More than 13 million units of blood are transfused each year. Although transfusion can certainly be lifesaving, numerous studies over the past 20 years have shown significant, dose-dependent increases in morbidity, mortality, and cost with each unit of packed red blood cells (pRBCs) transfused.1 Transfusion is one of the most common interventions in the critically ill population; however the negative effects of transfusion-related infection are well documented in the recent literature.1-7 There is no question that transfusion of blood products can be lifesaving to acutely ill trauma patients, but there is little evidence regarding when transfusions are indicated in asymptomatic anemic patients who are no longer in need of acute resuscitation.
Several studies have analyzed healthy individuals with an isovolemic reduction in hemoglobin (Hgb) level to 5.0 g/dL.8,9 They have found no significant compromise in oxygen delivery to the tissues. Currently, there is a lack of clinical data to suggest adequate RBC transfusion endpoints in trauma surgery.10 Given the lack of evidence to support transfusion triggers for young, healthy, asymptomatic orthopedic trauma patients, we decided to investigate whether a more conservative transfusion strategy might be as safe as a more liberal strategy.
Materials and Methods
After obtaining approval from our institutional review board, we performed a retrospective observational cohort analysis of patients treated at a level I trauma center between September 2006 and February 2009. The trauma registry included all patients who underwent surgery performed by a single orthopedic fellowship–trained trauma surgeon. All patients who had a recorded Hgb level of 9.0 g/dL or less at any time during their admission were included; they were considered no longer volume-depleted after initial resuscitation. Exclusion criteria were age under 18 years or over 50 years; pregnancy; head injury; and preexisting heart, pulmonary, or renal disease.
Initially, 963 patients were identified as orthopedic trauma patients treated by Dr. Mullis within the defined period. After inclusion and exclusion criteria were used to limit this database, the charts of the 109 patients who met the above criteria were reviewed. By chart review or telephone follow-up, 104 patients with 1-year follow-up were identified, and their cases became the basis for our analysis. Demographic information, length of hospital stay, surgeries performed, number of pRBC units transfused, Hgb level prompting transfusion, lowest recorded Hgb level, complications, and Injury Severity Score (ISS) were recorded for each patient. Seventy-two patients (69%) were male, 32 (31%) female. Mean age of the study population was 33 years.
Patients were divided into 2 groups by lowest Hgb level before first transfusion—under 7.0 g/dL and 7.0 g/dL or higher—and then by whether they had been transfused. General guidelines for erythrocyte transfusion on the orthopedic trauma service included patients who were symptomatic at rest (headache, dizziness, or shortness of breath) and asymptomatic patients with Hgb levels under 5.0 g/dL. For patients with varying (lesser) degrees of anemia, transfusion typically depended on clinical symptoms and overall decrease in Hgb level from that recorded on admission.
Patient charts were reviewed for complications extending through a 1-year period after initial discharge from the inpatient service. Patients who had not received follow-up treatment through a known outpatient clinic were contacted by telephone to ascertain outcome. Overall, 5 of the 109 patients were lost at 1-year follow-up, leaving 104 patients with 1-year follow-up (95%). Primary outcome of the study was postoperative complications. Superficial wound infection was defined as cellulitis near the surgical site within 1 year, requiring oral antibiotics; deep wound infection was defined as any related infection within 1 year of injury, requiring intravenous antibiotics or surgical débridement in the operating room. The review for complications included superficial infection, deep infection, urinary tract infection, pneumonia, pulmonary embolism, deep venous thrombosis, acute renal failure or insufficiency, nonunion, delayed union, compartment syndrome, osteomyelitis, nerve palsy, anoxic brain injury, cardiac ischemia or infarct, pancreatitis, and death.
Statistical Methods
The primary focus of this analysis was to determine if patients’ risk of complication at 1-year follow-up was affected by anemia—lowest recorded Hgb level before first transfusion for transfused patients, or lowest Hgb level during hospital stay for nontransfused patients—or whether transfusion itself might be a risk factor for complication. Multiple logistic regression models were used to determine the likelihood each group would have a complication. The dependent variable was complication rate; the explanatory variables included whether the patient was transfused, anemia/Hgb level (under 7 g/dL vs 7 g/dL or higher), and the 2-way interaction. Other possible explanatory variables entered into the model were age, sex, ISS, and whether the patient had had multiple surgeries. As the sample size was small, these variables were entered into the regression model one at a time. Results are presented as odds ratios (ORs) with corresponding 95% confidence intervals (CIs) and P values. The analysis was performed with SAS Version 9.1 (SAS Institute, Cary, North Carolina). Tests were considered statistically significant with P < .05 and marginally significant with P < .10. OR above 1 indicated that the odds of a complication occurring were higher in the exposed group (transfused patients) than in the unexposed group (nontransfused patients).