Automated surveillance for healthcare-associated infections: opportunities for improvement

Abstract

Surveillance of healthcare-associated infections is a cornerstone of infection prevention programs, and reporting of infection rates is increasingly required. Traditionally, surveillance is based on manual medical records review; however, this is very labor intensive and vulnerable to misclassification. Existing electronic surveillance systems based on classification algorithms using microbiology results, antibiotic use data, and/or discharge codes have increased the efficiency and completeness of surveillance by preselecting high-risk patients for manual review. However, shifting to electronic surveillance using multivariable prediction models based on available clinical patient data will allow for even more efficient detection of infection. With ongoing developments in healthcare information technology, implementation of the latter surveillance systems will become increasingly feasible. As with current predominantly manual methods, several challenges remain, such as completeness of postdischarge surveillance and adequate adjustment for underlying patient characteristics, especially for comparison of healthcare-associated infection rates across institutions.

Keywords: electronic; healthcare-associated infection; methodology; prediction; surveillance.