Femoral blood gas analysis, another tool to assess hemorrhage severity following trauma: an exploratory prospective study

Abstract

Background

Veno-arterial carbon dioxide tension difference (ΔPCO₂) and mixed venous oxygen saturation (SvO₂) have been shown to be markers of the adequacy between cardiac output and metabolic needs in critical care patients. However, they have hardly been assessed in trauma patients. We hypothesized that femoral ΔPCO₂ (ΔPCO_{2 fem}) and SvO₂ (SvO_{2 fem}) could predict the need for red blood cell (RBC) transfusion following severe trauma.

Methods

We conducted a prospective and observational study in a French level I trauma center. Patients admitted to the trauma room following severe trauma with an Injury Severity Score (ISS) > 15, who had arterial and venous femoral catheters inserted were included. ΔPCO_{2 fem,} SvO_{2 fem} and arterial blood lactate were measured over the first 24 h of admission. Their abilities to predict the transfusion of at least one pack of RBC (pRBC_H6) or hemostatic procedure during the first six hours of admission were assessed using receiver operating characteristics curve.

Results

59 trauma patients were included in the study. Median ISS was 26 (22–32). 28 patients (47%) received at least one pRBC_H6 and 21 patients (35,6%) had a hemostatic procedure performed during the first six hours of admission. At admission, ΔPCO_{2 fem} was 9.1 ± 6.0 mmHg, SvO_{2 fem} 61.5 ± 21.6% and blood lactate was 2.7 ± 1.9 mmol/l. ΔPCO_{2 fem} was significantly higher (11.6 ± 7.1 mmHg vs. 6.8 ± 3.7 mmHg, P = 0.003) and SvO_{2 fem} was significantly lower (50 ± 23 mmHg vs. 71.8 ± 14.1 mmHg, P < 0.001) in patients who were transfused than in those who were not transfused. Best thresholds to predict pRBC_H6 were 8.1 mmHg for ΔPCO_{2 fem} and 63% for SvO_{2 fem}. Best thresholds to predict the need for a hemostatic procedure were 5.9 mmHg for ΔPCO_{2 fem} and 63% for SvO_{2 fem}. Blood lactate was not predictive of pRBC_H6 or the need for a hemostatic procedure.

Conclusion

In severe trauma patients, ΔPCO_{2 fem} and SvO_{2 fem} at admission were predictive for the need of RBC transfusion and hemostatic procedures during the first six hours of management while admission lactate was not. ΔPCO_{2 fem} and SvO_{2 fem} appear thus to be more sensitive to blood loss than blood lactate in trauma patients, which might be of importance to early assess the adequation of tissue blood flow with metabolic needs.

Background

Severe trauma remains the leading cause of death before the age of 50 [1]. Approximately 40% of trauma deaths involve hemorrhagic shock and are possibly preventable through damage control resuscitation and early bleeding control [2, 3]. Hemorrhage-induced hypovolemia leads to a drop in cardiac output (CO) responsible for tissue hypoperfusion and impaired oxygen delivery to the organs. When the adaptive mechanisms to hypovolemia are overwhelmed and can no longer compensate for the decrease in oxygen delivery, alterations of cellular homeostasis may ultimately lead to refractory shock and multiorgan dysfunction [4]. It is therefore necessary to identify patients with severe blood loss as early as possible in order to quickly start appropriate resuscitation.

At the initial phase of severe trauma management, CO monitoring is not readily available. Several biomarkers have been proposed to assess the adequacy between oxygen delivery and demand in these conditions. Thus, admission lactate has been shown to be predictive of severe hemorrhage associated with massive transfusion [5]. However, during hemorrhage, blood lactate increases only beyond a critical threshold of oxygen delivery when cell anaerobic metabolism is triggered to maintain ATP production. Lactate production therefore only rises when blood loss gets significant [6]. Other metabolic parameters have been proposed to assess the adequacy between oxygen demand and supply in different critical conditions such as sepsis and major surgery. These parameters include the venous to arterial carbon dioxide gradient (ΔPCO₂) [7, 8] and venous oxygen saturation (SvO₂) [9, 10] but none of them has been studied in severe trauma patients. A drop in CO is responsible for a decrease in tissue blood flow leading to venous carbon dioxide accumulation and an increase in the fraction of extracted oxygen that cause an increase in ΔPCO₂ [11] and a decrease in SvO₂ respectively [12].

Femoral arterial and central venous catheters are commonly inserted during initial management of severe trauma patients in the trauma bay [13]. They make venous and arterial blood gas available to appraise the venous to arterial difference in CO₂ (ΔPCO_{2 fem}) as well as femoral venous oxygen saturation (SvO_{2 fem}).

The aim of the present study was to assess the ability of ΔPCO_{2 fem}, SvO_{2 fem} and arterial blood lactate to predict transfusion of red blood cells (RBC) over the first hours following severe trauma.

Methods

Study design

This observational, prospective and single-centre study was conducted in the surgical Intensive Care Unit of Bicêtre Hospital, an academic level-1 trauma center, from September 2015 to May 2020. This hospital provides 24-h availability of all essential trauma specialties, staff and equipment for trauma patient care. This study was approved by the ethics committee (“Comité de Protection des Personnes”) of the hospital (SC13-014 RCB: 2013-A01171-44) with waiver of participant consent. Patients or relatives were informed and we obtained confirmation of non-objection to data use.

Study population

Patients over 18, admitted from the trauma scene, with an Injury Severity Score (ISS) greater than 15, and for whom initial management in the trauma room required the insertion of an arterial and a venous catheter at the femoral site, were included.

Pregnant women, patients with NYHA III or IV heart failure or chronic kidney disease (clearance < 30 ml/min) and patients with lower limb amputation or severe crush injury on the same side the femoral catheters were inserted were not included in the study.

Clinical management

Following the emergency call, a physician-staffed mobile intensive care unit was sent to the trauma scene. After clinical assessment and medical care adjusted to trauma severity, patients were transported to the study center. Upon arrival in the trauma room, the insertion of femoral catheters was decided based on the patient’s history, clinical examination, and FAST ultrasound results, at the discretion of the clinician in charge as detailed elsewhere [13]. Briefly, the catheter lines were prepared in a sterile manner before patients’ arrival. Arterial (5 Fr, 11 cm) and venous femoral catheters (7 Fr, 20 cm) were simultaneously inserted by a trained resident under supervision of the trauma leader (consultant). The decision to transfuse was made by the physician in charge, based on his evaluation of the patient’s clinical situation using standard parameters (including clinical signs of shock, positive FAST ultrasound or obvious external hemorrhage and bed-side measurement of hemoglobin level) and blinded to the ΔPCO_{2 fem}, SvO_{2 fem} results.

Data collection

For each patient, demographics, past medical history, trauma characteristics, pre-hospital management data such as time to hospital admission, initial Glasgow Coma Scale (GCS) score, minimum systolic blood pressure, peripheral oxygen saturation, maximum heart rate, need for mechanical ventilation, volume of fluid resuscitation, catecholamines use were collected. We also reported the following scores: Injury Severity Score (ISS) [14], details of injuries for each organ (Abbreviated Injury Scale 2015, AIS) [15] and simplified acute physiology score (SAPS II) [16]. Severe organ injury was defined by an AIS for the correspondent organ over 2. On hospital admission, mean arterial blood pressure, heart rate, peripheral oxygen saturation, laboratory parameters (hemoglobin, prothrombin time, fibrinogen, myoglobin) were reported. Fluid administration, dose of catecholamines, transfusion therapy (type of product and amount) and hemostatic procedures were collected during the first 6 h of admission. The following outcome variables were also reported: length of mechanical ventilation, length of intensive care unit (ICU) stay and hospital mortality.

Blood gas measurement

Venous and arterial blood gas were sampled from femoral catheters at insertion and regularly during the first 24 h, between 4 and 8 h from ICU admission (H6), 10 and 14 h from ICU admission (H12) 20 and 26 h from ICU admission (H24). pH, arterial (PaCO_2fem) and femoral venous carbon dioxide partial pressure (PvCO_2fem), arterial oxygen partial pressure (PaO₂), venous oxygen saturation (SvO_2fem), arterial oxygen saturation (SaO₂)_, hemoglobin concentration and blood lactate were measured on both samples using a point-of-care blood gases analyzer (ABL 800, Radiometer). The femoral venous to arterial difference in carbon dioxide pressure (ΔPCO_{2 fem}) was calculated as femoral PvCO₂–PaCO₂.

Outcome

The primary outcome of interest was the transfusion of at least one pack of RBC during the first six hours of admission (pRBC_H6). The secondary outcome was the need for an emergency hemostatic procedure. It was defined as performing angioembolization or hemostatic surgery within the first six hours of management from admission. Orthopedic surgery for fracture osteosynthesis was not considered as a hemostatic procedure.

There is no data regarding ΔPCO_{2 fem} and SvO_2fem in trauma patients. Since trauma patients present a wide variety of traumatic injuries, we considered that a sample of 60 patients would be representative for this exploratory study.

Statistical analysis

Qualitative variables were expressed as counts (proportions). Quantitative variables were expressed as mean (SD) or median (25th–75th interquartile range) according to their distributions. Correlations between blood lactate, ΔPCO_{2 fem} and SvO_{2 fem} at admission were analyzed by Pearson correlation.

In the overall population, volumes of pRBC_H6 transfused were reported according to quartile distribution of blood lactate, ΔPCO_{2 fem} and SvO_{2 fem}. Then, patients were separated into two groups according to whether they had been transfused with pRBC_H6 or not. Their characteristics were compared by a Student t-test or a Mann–Whitney test for quantitative variables (according to data distribution) and by a Chi-square test for qualitative variables. Evolution of lactate, ΔPCO_{2 fem} and SvO_{2 fem} during the 24 first hours of admission were compared over time and between the two groups with a mixed-effect model. Time, group (transfusion, no transfusion) and interaction (time x group) were set as fixed effect and patient as random effect.

To evaluate the ability of admission lactate, ΔPCO_{2 fem} and SvO_{2 fem} to predict the need for pRBC_H6 transfusion or an emergency hemostatic procedure, receiver operating characteristic curves (ROCs) were built and their areas under the curve (AUC) were calculated. The best threshold for the prediction of pRBC_H6 transfusion or to predict the need for an emergency hemostatic procedure was defined as the value maximizing the Youden index (Y = Sensitivity + Specificity-1). Sensitivity, specificity, positive and negative predictive values (PPV and NPV), were reported for each variable. Two-sided level of significance was fixed at P = 0.05. Data were analyzed using Prism (GraphPad Software, San Diego, California, USA).

Results

Population characteristics

Seventy patients were enrolled in the study. Eleven patients were secondarily excluded because their ISS was less than 15, leaving 59 patients in the final analysis (Additional file 1). Patients were 46 ± 20 years old and one third were women (17(29%)). Patients experienced predominantly blunt trauma (87%) with a median ISS of 26 (22–32). Median ICU length of stay was 10 (5–15.5) days and the overall hospital mortality was 26%. Two patients died in the first 24 h. The main characteristics of the population are presented in Table 1.

Table 1 General characteristics in the overall population and in both transfused and non-transfused patients

Incidence of transfusion

Over the first six hours of admission, 28 patients (47%) received at least one pack of RBC. The characteristics of transfused and non-transfused patients are shown in Table 1. Transfused and non-transfused patients were similar with regard to age, gender, SAPSII, type of trauma and on-scene hemodynamic parameters. Transfused patients presented more severe abdominal (43% vs. 10%, P = 0.006) and pelvic trauma (54% vs. 29%, P = 0.020), but they had significantly less traumatic head injuries (28.6% vs. 64.5%, P = 0.006) than non-transfused patients.

Mean blood pressure at hospital admission was significantly lower in the transfused group (P = 0.020). The given volume of crystalloids (P < 0.001) and colloids (P = 0.020) over the first six hours of management were significantly greater in patients who got transfused. Nineteen patients (67.8%) required an emergency hemostatic procedure in the transfused group and only 2 (6.5%) in the non-transfused group (P < 0.001). Mortality was not different between the two groups (P = 0.060) (Table 1).

Blood gas analysis

In the whole cohort, blood lactate was 2.7 ± 1.9 mmol/l, ΔPCO_{2 fem} was 9.1 ± 6.0 mmHg and SvO_{2 fem} was 61.5 ± 21.6% on admission. As shown in Fig. 1, volume of pRBC_H6 was associated with a greater ΔPCO_{2 fem} or blood lactate level and a lower SvO_{2 fem}. Volume of pRBC_H6 transfused was larger when ΔPCO_{2 fem} was greater than 13 mmHg, SvO_{2 fem} was less than 50% or blood lactate was greater than 3.5 mmol/l (Fig. 1).

Fig. 1

Volumes of red blood cells transfused during the first six hours of admission. Volumes are shown according to quartile distribution of A ΔPCO_{2 fem}. B SvO_{2 fem} and C lactate. All data are reported as mean ± SD. ΔPCO_{2 fem} femoral venous-arterial difference in carbon dioxide pressure. SvO_{2 fem} femoral venous oxygen saturation

At admission, in the overall population ΔPCO_{2 fem} was at its highest level and then decreased to a mean minimal value at H24 of 5.4 ± 3.5 mmHg. SvO_{2 fem} was the lowest at admission and then plateaued from H6 with a mean value of 68.5 ± 16.8%. Maximum lactate was reached at H6 (2.8 ± 2.1 mmol/) and then decreased with a mean value of 2.0 ± 1.9 mmol/l at H24 (Additional file 1). At admission, ΔPCO_{2 fem} (r = 0.48; P < 0.001) and SvO_{2 fem} (r = 0.40; P < 0.004) were correlated with lactate and they were correlated together (r = 0.63; P < 0.001).

At admission, ΔPCO_{2 fem} (11.6 ± 7.1 mmHg vs. 6.8 ± 3.7 mmHg; P = 0.003), SvO_{2 fem} (50.0 ± 23.0% vs. 71.8 ± 14.1%; P < 0.001) and lactate (3.3 ± 2.4 mmol/l vs. 2.2 ± 1.3 mmol/l; P = 0.04) were significantly different in transfused compared to non-transfused patients (Table 2).

Table 2 Hemodynamic variables in the overall population and in transfused and non-transfused patients

Over the first 24 h, ΔPCO_{2 fem} decreased significantly and was statistically different between transfused and non-transfused groups without interaction between time and groups (time effect, P = 0.002; transfusion effect, P = 0.008; time x transfusion effect, P = 0.1). SvO_{2 fem} remained stable in the non-transfused group, over the first 24 h, while it significantly increased over time in the transfused group (time effect, P = 0.04; transfusion effect, P = 0.02; time x transfusion effect, P < 0.001). In the transfused group, blood lactate peaked at H6 and then decreased over the first 24 h while it decreased, from admission to H24, in the non-transfused group. Blood lactate was different between non-transfused and transfused patients without interaction between time and groups (time effect, P = 0.01; transfusion effect, P = 0.007; time x transfusion effect, P = 0.7) (Fig. 2 and Additional files 2 and 3).

Fig. 2

Evolution during the first 24 h of hemodynamic variables in transfused and non-transfused patients. A ΔPCO_{2 fem}. B SvO_{2 fem} and C lactate. All data are reported as mean ± SD. *P < 0.05; **P < 0.005; ***P < 0.001. Parameters were measured at hospital admission and over the first 24 h. ΔPCO_{2 fem} femoral venous-arterial difference in carbon dioxide pressure. SvO_{2 fem} femoral venous oxygen saturation

Transfusion prediction

The abilities of admission ΔPCO_{2 fem}, SvO_{2 fem} and blood lactate to predict the need for pRBC_H6 transfusion and the need for a hemostatic procedure during the first six hours of admission are presented in Fig. 3 and Additional file 4.

Fig. 3

ROC curves for prediction of RBC transfusion and hemostatic procedure by ΔPCO_2fem and SvO_2fem. A ROC curve for prediction of pRBC_H6 by ΔPCO_{2 fem} at admission. B ROC curve for prediction of pRBC_H6 by SvO_{2 fem} at admission. C ROC curve for prediction of hemostatic procedure during the first 6 h of admission by ΔPCO_{2 fem} at admission. D ROC curve for prediction of hemostatic procedure during the first 6 h of admission by SvO_{2 fem} at admission. AUC area under the curve. ΔPCO_{2 fem} femoral venous-arterial difference in carbon dioxide pressure. pRBC_H6 transfusion of at least 1 pack of red blood cell during the first 6 h of admission. ROC Receiver operating characteristics. SvO_{2 fem} femoral venous oxygen saturation

Admission ΔPCO_{2 fem} significantly predicted pRBC_H6 transfusion with an AUC of 0.71 (CI95% 0.57–0.86; P = 0.008) and SvO_{2 fem} significantly predicted pRBC_H6 transfusion with an AUC of 0.77 (CI95% 0.64–0.91; P < 0.001). At admission, the optimal thresholds of ΔPCO_{2 fem} to predict the need for pRBC_H6 transfusion was 8.1 mmHg and the optimal threshold of SvO_{2 fem} to predict the need for pRBC_H6 transfusion was 63%. The predictive performances of ΔPCO_{2 fem} and SvO_{2 fem} are given in Table 3.

Table 3 Predictive performances of hemodynamic variables for transfusion of red blood cells during the first six hours of admission

Admission ΔPCO_{2 fem} and SvO_{2 fem} significantly predicted the need for an emergency hemostatic procedure with respective AUCs of 0.74 (CI95% 0.58–0.90, P = 0.02) and 0.72 (CI95% 0.57–0.87, P = 0.01). A threshold of 5.9 mmHg for ΔPCO_{2 fem} and 63% for SvO_{2 fem} at admission were the optimal values to predict the need for an emergency hemostatic procedure.

Admission blood lactate did not significantly predict the need for pRBC_H6 transfusion (AUC = 0.64 CI95% 0.49–0.79; P = 0.07) or the need for a hemostatic procedure (AUC = 0.60 CI95% 0.42–0.78; P = 0.3) in the first six hours of admission.

Discussion

We performed this prospective and observational study to assess the ability of ΔPCO_{2 fem}, SvO_{2 fem} and arterial blood lactate to predict transfusion of red blood cells over the first six hours (pRBC_H6) following severe trauma. We found that the difference in femoral venous to arterial PCO₂ (ΔPCO_{2 fem}) and venous oxygen saturation (SvO_{2 fem}) at admission were predictive of RBC transfusion within the first six hours of admission, with respective optimal thresholds of 8.1 mmHg and 63%. Second, we also found that ΔPCO_{2 fem} and SvO_{2 fem} were predictive of an emergency hemostatic procedure. Third, though admission blood lactate correlated with the volume of transfused RBC over the first 6 h, it was neither predictive of pRBC_H6 transfusion nor predictive of an emergency hemostatic procedure. Fourth, ΔPCO_{2 fem} and SvO_{2 fem} normalized earlier than lactate after transfusion.

Relationship with previous studies

Venous to arterial PCO₂ difference and venous O₂ saturation have hardly been studied in the setting of severe trauma. We found that admission ΔPCO_{2 fem} and SvO_{2 fem} were predictive of pRBC_H6 transfusion. These results are consistent with studies carried out on animal models of hemorrhagic shock [17, 18] where incremental blood loss led to a progressive increase in ΔPCO₂ and decrease in SvO₂. According to Fick equation, generated CO₂ (VCO₂) equals the product of CO by the difference between mixed venous and arterial CO₂ contents (Cv-aCO₂) and is constant in aerobic and anaerobic conditions. ΔPCO₂ has been shown to be linearly related to Cv-aCO₂. Thus, increase in ΔPCO₂ during hemorrhage results from a drop in CO leading to CO₂ stagnation at tissue level with an increase in venous CO₂. During hemorrhage, an increase in ΔPCO₂ thus reflects the inadequacy between CO and metabolic activity. Hemorrhage also results in a decrease in oxygen transport to tissues responsible for an increase in oxygen extraction ratio and therefore a decrease in SvO₂. Thus, increased ΔPCO_{2 fem} and decreased SvO_{2 fem} likely reflect the decrease in tissue blood flow to the lower limbs related to the magnitude of blood loss in trauma patients.

We measured femoral ΔPCO₂ and SvO₂ (ΔPCO_{2 fem} and SvO_{2 fem}) while most clinical and experimental studies reported about central ΔPCO₂ and ScvO₂. Normal central ΔPCO₂ ranges from 2 to 5 mmHg [19]. In a study conducted in 14 healthy volunteers, ΔPCO_{2 fem} ranged from 2 to 4 mmHg while volunteers were passively cycling [20]. Several studies reported conflicting differences between ScvO₂ and SvO_{2 fem} with bias of 2.7 ± 7.9% in patients undergoing elective right heart catheterization [21] to bias of 2 to 8% in critical care patients [22, 23]. However, though SvO_{2 fem} is not a reliable surrogate of ScvO₂, we found it to be more predictive of pRBC_H6 transfusion than admission ΔPCO_{2 fem} or blood lactate.

Although there is a significant correlation between blood lactate, ΔPCO_{2 fem}, SvO_{2 fem}, these measurements are not interchangeable. Unlike what Régnier et al. reported, admission lactate was not found as a predictive marker of pRBC_H6 transfusion [5]. However, we used a lower threshold (one pRBC_H6 as compared to massive transfusion defined by 6 pack of RBC in 24 h). ΔPCO_{2 fem} and SvO_{2 fem} appeared thus to be more sensitive than blood lactate to predict hemorrhage requiring RBC transfusion. A hypothesis is that these two parameters are a direct reflection of tissue blood flow to the lower limbs while aerobic metabolism is still present [24], whereas blood lactate increases at a later stage of hemorrhage, when patient blood loss is more pronounced with concomitant triggering of anaerobic metabolism [6, 25]. Consistently with what has been shown in animals [6], following an active hemorrhage, it would seem that the drop in CO, responsible for peripheral vasoconstriction and a drop in blood flow to the lower limbs can be more rapidly detected by an elevation of ΔPCO_{2 fem} and a decrease of SvO_{2 fem} than using blood lactate level. Blood flow is indeed early reduced in musculo-cutaneous tissue (i.e. lower limbs) to preserve flow to the noble organs (i.e. heart, brain) [26], which may explain the early increase in ΔPCO_{2 fem} and decrease in SvO_2fem.

It is interesting to note that patients did not present with the same injuries in the two groups. The transfused group presented mainly abdominal and pelvic injuries, more often responsible for hemorrhage, while the non-transfused group presented mainly intracranial injuries. Although mean ΔPCO_{2 fem} was higher in the transfused group compared to the non-transfused group (11.6 vs. 6.8 mmHg), some patients had a high ΔPCO_{2 fem} in the non-transfused group. These patients presented a severe head injury with an AIS head greater than 3, or had bowel perforation without hemorrhage and others had severe thoracic injuries with a hemo- or pneumothorax. In patients with no clinical or CT evidence of hemorrhage, the presence of a high ΔPCO_{2 fem} should therefore lead to the search for another cause of cardiovascular dysfunction. Severely injured trauma patients can indeed present causes of shock other than hemorrhage like tamponade, peritonitis associated with bowel perforation, spinal cord injury but also neurogenic cardiovascular dysfunction reported in several studies in patients with severe head injury [27].

Consistent with what has been observed in animals, ΔPCO_{2 fem} and SvO_{2 fem} corrected more rapidly than lactate, which increased until H6 in the transfused group [25]. ΔPCO_{2 fem} and SvO_{2 fem} are likely corrected in parallel with the improvement of tissue blood flow following transfusion, whereas hyperlactatemia is an unreliable marker of hypoxia and hypoperfusion. Indeed, lactate production is also a consequence of cellular dysoxia secondary to injury-induced inflammation and microcirculation alterations, frequently observed after severe trauma, despite tissue perfusion improvement [28]. Unrelated to tissue dysoxia, hyperlactatemia is also frequently observed in shock state, following aerobic glycolysis activation through catecholamine-dependent ß2-receptor stimulation [29].

Implication of study findings

Our findings imply that ΔPCO_{2 fem} and SvO_{2 fem} are more sensitive than blood lactate to predict the need for early blood transfusion or hemostatic procedure, on admission of severe trauma patients. Seven patients (25%) who were transfused with RBC during the first six hours of management had indeed a high ΔPCO_{2 fem} and/or a low SvO_{2 fem} but a normal blood lactate. Moreover, the evolution of ΔPCO_{2 fem} and SvO_{2 fem} over time may help to assess transfusion effectiveness to restore adequate tissue perfusion. Cardiac output monitoring using standard tools such as cardiac ultrasound or thermodilution are not available in the trauma bay during the initial phase, and macro-hemodynamic parameters such as heart rate and mean blood pressure are poorly correlated with CO during hemorrhagic shock states. Thus, in this study, we showed that ΔPCO_{2 fem} and SvO_{2 fem} could be easily used to assess the adequation of tissue blood flow (i.e. to the lower limbs) with metabolic needs following severe trauma.

Strengths and limitations

This study has several strengths. To our knowledge, this is the first study evaluating ΔPCO₂ and SvO_{2 fem} in severe trauma patients and their capacities to predict early transfusion. Moreover, we provided new data on the level of ΔPCO_{2 fem} and SvO_{2 fem} values in such patients. These results allowed us to propose alternative markers to assess the adequacy between tissue blood flow and metabolic needs in trauma patients.

This study also has several limitations. First, it is a single-centre study with a small sample-size, in patients who required the insertion of an arterial and a venous catheter at the femoral site which prevents generalization of these results. Indeed, only patients which ICU clinician judged, severe at hospital admission [13], were included which make this study applicable only to these patients. Second, we studied ΔPCO₂ and SvO₂ at the femoral site which are not good surrogates of central ΔPCO₂ and ScvO₂, considered as standards. However, we were able to report cut-off values for these markers at femoral site for transfusion prediction. Third we do not know the normal values for ΔPCO_{2 fem} and SvO_{2 fem} at baseline but we highlighted significant differences in ΔPCO_{2 fem} and SvO_{2 fem} according to the need for transfusion, reflecting the hemorrhage-induced drop in tissue blood flow and allowing to assess resuscitation efficacy over time. Nevertheless, additional studies will be necessary to determine the range of normal value of ΔPCO_{2 fem} and SvO_{2 fem} at the femoral level. Finally, since SvO₂ is dependent on hemoglobin, its predictive value for red blood cell transfusion may have been overestimated by the effect of anemia in patients with severe hemorrage [30]. Moreover, due to the Haldane effect, it is possible that in the most severe patients with low hemoglobin level and significant metabolic acidosis, the ΔPCO_{2 fem} may have been overestimated compared to veno-arterial CO₂ content (Cv-aCO₂). Indeed, for the same level of Cv-aCO₂, ΔPCO₂ increases in case of metabolic acidosis and anemia [31] making the relationship between the two values no longer linear. This was however observed at the late phase of hemorrhage, when bloodloss was major [17] and, in a way, reflect the seriousness of hemorrhage which is consistent with what we expected to predict.

Conclusion

In severe trauma patients, ΔPCO₂ and SvO₂ measured at the femoral level at admission were predictive for the need of RBC transfusion and hemostatic procedures during the first six hours of management while admission lactate was not. In this study, ΔPCO_{2 fem} and SvO_{2 fem} appear thus to be more sensitive to blood loss than lactate in patients requiring femoral venous and arterial catheters insertion, which might be of importance to assess the adequation of tissue blood flow with metabolic needs following severe trauma.