Epidemiology of infection in critically ill patients with acute renal failure

Objectives Critically ill patients with infection are at increased risk for developing acute renal failure (ARF), and ARF is associated with an increased risk for infection. Both conditions are associated with prolonged length of stay (LOS) and worse outcome; however, little data exist on the epidemiology of infection in this specific cohort. Therefore, we investigated the occurrence of infection in a cohort of critically ill patients with ARF treated with renal replacement therapy (RRT). In addition, we assessed whether this infection worsened outcome. Design Retrospective cohort study. Setting General intensive care unit (ICU) in an academic tertiary care center comprising a 22-bed surgical ICU, eight-bed cardiac surgery ICU, 14-bed medical ICU, and six-bed burn center. Patients Six hundred forty-seven consecutive critically ill patients with ARF treated with RRT, admitted between 2000 and 2004. Interventions None. Measurements and main results total of 519 (80.2%), 193 (29.8%), 66 (10.2%), and ten (1.5%) patients developed one, two, three, and four episodes of infection, respectively. Of 788 episodes of infection observed, 364 (46.2%) occurred before, 318 (40.3%) during, and 106 (13.4%) after discontinuation of RRT. Pneumonia (54.3%) was most frequent, followed by intra-abdominal (11.9%) and urinary tract infections (9.7%). Infections were caused by Gram-negative organisms in 33.7%, Gram-positive organisms in 21.6%, and yeasts in 9.8%. Patients with infection had higher mortality (p = 0.04) and longer ICU and hospital LOS. They needed more vasoactive therapy and spent more time on mechanical ventilation and RRT (all p < 0.001) than patients without infection. After adjustment for potential confounders, Acute Physiology and Chronic Health Evaluation II score, age, mechanical ventilation, and vasoactive therapy were associated with worse outcome, but infection was not. Conclusions Infection occurred in four fifths of critically ill patients with ARF treated with RRT and was in an unadjusted analysis associated with longer LOS and higher mortality. After correction for other covariates, infection was no longer associated with in-hospital mortality.