A comparison of methods to detect urinary tract infections using electronic data

Area of Science:

• Infectious Disease Epidemiology
• Health Informatics
• Public Health Surveillance

Background:

• Electronic medical records (EMRs) offer potential for automated surveillance of healthcare-associated infections (HAIs).
• Traditional HAI surveillance is labor-intensive; computational methods can screen EMR data for actionable insights.
• Public health reporting and quality improvement initiatives increasingly focus on HAIs.

Purpose of the Study:

• To compare the effectiveness of seven computer-based decision rules for identifying urinary tract infections (UTIs) in hospitalized patients.
• To evaluate the utility of different data sources (laboratory, clinical, administrative) within these algorithms.

Main Methods:

• A retrospective analysis of 33,834 hospital admissions was conducted.
• Seven distinct computer-based decision rules for UTI identification were applied.
• Data sources included laboratory results, patient clinical data, and International Statistical Classification of Diseases and Related Health Problems, Ninth Revision (ICD-9) codes.

Main Results:

• Out of 33,834 admissions, 3,870 UTIs were identified by at least one rule.
• ICD-9 codes alone identified 2,614 UTIs, with a sensitivity of 55.6% compared to culture- and symptom-based definitions.
• Laboratory-based definitions identified 2,773 UTIs, reduced to 1,125 when fever was included.

Conclusions:

• The choice of algorithm for UTI identification depends on the specific surveillance objective.
• Electronic surveillance methods show promise for mandatory reporting, process improvement, and economic analyses of HAIs.