Impact of antibiotic-resistant Gram-negative bacilli infections on outcome in hospitalized patients

Abstract

The impact of resistant (vs. nonresistant) Gram-negative infections on mortality remains unclear. We sought to define risk factors for and excess mortality from these infections. Prospective cohort study. Inpatient surgical wards at a university hospital. All patients in the general, transplant, and trauma surgery services diagnosed with Gram-negative rod (GNR) infection. All culture-proven GNR infections (n = 924) from December 1996 to September 2000 were studied. Characteristics and outcomes were compared between GNR infections with and without antibiotic resistance. Univariate and logistic regression analysis identified factors associated with antibiotic-resistant GNR (rGNR) infection and mortality. rGNR infection (n = 203) was associated with increased Acute Physiology and Chronic Health Evaluation (APACHE) II scores (17.8 +/- 0.5), multiple comorbidities, pneumonia and catheter infection, coexistent infection with antibiotic-resistant Gram-positive cocci and fungi, and high mortality (27.1%). Only seven isolates were resistant in vitro to all available antibiotics. Logistic regression demonstrated that rGNR infection was an independent predictor of mortality (odds ratio, 2.23; 95% confidence interval, 1.35-3.67; p =.002). Analysis of rGNR infection with controls matched by organism, age, APACHE II score, and site of infection, however, revealed that antibiotic resistance was not associated with increased mortality (23.6% vs. 29.2%, p =.35). Furthermore, analysis of all Pseudomonas aeruginosa infections demonstrated no significant difference in mortality between resistant and sensitive strains (18.9% vs. 20.0%, p =.85). rGNRs are associated with prolonged hospital stay and increased mortality. Infection with rGNRs independently predicts mortality; however, this may be more closely related to selection of certain bacterial species with a high frequency of resistance rather than actual resistance to antibiotic therapy. Therefore, altering infection-control practices to limit the dissemination of certain bacterial species may be more effective than attempts to control only antibiotic-resistant isolates.