Impacts of population ageing on bloodstream infection epidemiology and outcomes: A machine learning and statistical modelling study

OBJECTIVES: To investigate the effect of population ageing on pathogen distribution and clinical outcomes in bloodstream infections (BSIs). METHODS: We analysed 37,100 adult patients with BSI from two university hospitals in South Korea (2006-2023) and used statistical and machine learning (ML) approaches to assess temporal trends, age-related changes in causative pathogens, and 30-day mortality. RESULTS: The mean age of patients was 64.6 years (standard deviation 15.0), with 55.7% aged ≥65. Advanced age was independently associated with a high 30-day mortality via a shift in pathogen distribution. ML models predicted an increasing proportion of BSIs caused by Escherichia coli, Enterococcus faecalis, coagulase-negative staphylococci and fungi with ageing and a decreasing proportion of those caused by Staphylococcus aureus, streptococci, Pseudomonas aeruginosa and Acinetobacter spp. Fungaemia contributed to the highest adjusted mortality rate. The advantage of E. coli-BSI being associated with low 30-day mortality was diminished in strains not susceptible to third-generation cephalosporins. CONCLUSION: Population ageing is associated with shifts in BSI epidemiology and outcomes. Our findings suggest that tailored antimicrobial stewardship and infection management are necessary to address the burden of BSIs in ageing populations.