Association between primary infection source and in-hospital mortality among adults with secondary bloodstream infection: a retrospective cohort study

Background The primary infection source may influence microbiology and prognosis in secondary bloodstream infection (SBSI), but evidence across microbiologically heterogeneous SBSI populations remains limited. We examined source-specific clinical profiles and in-hospital mortality. Methods This single-center retrospective cohort study included adults with SBSI admitted to a 4,500-bed tertiary hospital in China from 2017 to 2024. Primary infection sources were classified as lower respiratory tract, intra-abdominal, urinary tract, or other infections. In-hospital mortality was analyzed using cumulative incidence functions and Fine–Gray regression, with discharge alive treated as a competing event. Cause-specific Cox regression was used for sensitivity analysis. Results Among 759 patients, overall in-hospital mortality was 31.8% (241/759). Mortality was 52.5% for lower respiratory tract, 24.0% for intra-abdominal, 22.2% for other, and 10.0% for urinary tract infections. The lower respiratory tract group also had the highest proportions of hospital-acquired infection (45.7%), ICU admission (44.3%), polymicrobial BSI (22.1%), and multidrug-resistant organism infection (57.9%). Cumulative incidence differed across infection sources (Gray’s test, P < 0.001). In the most extensively adjusted model (n = 704), lower respiratory tract infection was associated with higher mortality than intra-abdominal infection (sHR, 2.37; 95% CI, 1.60–3.50), urinary tract infection (sHR, 4.40; 95% CI, 2.33–8.32), and other infections (sHR, 2.00; 95% CI, 1.12–3.55). Cause-specific estimates were directionally consistent. Conclusions Primary infection source was associated with in-hospital mortality in SBSI. Lower respiratory tract infection identified a high-risk clinical and microbiological phenotype characterized by greater hospital-acquired infection, ICU admission, polymicrobial BSI, and multidrug resistance. Recognition of this profile may support early severity reassessment and review of empirical antimicrobial coverage, although the absolute risks require validation in other settings.