Case ascertainment of a potential centrally-implemented, automated system for national surveillance of healthcare-associated infections, England, 2016 to 2023

BACKGROUNDMandatory reporting of healthcare-associated infections (HCAI) in England is conducted locally by acute hospital groups and can be a large burden on healthcare staff.AIMWe aimed to determine the case ascertainment of a potential centrally-implemented, automated HCAI surveillance system in England using preexisting data feeds at the UK Health Security Agency.METHODSWe compared monthly case numbers submitted between 1 April 2016 and 31 March 2023 by acute hospital groups (locally-implemented surveillance) to routinely-collected laboratory and hospital encounter records (centrally-implemented surveillance) for all infections under mandatory surveillance in England. Since laboratories can serve multiple hospitals, we compared several methods of assigning laboratory-confirmed cases to hospital groups.RESULTSLocally-implemented vs centrally-implemented surveillance identified: meticillin-resistant <i>Staphylococcus aureus</i> bacteraemias 5,453 vs 5,859 (ratio 1.07), meticillin-susceptible <i>S. aureus</i> bacteraemias 84,680 vs 83,326 (0.98), <i>Escherichia coli</i> bacteraemias 281,100 vs 275,133 (0.98), <i>Klebsiella</i> species bacteraemias 65,877 vs 67,301 (1.02), <i>Pseudomonas aeruginosa</i> bacteraemias 25,862 vs 25,715 (0.99), <i>Clostridioides difficile</i> infections (CDI) 94,054 v 90,942 (0. 97) respectively. Assigning hospital groups by linking laboratory records to hospital encounters produced lower monthly mean absolute difference (MAD) vs locally-implemented surveillance than using laboratory records alone. MAD was 0.65 cases/month for bacteraemias, 2.99 for CDI; differences occurred in both directions. MAD decreased over time for bacteraemias but increased from April 2021 onwards for CDI.CONCLUSIONCentrally-implemented surveillance could be feasible for bacteraemias in England due to comparable case numbers with local surveillance. However, more research is needed around understanding and managing data quality of automated feeds, particularly for CDI.