In order to characterize the impact and pattern of Gram-negative bacteraemi
a (GNB) at a Swiss University hospital and to assess the effect of multi-re
sistance on mortality, we conducted a 6-y retrospective cohort study using
linear regression and multivariate Cox-proportional hazard analysis. 1766 p
atients had 1835 episodes of GNB; 61% were community-acquired. The incidenc
e of GNB increased linearly (r(2) = 0.90, p = 0.014) from 7.07 episodes to
8.32 episodes per 1000 admissions, but this trend was no longer significant
after adjustment for the number of blood cultures drawn/y. The in-hospital
mortality for patients with GNB decreased from 20% in 1989 to 16% in 1994
(r(2) = 0.94, p = 0.005), The risk ratio for death remained unchanged over
the study period and was 7-fold higher for patients with GNB than for patie
nts without GNB. Factors independently associated with an increased hazard
of death after GNB were: severity of illness as measured by exposure to int
ensive care (hazard ratio \HR\, 1.5); age=66-79 y (HR 1.8); GNB due to Kleb
siella spp, (HR 1.7) or Pseudomonas aeruginosa (HR 1.6); and polymicrobial
infection (HR 1.6). Multi-resistance was not associated with an increased r
isk of death (HR 1.0). Although the crude mortality of GNB decreased, the p
opulation-attributable risk ratio for death remained significant. These dat
a suggest the absence of a major impact of multi-resistant GNB on patient m
ortality.