283 - Epidemiology and Clinical Significance of Persistent Bacteremia in Severely Burned Patients

Background: Recent literature questions the utility of follow-up blood cultures (FUBC), especially for gram-negative bloodstream infections (BSIs).  This has yet to be evaluated in the burn intensive care unit (BICU), where many BSIs are gram-negative.  We evaluated the FUBC frequency, positivity rate, and clinical significance of persistent BSI (p-BSI) in BICU patients.

Methods: Patients ≥ 18 years old admitted to the US Army Institute of Surgical Research for combat-related thermal burns from 1/2003-6/2014 were included.  P-BSI was defined as the same organism isolated from initial and FUBC (within 1-5 days).  Non-p-BSI (np-BSI) included patients without subsequent isolation of the same organism between 1-5 days post-positive blood culture.  Exclusion criteria were initial blood culture with usual skin flora, polymicrobial BSI, fungemia, and death within 24 hours of notification of initial positive blood culture.  Those factors significantly associated with mortality on univariate analysis were evaluated with binomial logistic regression (BLR).

Results: Of 126 patients meeting inclusion criteria with BSI, 53 (42.1%) had p-BSI and 73 (57.9%) had np-BSI (table 1).  50 (67.6%) np-BSI patients had FUBC.  P-BSI and np-BSI patients did not differ in age, gender, or race, but p-BSI and np-BSI patients had median total body surface area (TBSA) burns of 47 (IQR 34-63) and 35.3 (IQR 23.3-56.6), respectively (p=0.021).  P-BSI patients had longer hospitalizations, ICU stays, and intubations (p< 0.01; table 1).  Microbiology did not differ between p-BSI and np-BSI (p=0.517).  Notably, 20 (37.7%) p-BSI patients died compared to 8 (10.8%) np-BSI patients (p< 0.001; table 2).  BLR revealed that p-BSI (p=0.031), TBSA (p< 0.001), ISS (p=0.008), and length of ICU stay (p=0.002) and intubation (p< 0.001) were independently significantly associated with mortality.Table 1: Clinical characteristics of burn patients with and without persistent bacteremia
Table 2: Univariate analysis evaluating associations with mortality in burn patients with bacteremia


Conclusion: P-BSI was common in this burn population.  Severe burns and longer duration of hospitalization, ICU stays, and intubation, but not microbiology were associated with p-BSI.  However, p-BSI (in addition to more traditionally identified risk factors like TBSA and duration of hospitalization, ICU, and ventilator days), was independently associated with increased mortality.  FUBC may serve as an additional prognostic factor in burn patients with BSI.