Multivariate logistic regression analyses for endpoints 30-day mortality, severe sepsis or septic shock and disseminated disease.

§Three patients were lost to follow-up.&Mandatory variable.#Due to an extremely high correlation of blaI and blaR (contingency coefficient 1.0 of blaR with blaZ), only blaZ was included in the multivariate logistic regression model (endpoint 30-day mortality).§Due to a high correlation of sed/sej/ser and aadD (contingency coefficient 0.71), only sed/sej/ser was included in the multivariate logistic regression model (endpoint severe sepsis/septic shock). If aadD is included instead, results are OR 4.23 (95% CI 0.91–19.61, p = 0.07) for aadD and OR 1.80 (95% CI 1.04–3.10, p = 0.03) for fosB.†Due to a contingency coefficient of 1.0 of sed, sej, ser, only sed was included in the multivariable logistic regression model (endpoint disseminated disease).‡If 22 SAB cases with initiation of adequate antibiotic therapy ≥3 days after onset of SAB were excluded from the analysis CC15 exhibited a marginally significant effect with OR 2.81 (95% CI 1.0–7.91, P = 0.05) whereas mecA lost its significant impact (OR 3.35, 95% CI 0.79–14.21, p = 0.10). For the other endpoints no significant changes were observed.