Explaining sex differences in risk of bloodstream infections using mediation analysis in the population-based HUNT study in Norway

Previous studies indicate sex differences in incidence and severity of bloodstream infections (BSI). We examined the effect of sex on risk of BSI, BSI mortality, and BSI caused by the most common infecting bacteria. Using causal mediation analyses, we assessed if this effect is mediated by health behaviours (smoking, alcohol consumption), education, cardiovascular risk factors (systolic blood pressure, non-HDL cholesterol, body mass index) and selected comorbidities (cardiovascular disease, chronic kidney disease, diabetes, cancer history and chronic lung disease). This prospective study included 64,040 participants (46.8% men) in the population-based HUNT2 Survey (1995−97) linked with hospital records in incident BSI. During median follow-up of 14.8 years, 1840 (2.9%) participants (51.3% men) experienced a BSI and 396 (0.6%) died (56.6% men). Men had 41% higher risk of first-time BSI (95% confidence interval (CI), 28−54%) than women. Together, health behaviours, education, cardiovascular risk factors and the selected comorbidities mediated 34% of the excess risk of BSI observed in men. The HR of BSI mortality was 1.87 (95%CI, 1.53−2.28), for BSI due to S. aureus 2.09 (1.28−2.54), S. pneumoniae 1.36 (1.05−1.76), and E. coli 0.97 (0.84−1.13) in men vs women. This population-based study shows that men have higher risk of BSI and BSI mortality than women. One-third of this effect was mediated by potential modifiable risk factors for incident BSI