National Automated Surveillance of Hospital-Acquired Bacteremia in Denmark Using a Computer Algorithm

Abstract

BACKGROUNDIn 2015, Denmark launched an automated surveillance system for hospital-acquired infections, the Hospital-Acquired Infections Database (HAIBA).OBJECTIVETo describe the algorithm used in HAIBA, to determine its concordance with point prevalence surveys (PPSs), and to present trends for hospital-acquired bacteremiaSETTINGPrivate and public hospitals in DenmarkMETHODSA hospital-acquired bacteremia case was defined as at least 1 positive blood culture with at least 1 pathogen (bacterium or fungus) taken between 48 hours after admission and 48 hours after discharge, using the Danish Microbiology Database and the Danish National Patient Registry. PPSs performed in 2012 and 2013 were used for comparison.RESULTSNational trends showed an increase in HA bacteremia cases between 2010 and 2014. Incidence was higher for men than women (9.6 vs 5.4 per 10,000 risk days) and was highest for those aged 61–80 years (9.5 per 10,000 risk days). The median daily prevalence was 3.1% (range, 2.1%–4.7%). Regional incidence varied from 6.1 to 8.1 per 10,000 risk days. The microorganisms identified were typical for HA bacteremia. Comparison of HAIBA with PPS showed a sensitivity of 36% and a specificity of 99%. HAIBA was less sensitive for patients in hematology departments and intensive care units. Excluding these departments improved the sensitivity of HAIBA to 44%.CONCLUSIONSAlthough the estimated sensitivity of HAIBA compared with PPS is low, a PPS is not a gold standard. Given the many advantages of automated surveillance, HAIBA allows monitoring of HA bacteremia across the healthcare system, supports prioritizing preventive measures, and holds promise for evaluating interventions.Infect Control Hosp Epidemiol 2017;38:559–566