# Advancing hospital-onset bacteraemia surveillance: a five-year retrospective study following the hospital-wide implementation of an automated surveillance system at a German university hospital

## TL;DR

This study analyzed hospital-onset bacteraemia cases over five years at a German hospital using an automated system, revealing high infection rates and antimicrobial resistance.

## Contribution

The study provides real-life European data on hospital-onset bacteraemia using an automated surveillance system.

## Key findings

- Hospital-onset bacteraemia rates were highest in intensive care units at 6.0 per 1,000 patient days.
- 34.5% of cases were potentially secondary, with respiratory tract as the most common source.
- 13.2% of microorganisms were multidrug-resistant, including 60.5% of Acinetobacter baumannii resistant to carbapenems.

## Abstract

Hospital-onset bacteraemia and fungaemia (HOB) has emerged as a novel surveillance metric in recent years and a prime target for automation of surveillance of healthcare-associated infections. However, real-life HOB data from European institutions remain scarce. This study explores the epidemiology of HOB at a German university hospital and describes characteristics of HOB cases.

A retrospective single-centre study was conducted by applying an extended version of the Providing a Roadmap for Infection Surveillance in Europe (PRAISE) automated HOB algorithm to data from the electronic health records of all in-hospital patients admitted to Charité university hospital between 2018 and 2022. HOB rates per 1,000 patient days were calculated for different groups of wards. Furthermore, the distribution of microorganisms, share of antimicrobial resistance, and source of possible secondary HOB (defined as HOB-causing pathogens detected in relevant clinical samples other than blood) were analysed. Additionally, patient characteristics and outcomes were investigated.

A total of 3,648,254 patient days and 7,256 HOB with 8,357 microorganisms were included. The pooled HOB rate was 6.0 per 1,000 patient days in intensive care units, and between 0.9 and 2.0 in the various groups of non-intensive care units. Around 34.5% (n = 2,505) of HOB were deemed potentially secondary, with respiratory tract (37.6%, n = 943) being the most common source. A total of 1,106 of 8,357 (13.2%) microorganisms were classified as multidrug-resistant, including 60.5% (23 of 38) of Acinetobacter baumannii with resistance to carbapenems. Case fatality within 14 days of HOB onset was 16.2% (990 of 6,093 patients).

Analysis of electronic health record data provides important insights into the epidemiology and characteristics of HOB cases. Substantial rates of antimicrobial resistance and case fatality underscore the relevance of HOB as an IPC metric. Results from this study may inform refinement of algorithms for automated HOB surveillance.

The online version contains supplementary material available at 10.1186/s13756-026-01708-9.