Automated Surveillance for Catheter-Associated Urinary Tract Infections

Background: Urinary tract infections (UTI) are the most common hospital-acquired infections (HAI), causing increased morbidity and hospital costs among patients. Manual surveillance for catheter-associated urinary tract infections (CAUTI) is labor-intensive and often requires a large proportion of infection preventionist’s time. Automated surveillance of patient medical records and microbiology data regarding urinary catheter usage offers a more efficient method of surveillance.

Objective: To compare electronic CAUTI surveillance and manual chart abstraction

Methods: Patients with a positive urine culture in three inpatient units at Progress West HealthCare Center, a 72-bed community hospital, between August 1, 2008 and July 31, 2009 were evaluated. Patient admission, microbiology, and urinary catheter usage data were collected manually by chart review. Infection prevention personnel used the Centers for Diseases Control National Health Safety Network definitions to determine "true" CAUTI for manually abstracted data. An automated CAUTI surveillance report was generated through the medical record McKesson Horizon Clinicals database (McKesson, San Francisco). For all inpatients, rules were applied to the electronic data to determine the presence of CAUTI. Condition rules included: (1) patient had a urinary catheter; (2) positive culture results > 48 hours after catheter insertion; and (3) colony count ≥ 100,000 colonies. Sensitivity, specificity, and predictive values, were calculated based on the automated results using manual chart abstraction as the reference standard.

Results: During the study period, 1,123 positive urine cultures were reported of which 224 cultures were from inpatients.  Of the 224 positive urine cultures, 44 inpatients were evaluated and ten (23%) CAUTI were confirmed by manual chart review (3.8 CAUTI per 1,000 urinary catheter days). Results included symptomatic UTI (SUTI) and asymptomatic bacteriuria (ASB). Automated surveillance generated results with a sensitivity of 100% and specificity of 98%. Table 1.  Performance of Automated Surveillance for CAUTI

	Manual Surveillance
Automated Report	Positive	Negative	TOTAL
Positive	10	5	15
Negative	0	209	209
TOTAL	10	214	224

Conclusions: Automated surveillance accurately reported all true CAUTI from manual chart abstraction, and 98% of truly non-diseased patients were identified as having no CAUTI. A negative predictive value of 100% indicates that automated surveillance accurately identified and excluded positive cultures not associated with CAUTI. Automated CAUTI surveillance with perfect sensitivity can increase the efficiency of Infection Preventionists and reallocate time spent on infection prevention strategies to reduce HAI. Prevention efforts focused on reducing cultures of ASB patients would improve predictions of true SUTI.