Infection: A Major Contributor to Anemia of Critical illness

Authors:
Conor Dillon, Elizabeth Wang, Harshank Patel, Hassan Nadeem, Monica Patten, Robert Sawyer, Sara Hennessy, Tjasa Hranjec

Body of Abstract:
Background: Critically ill patients often receive transfusions beyond the initial resuscitation (>48 hours), not always associated with hemorrhage or perioperative blood loss. We hypothesized that injured, infected patients would receive fewer products due to their hypercoagulable state.

Methods: A single center, chart review identified all adult, trauma patients, admitted to our trauma Intensive Care Unit (ICU) with hospital length of stay (LOS) ≥7 days over a 12-month period. Infections were divided into early (≤72 hours post-trauma) and late (>72 hours post trauma). Univariate analysis was used to evaluate an association between anemia (hemoglobin ≤ 7.9) outside of the initial resuscitation window (>48 hours), infection, and transfusion. Multivariate analysis identified risk factors for transfusion.

Results: A total of 145 patients were included: mean age 50±1.7 years, mostly male (71%) with mean LOS 14.7±0.9 days. Among 79 (54.5%) injured patients requiring no transfusion during hospitalization, 9 (11.4%) were treated for an infection (Figure 1). Comparatively, 66 patients were transfused beyond 48 hours, with 52 patients (78.8%) infected, p<0.0001. Pneumonia was most frequently diagnosed (70%), followed by intra-abdominal infection (17.3%). Infected patients received a total of 271 RBC transfusions, compared to 28 in non-infected patients, p<0.0001. Patients with early infections received less red blood cells (RBC) compared to late infections (1.3±0.6 vs 5.1±1.0, p=0.0024). Early diagnosis and treatment of infections was associated with less RBC (3.1±0.9 vs 6.0±1.5, p=0.0824). Most episodes of anemia had no identifiable bleeding source, however, 77.5% of infections were diagnosed within 72 hours of the lowest, documented hemoglobin. Nearly one-third of the patients experienced severe bleeding (GI hemorrhage, retroperitoneal hematoma, hemoptysis) and required life-saving interventions. Most (93%) patients with severe bleeding were found to have late or persistent infections that despite treatment and clinical improvement, were not abolished. When adjusted for age and gender, perioperative bleeding (OR 13.8, CI 4.1-46) and infection (OR 37.7, CI 12.5-113.4) were found to be the most significant predictors of transfusion beyond the resuscitation period (c-statistic 0.908). Of note, only patients with late infections developed documented venous thrombosis (n=6). Conclusions: A strong association exists between infection and significant hemorrhage in the ICU. An acute decrease in hemoglobin (anemia of critical illness) without obvious cause may be considered as a trigger for infectious work-up in patients whose clinical signs, symptoms and laboratory/radiologic findings are unclear. Prompt diagnosis and treatment of infection may decrease the risk of severe hemorrhage and volume of transfusions. Figure 1. Venn diagram of patients based on their infection, transfusion or perioperative status