Computer aided decision making – ICU surveillance in Wales

Background: In 2005 the National Leadership and Innovation Agency for Healthcare (NLIAH) launched care bundles for the insertion and maintenance of central venous catheters (CVC) and prevention of ventilator associated pneumonia (VAP) in Welsh Intensive Care Units (ICU). Mandatory surveillance of CVC-related infections in Welsh ICUs began in 2007, VAPs in 2008, as a collaboration between the Welsh Healthcare Associated Infection Programme (WHAIP) and NLIAH, providing useful infection data for ICU and infection control staff, serving as outcome measures for care bundles.

For surveillance benchmarking, Hospital in Europe Link for Infection Control through Surveillance (HELICS) definitions were used.  Due to standard practices in Welsh ICUs, initial results showed few matches to HELICS definitions limiting the benefit of surveillance to hospital staff, as many of the locally deemed infections were excluded. Problems categorising clinical diagnosis of CVC infections to the complex HELICS definitions also existed.

Objective: To develop an ICU surveillance system for collecting CVC infection and VAP data that conforms to HELICS definitions and local infection definitions. To ensure accuracy and consistency in validating and categorising data prior to analysis and generate timely useful infection reports to provide outcome measures for the care bundles, leading to improvements in patient care.

Methods: Data is captured via paper forms, completed by ICU staff, recording microbiological and clinical results.  These are returned to WHAIP for scanning using an optical mark reader to produce data files and are loaded into a bespoke SQL Server 2005 database with a Windows vb.net interface.  A series of database rules have been developed for data validation, infection identification (i.e. no infection, HELICS infection or local infection) and categorisation (e.g. CRI1, PN1). The database generates infection reports developed in consultation with ICU staff and allows for regular feedback to ICUs.

Results: The system allows error reports to be run each time data is loaded. Many errors (e.g. date discrepancies) can be rectified in-house or with assistance from hospital surveillance coordinators to provide a robust dataset. Automated reports of HELICS and non-HELICS defined CVC infections and VAPs are produced bi-monthly. HELICS defined data is used for all Wales reporting and fed into the HELICS database for ICU infection surveillance.  Locally defined infection rates are used for monitoring local practice.

Conclusions: The system ensures rigorous data validation before regular infection rate reports are fed back to hospitals. The use of built-in rules to define and categorise infections ensure consistency of diagnosis between practitioners. Added value is provided by the facility to diagnose and report local infections allowing practitioners to monitor infections that would otherwise be missed by surveillance of HELICS defined infections.