” Dr. Evans points out. “We have been able to identify 10% who are high-risk, some 20% of whom are carriers.” According to the CDC, 19,000 patients died of MRSA in the U.S. in 2005 and 95,000 were infected.

High yield indicators
The high-risk criterion with the highest predictive value was previous MRSA colonization or infection, which accounted for 61% of the high-risk patients identified. That was followed by previous antibiotic use plus hospitalization of more than 10 days, which applied to 33% of patients flagged by the surveillance system as high risk. The criterion with the lowest positive predictive value was hemodialysis, which applied to only 6.5% of asymptomatic carriers identified.

While the study demonstrated the feasibility of using technology for surveillance, Dr. Evans points out that “other issues” became apparent during the study. For instance, some members of the Intermountain community worried that identifying high-risk patients might violate patient privacy. There were also concerns about who would pay for the laboratory test to confirm carriage.

Grants solved the financial issue for the study. As for privacy, “we pointed out the high possibility that these asymptomatic patients could re-infect themselves with a line or other infection,” says Dr. Evans. “We made the case that identification was good preventive care for those patients, not just a strategy to head off transmitting the infection to others.”

Logistical problems
Researchers also uncovered a host of implementation issues. On newly admitted patients, for instance, nurses were so busy attending to patients’ other needs that they would acknowledge an alert to get a nasal swab on a high-risk patient, but then forget to collect and order the test. Nurses ended up requesting that such alerts be delayed for 45 minutes after admission.

The research team also realized that it had to contend with another unexpected factor: “alert fatigue.” Physicians and nurses at LDS had already been using flashing computer screens to flag ventilator disconnects and programming problems with infusion pumps.

When the research team used that same alert system for MRSA alerts, it found that some staff wanted to preserve flashing screens for medical emergencies. The bottom line was that nurses didn’t always consider MRSA a high-risk problem.

“We plan to test alternate deployment methods in future versions,” the study authors wrote, “to avoid the potential for alert fatigue.”

Moving forward
The study laid the groundwork for future research in how technology can target MRSA screening. It also showed, Dr. Evans points out, that a hospital does not have to test everyone to get a result worth pursuing.

Almost two years after the study took place, LDS continues to use the data-mining software, but it only sends out alerts to its ICU. Doctors in the ICU asked that the surveillance program continue, so those physicians now receive both flashing-screen alerts and e-mails when high-risk patients are identified.

The issue of whether to make the surveillance standard throughout the hospital, let alone the rest of the Intermountain system, is still being discussed. Despite the system’s unique capacity, Dr. Evans says that using the technology raises bottom-line questions that will sound familiar to hospitalists who work in facilities without such state-of-the-art resources.

“During the study period, a grant paid for the nasal swabs,” he says. “So the question is: How are we going to do follow-up, and where’s the return on investment? Everybody thinks it’s a great thing to do, but who’s going to pay for it?”

Judi Hasson is a freelance writer based in McLean, Va.