None of those receiving empirical b-lactam/blactamase inhibitor or cephamycin therapy died. In a subgroup analysis, the patients with rapidly fatal underlying disease (11.8% vs. 30%, p Z 0.022), solid tumor (17.6% vs.35.0%, p Z 0.054), and E. coli infections (67.6 % vs. 47.3 %,p Z 0.047) were more likely to receive empirical inappropriate therapy