Automated surveillance of hospital-onset bacteremia and fungemia: feasibility and epidemiological results from a Dutch multicenter study

Abstract

Objective: Hospital-onset bacteremia and fungemia (HOB) has been suggested as a suitable and automatable surveillance target to include in surveillance programs, however differences in definitions across studies limit interpretation and large-scale implementation. We aimed to apply an automated surveillance system for HOB in multiple hospitals using a consensus definition, and describe HOB rates.

Design and setting: Retrospective cohort study in four Dutch hospitals: 1 tertiary hospital and 3 secondary hospitals.

Patients: All patients admitted for at least one overnight stay between 2017 and 2021 were included, except patients in psychiatry wards.

Methods: Data from the electronic health records and laboratory information system were used to identify HOBs based on the PRAISE consensus definition. HOB rates were calculated at ward and micro-organism-level.

Results: Hospital-wide HOB rates varied from 1.0 to 1.9, and ICU rates varied from of 8.2 to 12.5 episodes per 1000 patient days. The median time between admission and HOB was 8-13 days. HOBs were predominantly caused by Enterobacterales, Enterococci, S. aureus and coagulase-negative staphylococci. Longitudinal HOB surveillance detected differences over time at ward and micro-organism level; for example increased HOB rates were observed during the COVID-19 pandemic. Sensitivity analyses demonstrated the impact of assumptions regarding the collection of confirmatory blood cultures for common commensals.

Conclusions: Applying a fully automated definition for HOB surveillance was feasible in multiple centers with different data infrastructures, and enabled detection of differences over time at ward and micro-organism-level. HOB surveillance may lead to prevention initiatives in the future.

Figure 1.

Flowchart of the algorithm identifying hospital-onset bacteremia and fungemia. Blood cultures were defined based on set-level, ie, 1 or 2 vials. A blood culture is considered positive if a micro-organism was determined. Micro-organism events are defined by either a pathogen in 1 blood culture OR the same common commensal in 2 blood cultures (different sample ID’s) within 2 calendar days of each other. Micro-organism episodes are defined including an episode duration of 14 days. Bacteremia episodes are defined incorporating polymicrobial episodes. A bacteremia is classified as HOB if the start date is 2 or more days after hospital admission. These definitions are based on the PRAISE consensus definition, and the algorithm is described in more detail in Aghdassi et al. BC: blood culture, BC+: positive blood culture, HOB: hospital-onset bacteremia and fungemia, COB: community-onset bacteremia and fungemia. Figure adapted from Aghdassi et al.

Figure 2.

Flowchart from blood cultures to hospital-onset bacteremia. The percentage presented for HOB indicate the percentage of bacteremia episodes that are hospital-onset. For definitions, see Figure 1.

Figure 3.

Hospital-onset bacteremia incidences over time. HOB rates reflected by year and quarter. HOB rate: number of hospital-onset bacteremia episodes per 1000 patient days; reference: mean HOB rate 2 years before the specific timepoint; blood culture rate: number of blood cultures taken per 1000 patient days, reflected at the right y-axis. The band around the HOB rate reflects the 95% confidence interval.

Figure 4.

Micro-organism specific HOB rate. Hospitals were combined in this figure. HOB rate: number of hospital-onset bacteremia’s per 1000 patient days. ICU: intensive care unit.