Background: Great variability exists in the occurrence of antibiotic-resistant bacteria in ICUs around the world. The contribution of specific ICU care variables to these geographic variations is unknown.
Methods: ICU patients from two ICUs (in Jerusalem and Toronto) who were admitted for > 48 h and who grew a resistant bacteria in any culture during ICU admission were compared with those without resistant organisms across a range of demographic and ICU care interventions. Significant variables were investigated with logistic regression to identify factors predictive of infection/colonization with a resistant organism.
Results: Resistant organisms were acquired by 82/423 (19%) patients. Patients acquiring a resistant organism had a higher incidence of diabetes mellitus (21/82, 26% vs 52/341, 15%; P = .026), were more frequently admitted from another ICU (17/82, 21% vs 33/341, 10%; P = .005), received more antibiotics in the ICU (19 +/- 17 vs 14 +/- 14 days; P = .005), and had more ventilator (10 +/- 10 vs 7 +/- 8; P = .031) and central line days (10 +/- 8 vs 7 +/- 8; P < .001). These patients had a lower incidence of limitation-of-therapy orders (9/82, 11% vs 78/341, 23%; P = .015). Only the absence of a limitation-of-therapy order (odds ratio, 2.62; 95% CI, 1.21-5.68; P = .014) was independently associated with the acquisition of resistant organisms. Further, among ICU fatalities, 5/45 (11%) patients acquired a resistant organism prior to withdrawal vs 17/44 (39%) nonwithdrawal fatalities (P = .003). Nonwithdrawal fatalities received significantly more third-line antibiotics (7 +/- 14 vs 2 +/- 4; P = .031) despite similar ICU lengths of stay (15 +/- 21 days for nonwithdrawal fatalities vs 10 +/- 11 for withdraw fatalities; P = .210)
Conclusions: End-of-life treatment is independently associated with acquisition of resistant bacteria. Patients dying without withdraw orders receive more antibiotics and develop more resistant organisms. These patients may represent a reservoir of resistant bacteria in the ICU.