|| Checking for direct PDF access through Ovid
This review describes recent advances in the field of automated surveillance of healthcare-associated infections (HAIs), with a focus on data sources and the development of semiautomated or fully automated algorithms.The availability of high-quality data in electronic health records and a well-designed information technology (IT) infrastructure to access these data are indispensable for successful implementation of automated HAI surveillance. Previous studies have demonstrated that reliance on stand-alone administrative data is generally unsuited as sole case-finding strategy. Recent attempts to combine multiple administrative and clinical data sources in algorithms yielded more reliable results. Current surveillance practices are mostly limited to single healthcare facilities, but future linkage of multiple databases in a network may allow interfacility surveillance. Although prior surveillance algorithms were often straightforward decision trees based on structured data, recent studies have used a wide variety of techniques for case-finding, including logistic regression and various machine learning methods. In the future, natural language processing may enable the use of unstructured narrative data.Developments in healthcare IT are rapidly changing the landscape of HAI surveillance. The electronic availability and incorporation of routine care data in surveillance algorithms enhances the reliability, efficiency and standardization of surveillance practices.