Fig. 5

Hazard ratios for PVE based on multivariable Cox regression model including 90 day risk of PVE after any healthcare exposure. Figure shows the multivariable analysis based on any healthcare exposure defined as having an invasive procedure, hemodialysis, or hospitalization for 2 days or longer. n = 23,720. Bonferroni corrected critical p-value for multiple comparisons was 0.0031, and 95% confidence interval was Bonferroni adjusted. Index surgery after 2008 indicates after July 2008 (at the time of change in Australian antibiotic prescribing guidelines). CABG, coronary artery bypass surgery; CI, confidence interval; NA, not applicable; PVE, post cardiac valve surgery endocarditis. Excluding hemodialysis, a recurring exposure event from our definition of healthcare exposure made minimal change to the results: HR for PVE after healthcare exposure within 1 month of index valve surgery was 1.3 (95% CI 0.8–1.8) vs 1.3 (95% CI 0.95–1.9) if hemodialysis was excluded; HR after healthcare exposure 1–3 months after index valve surgery was 3.0 (95% CI 2.0–4.4) vs 2.8 (95% CI 1.9–4.2) if hemodialysis was excluded; HR after healthcare exposure > 3 months beyond index valve surgery was 3.9 (95% CI 3.4–4.4) vs 3.8 (95% CI 3.4–4.3) if hemodialysis was excluded