Hypertonic Resuscitation of Hypovolemic Shock After Blunt Trauma: A Randomized Controlled Trial

Background  The leading cause of late mortality after trauma is multiple organ failure syndrome, due to a dysfunctional inflammatory response early after injury. Preclinical studies demonstrate that hypertonicity alters the activation of inflammatory cells, leading to reduction in organ injury. The purpose of this study was to evaluate the effect of hypertonicity on organ injury after blunt trauma.

Design  Double-blind, randomized controlled trial from October 1, 2003, to August 31, 2005.

Setting  Prehospital enrollment at a single level I trauma center.

Patients  Patients older than 17 years with blunt trauma and prehospital hypotension (systolic blood pressure, ≤ 90 mm Hg).

Interventions  Treatment with 250 mL of 7.5% hypertonic saline and 6% dextran 70 (HSD) vs lactated Ringer solution (LRS).

Main Outcome Measures  The primary end point was survival without acute respiratory distress syndrome (ARDS) at 28 days. Cox proportional hazards regression was used to adjust for confounding factors. A preplanned subset analysis was performed for patients requiring 10 U or more of packed red blood cells in the first 24 hours.

Results  A total of 209 patients were enrolled (110 in the HSD group and 99 in the LRS group). The study was stopped for futility after the second interim analysis. Intent-to-treat analysis demonstrated no significant difference in ARDS-free survival (hazard ratio, 1.01; 95% confidence interval, 0.63-1.60). There was improved ARDS-free survival in the subset (19% of the population) requiring 10 U or more of packed red blood cells (hazard ratio, 2.18; 95% confidence interval, 1.09-4.36).

Conclusions  Although no significant difference in ARDS-free survival was demonstrated overall, there was benefit in the subgroup of patients requiring 10 U or more of packed red blood cells in the first 24 hours. Massive transfusion may be a better predictor of ARDS than prehospital hypotension. The use of HSD may offer maximum benefit in patients at highest risk of ARDS.

Trial Registration  clinicaltrials.gov Identifier: NCT00113685

Traumatic injury is the leading cause of death among Americans between the ages of 1 and 44 years, resulting in nearly 150 000 deaths per year in the United States.¹ Early deaths occur as a result of hypovolemic shock or severe traumatic brain injury, whereas late deaths result from progressive multiple organ dysfunction or nosocomial infection.^2,3 Early deaths resulting from traumatic brain injury may also be exacerbated by inadequate cerebral perfusion, which leads to a secondary ischemic injury to the brain.

Late deaths are affected by an initial excessive systemic proinflammatory response that contributes to the development of the acute respiratory distress syndrome (ARDS) and subsequent organ dysfunction leading to multiple organ failure syndrome. Whole-body ischemia followed by reperfusion, on resuscitation of hypovolemic shock, causes excessive, uncontrolled activation of the host inflammatory response resulting in organ injury. After this initial excessive inflammatory response, many patients experience a period of immunosuppression that is manifested, in part, by alterations in T-cell responsiveness.⁴ This results in increased susceptibility to nosocomial infection, which can provide the stimulus for a secondary aberrant immunoinflammatory response that further contributes to the development of ARDS and multiple organ failure syndrome. Strategies designed to influence outcome after injury must target early deaths by focusing on the acute resuscitation of hypovolemia, while minimizing secondary brain injury, and late deaths by immunomodulation of the systemic inflammatory response.

Hypertonic saline/dextran (HSD) (7.5% saline with 6% dextran 70) has been investigated as an alternative resuscitation fluid in critically injured patients.^5-9 Use of HSD increases the serum osmotic pressure, which results in the redistribution of fluid from the interstitial to intravascular space. This leads to rapid restoration of circulating intravascular volume. This osmotic effect of HSD has also been shown to reduce intracranial pressure in brain-injured patients. Thus, the combination of increased systemic perfusion, which increases cerebral perfusion, along with a decrease in the intracranial pressure should minimize the progression of secondary brain injury. In addition, recent studies have demonstrated an effect of hypertonicity on limiting the proinflammatory response of circulating inflammatory cells while enhancing T-cell function.^10-13 Thus, HSD may have additional beneficial effects by modulating the excessive immunoinflammatory response after systemic ischemia/reperfusion injury. Therefore, HSD has the potential to influence both early and late mortality after traumatic injury. Early trials of HSD resuscitation focused on the hemodynamic effects of this resuscitation strategy and did not report secondary outcomes such as the development of ARDS or multiple organ failure syndrome. The purpose of this trial was to determine whether HSD resuscitation decreases the risk of subsequent inflammatory organ injury in patients with hypovolemic shock after blunt trauma.

This was a randomized, single-center, double-blind efficacy study. Patients with blunt trauma and a prehospital systolic blood pressure (SBP) less than or equal to 90 mm Hg were randomized to receive either a 250-mL bolus of HSD or 250 mL of lactated Ringer solution (LRS) as their initial reperfusion fluid, followed by additional LRS as necessary during transport. The LRS used was nonracemic, containing L-lactate only. Inclusion criteria were blunt trauma, age older than 17 years (or adult size if age was unknown), at least 1 prehospital SBP measurement less than or equal to 90 mm Hg, and being transported directly to a single level I trauma center from the site of injury. Exclusion criteria included ongoing cardiopulmonary resuscitation, isolated penetrating trauma, known or suspected pregnancy, and receipt of more than 2000 mL of crystalloid before availability of study fluid. Patients with penetrating trauma were excluded in an effort to capture patients at higher risk of ARDS. The primary outcome was 28-day ARDS-free survival for patients receiving HSD compared with conventional resuscitation alone. Secondary outcome measures included multiple organ failure, nosocomial infection, 28-day mortality, ventilator-free days, duration of intensive care unit stay, and duration of hospital stay.

This study was conducted under the federal regulations for emergency medicine waiver of informed consent. Community notification and consultation were undertaken before the study, as stipulated by the University of Washington Institutional Review Board and the US Food and Drug Administration. Patients were identified and enrolled at the scene of injury by the prehospital providers. All trauma admission logs were reviewed daily by the study coordinator to identify any eligible patients who were not enrolled. Informed consent was obtained after hospital admission for continuation in the study from either the subject or the subject's legal representative.

The study fluid was purchased (RescueFlow; BioPhausia Inc, Stockholm, Sweden) in 250-mL infusion bags. The HSD marketed by this company has regulatory approval for use in 14 European countries. Investigational drug approval was obtained from the US Food and Drug Administration for use of this product for this study. The fluids were transferred into blinded intravenous fluid bags by the investigational pharmacy with appropriate stability and sterility testing. A random, computer-generated numeric code was applied to each bag and kept by the research pharmacist. The control fluid (250 mL of LRS; Baxter Healthcare, Deerfield, Illinois) was prepared in intravenous bags, which were identical to those for the HSD fluid with the same quality control procedures. Randomization numbers were computer generated in blocks of 6, and 6 bags were placed at each base station, where they were retrieved by the medic units. Only 1 bag of study fluid was kept on each ambulance and 2 on each helicopter. The numbering on each randomization block was sequential, and paramedics were instructed to take bags in order to avoid selection bias. To avoid the risk that blinding would be compromised by initial changes in serum sodium level, all caretakers were blinded to the serum sodium and chloride levels for the first 12 hours after injury. Critical serum sodium and chloride levels were reported to an independent safety monitor who was not responsible for the clinical treatment of the patient but who could advise the management team if intervention was required. If the clinician required the sodium value for critical further treatment of the patient, then this information was provided.

The intended primary clinical outcome variable was the incidence of ARDS within 28 days after injury. Operationally, to account for patients who were lost to follow-up or died within 28 days before ARDS status could be determined, ARDS-free survival up to 28 days was the actual measure analyzed. The ARDS-free survival was defined as the duration from study entry to development of ARDS or death, whichever occurred first. The duration was considered as having been “censored” at the time of last contact for patients lost to follow-up earlier than 28 days and before ARDS determination. Both an unadjusted analysis (log-rank test) and an adjusted analysis (Cox proportional hazards model regression accounting for relevant baseline variables) were performed. Results are reported as the relative hazard of developing ARDS or death for those receiving conventional therapy vs HSD resuscitation.

The diagnosis of ARDS was based on the American-European Consensus Conference on ARDS definition.¹⁴ These criteria include (1) hypoxia with a ratio of PaO₂ to fraction of inspired oxygen less than 200, (2) bilateral infiltrates on chest radiographs, and (3) no clinical evidence of increased left atrial pressure or a pulmonary artery wedge pressure of less than 18 mm Hg. For those without pulmonary artery catheter monitoring, clinical evidence of left atrial hypertension included (1) acute myocardial infarction or known cardiomyopathy or severely reduced ejection fraction (< 30%) or known critical valvular disease; and (2) chronic or acute oliguric renal failure with fluid input that exceeded output by 3 L or more in the previous 24-hour period. Acute lung injury has been defined as a milder form of ARDS with the same clinical criteria except for a ratio of PaO₂ to fraction of inspired oxygen less than 300.

Multiple Organ Failure Syndrome. The development of additional system organ dysfunction was tracked by the well-validated Multiple Organ Dysfunction Score (MODS).¹⁵ Its continuous nature allows detection of subtle differences in organ dysfunction not identified by dichotomous measures. The MODS assigns points to each of the 6 organ systems indicated, and the summary score is calculated by summing the worst scores of each organ system over the course of the intensive care unit stay. Because the MODS is designed to measure stable alterations in organ function, the first 48 hours after injury are excluded. Those who died in the first 48 hours were assigned the maximum MODS of 24, and those discharged before 48 hours were given a MODS of 0.

Nosocomial Infections. All nosocomial infections occurring within the first 28 days after injury were recorded, including bacteremia, pneumonia, urinary tract infection, surgical site infections, and intra-abdominal abscess. The Centers for Disease Control and Prevention definitions were used for these entities (Table 1).

Resource Utilization and Mortality. Additional secondary outcome variables included 28-day mortality, duration of hospital and intensive care unit stay, and ventilator-free days through day 28. Ventilator-free days were calculated as the number of days during which no ventilator support was required over the first 28 days.

Adverse Events and Noninfectious Complications. Serious adverse events were predefined to include any evidence of allergic reaction to HSD, seizure activity associated with infusion, a serum sodium level higher than 160 mEq/L (to convert to millimoles per liter, multiply by 1) requiring therapeutic intervention, or any death not consistent with injury severity. Additional data were collected regarding noninfectious complications, including acute renal failure, abdominal compartment syndrome, deep venous thrombosis, cardiac arrest, myocardial infarction, pulmonary embolism, and cerebral infarction.

On the basis of the reported incidence of ARDS after severe traumatic injury of 25% to 36%,^16-18 we assumed a 28-day ARDS-free survival rate of 65% for the control group and estimated a 15% improvement for the treatment group with an ARDS-free survival of 80%. We calculated that 381 uncensored observations would be required to detect this difference on the basis of a 2-sided log-rank analysis with a power of 0.9 (β = .1) and an α of .05. Thus, our planned enrollment was 400 patients, 200 in each treatment arm.

The trial was monitored by an independent data safety and monitoring board. Three interim analyses were planned at equal quartiles of 28-day completion, specifically, when 25%, 50%, and 75% of the targeted 400 patients completed the 28-day period. Stopping for futility or efficacy was based on formal group sequential stopping boundaries. The group sequential method described by O’Brien and Fleming¹⁹ was used to develop stopping rules to limit the effect of repeated testing on the probability of a type I error. In particular, at 25%, 50%, and 75% analysis, results with P values less than .000045, .0039, and .018, respectively, constituted evidence for consideration of early termination.

Data analysis was performed on an intent-to-treat basis. Thus, patients who were enrolled by the prehospital providers and who were subsequently identified as meeting exclusion criteria remained in the analysis. Unadjusted analyses of ARDS-free survival and mortality were presented by means of a Kaplan-Meier approach with log-rank test to determine statistical significance. To account for potential differences between the groups for baseline characteristics and injury severity, a Cox proportional hazards model was used to estimate the adjusted hazard ratio (HR). For comparison of demographic and injury severity data among the groups, the χ² test was used for categorical variables and unpaired t test or Wilcoxon rank sum test as appropriate for continuous variables. Where the data were not normally distributed, median and mean are both presented.

Preplanned subgroup analyses for the primary outcome were performed for the following groups: age greater than 55 years, head Abbreviated Injury Scale (AIS) score greater than 2, chest AIS score greater than 3, Injury Severity Score (ISS) greater than 25, massive transfusion (≥ 10 U of red blood cell transfusion during the first 24 hours after injury), and survival longer than 48 hours. To account for the fact that there may be an interaction between the treatment and the number of units of blood required, an additional analysis was conducted using the number of units transfused as an interaction term.

During the study period (October 1, 2003, to August 31, 2005), 4260 persons older than 17 years with blunt trauma were admitted to our trauma center directly from the scene of injury. Of these, 261 patients (6.1%) had a prehospital SBP less than or equal to 90 mm Hg and, of these, 209 were enrolled (Figure 1). The reasons for failure to enroll patients are indicated in Figure 1. Of the patients enrolled, 21 (9 in the HSD group and 12 in the LRS group) were subsequently found to meet 1 of the exclusion criteria (Table 2) but remained in the analysis on an intent-to-treat basis. For 1 patient in the HSD group we could not confirm that the entire volume of fluid had been infused. All providers were blinded as to the fluid administered, and we attempted to blind care providers to the serum sodium and chloride levels for the first 12 hours after injury. Complete 12-hour blinding was maintained for 78% of patients in both groups. Three patients were intentionally unblinded at the request of the neurosurgical service to allow for the administration of additional 3% saline to control elevated intracranial pressure. Two patients in the LRS arm of the study were pronounced dead in the field and were not transported to the hospital. They remained in the analysis for mortality and ARDS-free survival. Three patients (all from the LRS arm) who were ARDS-free at discharge were lost to follow-up before day 28: 1 was sent home on day 6 and subsequently had the telephone disconnected; the other 2 were homeless persons who left the hospital against medical advice on days 5 and 7.

The demographics and mechanism of injury of the study cohort are given in Table 3. There were no significant differences between the treatment groups relative to age, race, sex, or mechanism of injury. Although there were no significant differences in injury severity between the groups, by chance, the HSD group trended toward higher injury severity as manifested by the proportion of patients with an ISS greater than 25, head AIS score greater than 2, and chest AIS score greater than 3 (Table 4). Approximately 22% of patients in the HSD group and 15% of patients in the LRS group (P = .22) required massive transfusion, as defined by the need for 10 U or more of red blood cells during the first 24 hours after injury. The prehospital care provided is given in Table 5. A higher proportion of patients in the HSD group were transported by aeromedical transport (45.5% vs 34.3% in LRS group). Patients in the HSD group were also more likely to be intubated in the prehospital setting (72.7% vs 62.6%) and received a higher volume of total prehospital fluids.

There was no difference in the mean prehospital blood pressure before study drug administration or blood pressure on arrival to the emergency department (ED) (Table 5). Patients in the HSD group did tend to maintain their blood pressure better in the ED, as shown by the lowest systolic pressure recorded during this period. As expected, patients in the HSD group had a significant elevation in serum sodium level on admission (mean [SD], 147 [6] mEq/L vs 140 [4] mEq/L in the LRS group; P < .001). Admission hematocrit was also lower in the HSD group, but there was no significant difference in markers of coagulopathy. There was also no difference in the degree of metabolic acidosis in the ED, as indicated by initial lactate level or arterial pH.

The primary outcome of ARDS-free survival was assessed by a Kaplan-Meier approach (Figure 2). The estimated 28-day ARDS-free survival rate was 54% for the HSD group and 64% for the LRS group (P = .16, log-rank test). The unadjusted HR for LRS vs HSD was 0.75 (95% confidence interval [CI], 0.49-1.15). Nine of the 41 patients with ARDS (6 in the HSD arm and 3 in the LRS arm) subsequently died before day 28. To account for differences in injury severity, a Cox proportional hazards analysis was performed that included the following variables: age older than 55 years, head AIS score greater than 2, chest AIS score greater than 3, ISS greater than 25, and massive transfusion (≥10 U of packed red blood cells per 24 hours). This resulted in an LRS vs HSD HR of 1.01 (95% CI, 0.63-1.60). Factors that were significantly associated with decreased ARDS-free survival were as follows: age older than 55 years (HR, 1.88; 95% CI, 1.12-3.14), ISS greater than 25 (HR, 4.27; 95% CI, 2.12-8.59), and 10 U or more of packed red blood cells (HR, 2.65; 95% CI, 1.61-4.37).

There were no significant differences in the secondary outcome measures (Table 6). The most common nosocomial infection in both groups was pneumonia. There were no adverse events that were judged to be related to treatment. Specifically, there was no evidence of allergic reaction to HSD and no reports of seizures or elevated serum sodium levels that required therapeutic intervention. The noninfectious complications are listed in Table 7. There was a higher rate of deep venous thrombosis in the LRS group (P = .03). The estimated 28-day mortality was 29% for the HSD group and 22% for the LRS group (P = .30). The Kaplan-Meier curves for 28-day survival are shown in Figure 3. Sixty-five percent (35 of 54) of the deaths occurred within the first 48 hours after injury.

Preplanned observational analysis of relevant patient subgroups included the following: age older than 55 years, head AIS score greater than 2, chest AIS score greater than 3, ISS greater than 25, massive transfusion (≥10 U of red blood cell transfusion during the first 24 hours after injury), and survival longer than 48 hours. There was no significant advantage of HSD treatment in these sub groups, with the exception of the massive transfusion group. The Kaplan-Meier curve for ARDS-free survival for this subgroup is shown in Figure 4. The estimated 28-day ARDS-free survival rate was 13% for the HSD group vs 0% for the LRS group. This resulted in an HR of 2.03 (95% CI, 0.94-4.40). When the number of units transfused was considered as an interaction term, the HR was 2.18 (95% CI, 1.09-4.36). Given the small number of patients in this subgroup and the number of subgroup analyses, this cannot be considered statistically significant.

The study was closed on the basis of futility after the second Data Safety Monitoring Board interim analysis, which was based on analysis of the first 200 patients enrolled. At that time, the adjusted LRS vs HSD HR for ARDS-free survival based on treatment was 1.02 (95% CI, 0.64-1.63; P = .94). With P values of .25 by log-rank test for the ARDS-free survival duration up to 28 days and .18 for the binary 28-day rate, there were no indications that the null hypothesis of no treatment difference should be rejected for either outcome. To assess the study's sample size calculation assumption, the alternative hypothesis used to power the study (ie, the HDS arm's 28-day ARDS-free survival rate is higher than that of the control arm by 15% [eg, 80% vs 65%]) was also tested. The P value for this test was .0002, which constitutes evidence that the hypothesis of a 15% HDS advantage is highly unlikely. Furthermore, the P value for a 10% HDS advantage was .0027, indicating that a theoretical 10% HDS benefit was highly unlikely. Should there be a 10% difference (eg, 70% vs 60%), it would require 990 patients to have a power of 0.90 to observe a statistically significant difference. With 400 patients, the power to detect such a difference is 0.51. On the basis of these interim analyses results, the independent Data Safety Monitoring Board judged it futile to continue enrollment in a single-center trial and recommended early study termination. In addition, the 1 subgroup that did have some suggestion of benefit was the group requiring massive transfusion, and this subgroup represented only 19% of the population, suggesting that more specific inclusion criteria would be necessary to capture this severely injured population.

Previous to this study, there have been 8 clinical trials of HSD resuscitation after hypovolemic shock and 1 focused on outcome after traumatic brain injury.^5-9,20-23 Early studies of hypertonic resuscitation focused solely on the hemodynamic effects of restoring blood pressure with a smaller volume of resuscitation fluid. Of the 8 trials on hypovolemic shock, 2 involved administration of the fluid in the ED and 6 were prehospital studies. Data from animal models suggest that HSD resuscitation is most effective when given as the initial resuscitation fluid at the time of reperfusion.²⁴ This is consistent with the finding of a survival advantage in the prehospital trials that was not evident in the ED-based trials. Many of these studies were limited by their sample size, and several closed short of their power calculations for logistical reasons. The largest evaluation of HSD resuscitation was a multicenter trial by Mattox et al⁶ in 1991. This trial involved prehospital administration of HSD in 3 US cities. Although designed to be representative of the entire trauma population, this trial had a much higher percentage of patients with penetrating trauma (72%) than that seen in most studies. As a result, they were unable to evaluate any effect on traumatic brain injury. They did report a trend toward a decrease in the incidence of ARDS; however, only 2 patients in the cohort developed ARDS, which is a much lower incidence than that seen in the average blunt trauma population. This observation led us to restrict this trial to patients with blunt trauma. Subsequent meta-analyses of these data, however, did suggest a survival advantage to hospital discharge (odds ratio, 1.47; 95% CI, 1.04-2.08).²⁵ Furthermore, patients who required blood transfusion or immediate surgical intervention for bleeding showed an even greater survival benefit from HSD.

A recently completed trial by Cooper et al²⁰ from Australia focused on the effect of resuscitation with 7.5% saline (without dextran) on neurologic outcome in patients with a prehospital Glasgow Coma Scale score less than 8 and an SBP less than 100 mm Hg. This trial was also limited by a small sample size, with only 229 patients enrolled, and a 50% mortality, thus limiting the number of patients available for outcome assessment. Interestingly, although it was not statistically significant, they did observe a trend toward improved survival at 6 months in the hypertonic group (odds ratio, 1.17; 95% CI, 0.9-1.5; P = .23). Of the patients who survived to the ED, the long-term survival was 67% for those receiving hypertonic saline vs 55% for the LRS group (odds ratio, 1.72; 95% CI, 0.95-3.1; P = .07).

Subsequent to these early trials, a number of studies have demonstrated the profound effects of hypertonicity on the modulation of the inflammatory response. This includes a transient inhibition of the innate immune response with preservation and enhancement of the adaptive immune response.^13,26-29 Animal models of acute lung injury after hypovolemic shock have suggested that HSD resuscitation may significantly attenuate the development of inflammatory lung injury.^26,30 It is unknown whether these changes observed in the laboratory will translate into a decrease in the risk of ARDS or MODS in patients resuscitated from hypovolemic shock. Human studies of the administration of 7.5% saline to women undergoing elective hysterectomy failed to show any significant changes in plasma cytokine levels or the phenotype of immune cells.^31,32 More recent data on HSD resuscitation for patients in hypovolemic shock have demonstrated significant changes in neutrophil and monocyte function consistent with animal models.^10,33 Thus, these effects may be evident only in patients with a significant inflammatory insult.

A prominent theory for the development of MODS after injury involves a systemic activation of the inflammatory response due to whole-body ischemia/reperfusion injury after hemorrhagic shock coupled with release of inflammatory mediators from damaged tissue.³⁴ This results in dysfunctional activation of the innate immune response and a subsequent susceptibility to infection as a result of a counterinflammatory response. Hypertonic resuscitation has the potential to modulate this initial response and thus minimize susceptibility to a secondary insult. It is unknown whether this single initial dose is adequate to abate this process or whether maintenance of hypertonicity for a longer period is required. The addition of dextran to 7.5% saline is purported to prolong these effects for up to 4 hours. Furthermore, lung injury is the most common and initial manifestation of postinjury organ failure and thus may be most representative of the initial effects of hypertonic resuscitation.³⁵ Therefore, the purpose of this project was to determine whether HSD resuscitation at the time of reperfusion would reduce the incidence of ARDS and subsequent MODS in this patient population.

We selected the inclusion criterion of a prehospital SBP less than or equal to 90 mm Hg on the basis of previous studies that suggested that this value selects for a patient population with severe injury.^36-38 This was also the primary inclusion criterion for previous trials of HSD resuscitation. What we observed, however, was that a single prehospital SBP less than or equal to 90 mm Hg was not a specific marker for hypovolemic shock, with 44.5% of the patients enrolled in this trial not requiring any blood transfusions in the first 24 hours after injury. Thus, we maintain that the primary reason for the futility outcome of this trial was the failure to enroll the patient population at highest risk for ARDS and thus most likely to benefit from this intervention. Massive transfusion appears to be a much better predictor of ARDS than does prehospital SBP less than or equal to 90 mm Hg. This is consistent with several previous studies that have confirmed the high risk of ARDS and MODS associated with blood transfusion.^39,40 This finding has implications for the future design of trials evaluating resuscitation strategies in the prehospital setting. In response to these data, we have now altered the inclusion criteria for an upcoming multicenter trial of hypertonic resuscitation sponsored by the Resuscitation Outcomes Consortium (funded by the National Heart, Lung, and Blood Institute). This trial will enroll patients with a prehospital SBP less than 70 mm Hg or 70 to 90 mm Hg with a heart rate greater than 108 beats/min. These prehospital vital signs were associated with a higher proportion of patients requiring massive transfusion. Future efforts should focus on additional markers of shock in the prehospital setting to better identify patients who warrant early novel interventions. More sophisticated monitoring, such as evaluation of heart rate variability, measurement of tissue oxygen saturation, or rapid assessment of acid-base status using portable lactate or blood gas assessment tools, might be of use for future studies.^41-43

There were several limitations to the trial results as they stand. Despite randomization, there appears to be a chance inequity between the treatment groups, with a higher severity of injury and greater need for massive transfusion in the HSD group, which thus required multivariate analysis to evaluate the treatment effect. In addition, a higher proportion of patients received prehospital intubation in the HSD group. Prehospital intubation has been associated with hyperventilation, which may be detrimental to patients with severe traumatic brain injury.^44,45 In addition, despite efforts to blind care providers to the serum sodium and chloride levels for the first 12 hours after injury, this was successful only 78% of the time. Thus, if clinicians were aware of elevated serum sodium levels, they could have altered the resuscitation of the patients and thus diminished the effect of the intervention. All care providers were notified of the study and instructed not to alter therapy on the basis of hypernatremia alone, but it is possible that this may have occurred. The admission serum sodium level in this trial was on average 147 mEq/L. This is consistent with a recent trial in which HSD was administered to patients in hypovolemic shock in the ED and the mean serum sodium level 1 hour after administration was also 147 mEq/L.¹⁰ In the previous clinical trials of HSD in the prehospital setting, mean admission sodium level ranged from 148 to 154 mEq/L.^5-7,9,21,22 It has been argued that the efficacy of hypertonic resuscitation is reduced by the rapid administration of additional crystalloid solutions during ongoing resuscitation. Patients in the HSD group did receive more crystalloid solution in the prehospital setting, but this is likely owing to longer transport times as a result of a greater proportion of air transport in this group. We cannot rule out that the effects of hypertonicity on the inflammatory response may require a more significant period of hypernatremia. However, in the recently conducted ED trial,¹⁰ significant modulation of neutrophil and monocyte function occurred up to 6 hours after fluid administration with serum sodium levels comparable with those observed in this trial. A more prolonged or extreme initial hypernatremia may be required to affect clinical outcomes. To study this would require repeated dosing of HSD, which is currently limited because of the lack of available safety data in humans; such data are vital to support trials conducted under the emergency medicine waiver of informed consent regulations.

In conclusion, we demonstrate no difference in the primary or secondary outcomes assessed after hypertonic resuscitation in a blunt trauma population presenting with a prehospital SBP less than or equal to 90 mm Hg. There is the potential for benefit in the subgroup of patients requiring massive transfusion, which may be a marker for more severe hypovolemia or may directly affect the host immunoinflammatory response. It is, of course, possible that HSD has no relevant effect in humans at this dose; however, the weight of the current evidence still favors a survival advantage for those most severely injured. Further study is necessary to define this effect. We are now proceeding with a multicenter trial conducted by the Resuscitation Outcomes Consortium to allow adequate power to assess the effect of hypertonic resuscitation on survival after injury. This study is planned to enroll 3726 patients in hypovolemic shock after injury into 3 study arms: 7.5% saline with and without dextran vs isotonic sodium chloride solution. Secondary end points will include the development of ARDS and MODS and should provide a definitive answer regarding the role of this treatment strategy in the resuscitation of these severely injured patients.

Correspondence: Eileen M. Bulger, MD, Department of Surgery, Campus Box 359796, Harborview Medical Center, 325 Ninth Ave, Seattle, WA 98104 (ebulger@u.washington.edu).

Accepted for Publication: September 12, 2006.

Author Contributions: Dr Bulger was the principal investigator and thus had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Bulger, Jurkovich, Nathens, Copass, Awan, and Maier. Acquisition of data: Bulger, Copass, Hanson, Cooper, Neff, Awan, and Warner. Analysis and interpretation of data: Bulger, Jurkovich, and Liu. Drafting of the manuscript: Bulger, Copass, Awan, and Maier. Critical revision of the manuscript for important intellectual content: Bulger, Jurkovich, Nathens, Copass, Hanson, Cooper, Liu, Neff, Awan, Warner, and Maier. Statistical analysis: Liu. Obtained funding: Bulger and Nathens. Administrative, technical, and material support: Bulger, Jurkovich, Copass, Hanson, Cooper, Neff, Awan, Warner, and Maier. Study supervision: Bulger, Jurkovich, Copass, Neff, Awan, and Maier. Study drug preparation: Awan.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grant R01 HL073233-01 from the National Institutes of Health.

Previous Presentation: This study was presented at the Shock Society Meeting; June 4, 2006; Broomfield, Colorado.

Additional Contributions: We thank the paramedics and medical directors from the Seattle Fire Department, King County Medic One, and Snohomish County Medic One and the airlift nurses from Airlift Northwest for their participation in this trial. We also thank the members of the Data Safety Monitoring Board for their oversight.