Towards (semi-)automated surveillance of healthcare-associated infections

Abstract

Infections that develop within a healthcare institution – simply called healthcare-associated infections – can often be prevented. However, it is important to first find out how and how often these occur. This monitoring is also referred to as 'surveillance' and is often performed manually: an infection control practitioner examines each patient record to see whether a patient meets the requirements for a healthcare-associated infection. The studies in this dissertation show that this current method of surveillance is very labour-intensive, and that the infection rates are not suitable for making comparisons between hospitals.

An algorithm that uses existing information from hospital information systems can help to reliably find patients with healthcare-associated infections. Not only does this save time, this thesis also showed that more infections are found than when the surveillance is performed manually. These algorithms are also applicable in hospitals other than just an academic centre. However, there are still barriers that hinder large-scale implementation of these algorithms: the group of patients to which the algorithm is applied could not be selected automatically, and the extra information needed for the interpretation of the surveillance results was not always properly recorded in the hospital information system.