Background: Infection with resistant gram-negative bacteria is a growing threat to hospitalised patients.
Aim: To determine factors associated with mortality among infants infected by extended-spectrum beta-lactamase-producing Klebsiella species (Klebs-ESBL) and to assess whether selective empirical use of meropenem (MERO) is associated with high mortality.
Methods: Medical records of neonates admitted from January 2002 to December 2003 who had positive blood and/or cerebrospinal fluid (CSF) culture with Klebs-ESBL were reviewed for clinical, management and outcome information. Univariate and multivariate logistic regression analyses were performed to determine factors associated with mortality among infants with culture-proven Klebs-ESBL.
Results: A hundred patients had positive blood (n=97) and/or CSF cultures (n=9) owing to Klebs-ESBL. Overall mortality rate was 30%. The mortality rates among those who were empirically started on a combination of piperacillin-tazobactam and amikacin (Pip-Taz+Amik) (n=48), meropenem (MERO) (n=40) and in those not started on MERO or Pip-Taz+Amik) (n=12) were 25%, 32% and 42%, respectively. Non-survivors were younger (p=0.01), had cardio-respiratory compromise or required assisted ventilation at presentation (p<0.001), and were not started on antibiotics, MERO or Pip-Taz+Amik (p<0.001). On multivariate analysis, factors associated with mortality were vaginal delivery (OR -7.07, 95% CI 2.14-23.39), a need for assisted ventilation at onset of illness (OR -4.94, 95% CI 1.12-21.86) and not starting empirical MERO or Pip-Taz+Amik (OR -17.01, 95% CI 2.41-120.23).
Conclusion: While empirical use of carbapenems for nosocomial sepsis might be appropriate in areas where Klebs-ESBL is prevalent, their use can be restricted to those with cardio-respiratory compromise or severe sepsis without an increase in mortality.