ABSTRACT & COMMENTARY

Value of Electronic Surveillance for Hospital CAUTIs

By Joseph F. John, Jr., MD, FACP, FIDSA, FSHEA

Associate Chief of Staff for Education, Ralph H. Johnson Veterans Administration Medical Center; Professor of Medicine, Medical University of South Carolina, Charleston

Dr. John reports no financial relationships relevant to this field of study.

This article originally appeared in the July 2014 issue of Infectious Disease Alert. It was edited by Stan Deresinski, MD, FACP, FIDSA, and peer reviewed by Timothy Jenkins, MD. Dr. Deresinski is Clinical Professor of Medicine, Stanford University, Associate Chief of Infectious Diseases, Santa Clara Valley Medical Center, and Dr. Jenkins is Assistant Professor of Medicine, University of Colorado, Denver Health Medical Center. Dr. Deresinski does research for the National Institutes of Health, and is an advisory board member and consultant for Merck, and Dr. Jenkins reports no financial relationships relevant to this field of study.

SYNOPSIS: Compared to manual surveillance methods, an electronic surveillance tool for catheter-associated urinary tract infections had a high negative predictive value but a low positive predictive value.

SOURCE: Wald HL, et al. Accuracy of electronic surveillance of catheter-associated urinary tract infection at an Academic Medical Center. Infect Control Hosp Epidemiol 2014;35:685-91.

A group from the University Of Colorado School of Medicine constructed this study to determine if electronic surveillance for catheter-associated urinary tract infections (CAUTIs) was as good or better than standard surveillance. They identified 1695 patients from 2009 and 2010 that met inclusion criteria and that were used for this analysis. They developed an algorithm designed to try to detect UTI from the electronic health record and other sources of administrative data for these 1695 patients. The patients were included if they had a "high clinical suspicion" of having a CAUTI. The 425-bed hospital was in a urban setting. Patients were adults 18 years of age or older. Manual surveillance was the comparator arm. The average age was 57 years, and there was a male to female split of 49% to 42% with the remainder unknown. Of the 1695 patients included in this analysis, the electronic algorithm identified 64 cases thought to likely be CAUTI (15 were actually true positive urinary tract infections); in contrast, only 19 were identified through manual surveillance to have CAUTI. Electronic surveillance had a high negative predictive value (NPV) but a low positive predictive value (PPV = 23%). There was a 97% agreement between the electronic algorithm and the manual method. On the basis of these predictive values, the authors felt that electronic surveillance would be a good screening tool. The authors suggest that the test characteristics of the electronic algorithm could be improved in order to improve data pulls.

COMMENTARY

The best thing to say about this study is that, while creative, electronic surveillance could be used in its present form primarily for screening to eliminate negative cases, i.e., its high NPV. This conclusion is somewhat disappointing, but electronic surveillance is in its infancy so that the test characteristics when improved may raise the PPV and the tool could be a stand along.

In the meantime, manual surveillance has a wisdom that electronic surveillance cannot approach in documenting true infections. That does not mean that we should not try to continue to use innovative software to help us in this era of mega-data. This article used a Structured Query Language code in Microsoft Access to apply an algorithm that ends in either a CAUTI or an asymptomatic catheter-associated infection. To result in a CAUTI, the patient needs to have symptoms, and that is the challenging rub for the software to figure out. If there are no symptoms, but a positive blood culture, the diagnosis is considered at least a level of CAUTI. If there are no symptoms and the blood culture is negative, the urine culture positive for less than 2 organisms at a count of 100,000/cc, then there is not a CAUTI, but a CAASB, a catheter-associated asymptomatic bacteriuriaa. While all this process through the algorithm sounds complex — and it is — the use to Infection Control will be a final software that should be easy to apply.

Keep eyes peeled for use of algorithms in software that detects common hospital-acquired infections. For the time being, let us hope this electronic detection of CAUTI can be refined and demonstrate more sensitivity.