In a three-year period (January 2004 to December 2006), data from 573 consecutive adult multiple trauma patients admitted to an urban trauma center (Bergmannstrost Trauma Center, Halle/Saale, Germany) were collected retrospectively; 206 patients underwent blood alcohol screening, and 164 tested positive. We evaluated the diagnostic, therapeutic, and outcome characteristics of these patients and compared patients who received blood transfusions with patients who did not.

Inclusion criteria were suspected high-impact trauma, hospital admission on trauma Day One, and a positive BAC. Exclusion criteria were major burns, absence of an emergency physician at the accident scene, or missing data. The study protocol was approved by the hospital's data security manager and the local ethics committee.

All patients were transported to the hospital by physician-staffed ground ambulance or helicopter rescue. Some patients came directly from the scene of the accident (primary admission), and others were transported from smaller hospitals where they had received their initial treatment (secondary admission). All patients had a whole-body multi-slice CT-scan (Toshiba Aquilion 32), in accordance with the center's trauma protocol. Retrospective analysis was performed using computerized medical records, focusing on socio-demographic data and trauma- and treatment-specific variables. Personal patient information was treated in accordance with the Declaration of Helsinki.

'Multiple trauma' was defined as either two or more injuries (life-threatening in combination) or one life-threatening injury 14. An initial Glasgow coma scale (GCS) score was calculated at the scene of the accident by the emergency physician. A diagnosis of 'severe head injury' was made if there was a skull fracture or intracranial hemorrhage on head computed tomography (CT), as interpreted by a radiologist. The ISS was calculated after the initial resuscitation by a senior anesthesiologist (always the same person). Blood alcohol samples were taken in the emergency department. The lowest hemoglobin and prothrombin levels and the highest lactate levels in the 24 hours after admission were recorded. Airway management was classified into the following categories: tracheal intubation at the scene, in the emergency department, during transportation, or not performed. Admission and length-of-stay in the ICU, as well as the need for and duration of artificial ventilation, were noted. The number of packed red blood cell transfusions in the first 24 hours after admission and throughout hospitalization was calculated. For patients who died of their injuries, the time of survival after trauma was recorded.

Group comparisons were analyzed using the χ² test, Student's t-test, or the Mann-Whitney U-test. The alpha level of significance was set at 5%, two-tailed. To evaluate predictors for mortality, a binary logistic regression model was employed for the whole study population. Further exploratory analyses used Spearman rank correlations. All computations were made using SPSS 13.0.

From the cohort of 164 patients with a positive BAC, 19 patients were excluded: nine did not meet the requirement for the definition of multiple trauma, four had major burns, four were admitted after trauma Day One, and two had incomplete data. In the emergency department, 45 of the 145 patients had a BAC of >0.5 g/l and 36 had a BAC of >1.0 g/l. Seventy-six patients received a blood transfusion (transfusion group [TG]), and 69 patients did not (no transfusion group [NTG]). Both groups were comparable with respect to age, and most patients in both groups were male (89% of the whole study population). The characteristics of the two groups and statistical comparisons are summarized in Table 1.

The two groups did not differ in the causes of their trauma (Table 1): in both groups of patients, the major causes were traffic accidents and falls from a height. Four patients (n = 1 in TG vs. n = 3 in NTG) had other causes for their trauma: one patient was buried under a tumbling wall, two were injured in complex workplace accidents, and one was shot.

Most patients (77.2%) in both groups were primary admissions to the trauma center. Similar proportions of patients in both groups underwent helicopter rescue (table 1), but the TG patients were more likely to be secondary admissions (p = .002)

Transfused patients had lower GCS scores (p ≤ .001) and higher ISS (p ≤ .001) and were more likely to have a severe head injury (p ≤ .001). The BAC tended to be higher in TG patients (p = .053). Hemoglobin levels and prothrombin times in the first 24 hours were lower (p ≤ .001), and lactate levels were higher (p ≤ .001) in TG patients. Only 16 females were found in the study population; most of them received blood transfusions (81.3%).

Higher BACs correlated with lower GCS scores, but not with other clinical variables (correlations between BAC and age, laboratory results, ISS, transfusion amount, duration of intensive care, artificial ventilation, and days to decease in non-survivors were low and not significant).

The overall rate of tracheal intubation was 57.9%. Transfused patients had a higher rate of tracheal intubation, especially at the accident scene and in the emergency department. The number of transport-related intubations was similar in both groups. Furthermore, transfusion was associated with a higher admission rate to the ICU, longer ICU stays (p ≤ .001), and longer artificial ventilation times (p ≤ .001).

The average number of transfusions was four in the first 24 hours and 12 throughout hospitalization (Table 2). Twelve patients received massive transfusions (>10 units) in the first 24 hours. Only the laboratory parameters were significantly different between massively transfused patients and patients receiving lesser amounts of blood (lower hemoglobin and prothrombin time, higher lactate; p ≤ .001). BAC, age, ICU admission length, artificial ventilation times, GCS, and ISS were not related to transfusion amount.

Eighteen patients did not survive. Mortality was significantly higher in transfused patients (n = 15 in TG vs. n = 3 in NTG, p ≤ .001). The trauma contexts of deceased patients were traffic accident (n = 8), fall from height (n = 9), and shotgun injury (n = 1). Two patients died after discharge from the ICU. Non-survivors were more likely to have had a severe head injury and to have undergone tracheal intubation and artificial ventilation.

Significant differences between non-survivors and survivors were: ISS, age, intubation rate, lactate level, number of transfusions in the first 24 hours and total number of transfusions, GCS scores, and hemoglobin and prothrombin levels. BAC, ICU admission, length of ICU stay, and hours of artificial ventilation were equal in both groups. There were no differences in gender, admission type, or likelihood of severe head injury.

A binary logistic regression model was used to examine predictor variables for mortality (Table 3). The potential predictors assessed were age, GCS score, ISS, laboratory results, intubation, artificial ventilation, head injury, and transfusion. Since BAC must be considered as a confounder, we included it as an independent variable in the regression model. With a given fit of the regression model, only two parameters predicted mortality, age (p = .006) and ISS (p = .003), and these correctly predicted mortality in 66.7% of cases and survival in 98.9% of cases. If only TG (where most patients died) was used for predictive analysis, age (p = .023) and ISS (p = .006) remained the only significant predictors; they predicted mortality in 61.5% of cases and survival in 98.2% of cases.

Our study has several important limitations. Its retrospective character could have influenced the availability of the data collected. The decision to transfuse packed cells was at the discretion of the attending intensivist and not based on transfusion protocols or scoring systems 133233. Although regression analysis minimizes confounding in cohort studies, only prospectively performed randomization of patients to standard protocols can reduce the effects of subjective assessment and treatment by physicians 34. Additionally, our patients were not a homogeneous group, but rather presented with a variety of trauma etiologies. Also, our data reflect the experience of a single trauma center and are drawn from a relatively small sample size.

There was a large number of patients transferred from other facilities in whom BAC-determination was performed relatively late in the course of their management, and this may have led to unreliable laboratory results (such as false negatives, with BAC having declined to undetectable levels) 35. Finally, the low number of other potential predictive variables (e.g., patient hemodynamics, infusion of vasoactive agents, or the presence of organ failure) limits the general power of the study.

Nevertheless, the present study provides data in an under-researched field. Our findings suggest that morbidity is higher in transfused BAC-positive trauma patients and that transfusion use in BAC-positive patients correlates strongly with poorer laboratory measures and higher expenditure resulting from an increased need for intensive care. We found that ISS and age were robust predictors of overall mortality.