Presented at the Surgical Forum, American College of Surgeons Annual Meeting, Chicago, IL, October 2006.
Awarded first place, Minnesota State Trauma Paper Competition, American College of Surgeons, Minneapolis, MN, October 2006; and second place, Region V Trauma Paper Competition, American College of Surgeons, Indianapolis, IN, November 2006.
This work was funded by the Office of Naval Research (Grant N0001-14-02-1-0093 to GJB).
Address for correspondence and reprints: Greg Beilman, MD; e-mail: firstname.lastname@example.org.
Objectives: The aim of this study was to compare hypotensive and normotensive resuscitation in a porcine model of hemorrhagic shock.
Methods: This was a prospective, comparative, randomized survival study of controlled hemorrhagic shock using 28 male Yorkshire-Landrace pigs (15 to 25 kg). In 24 splenectomized pigs, the authors induced hemorrhagic shock to a systolic blood pressure (sBP) of 48 to 58 mm Hg (∼35% bleed). Pigs were randomized to undergo normotensive resuscitation (sBP of 90 mm Hg, n = 7), mild hypotensive resuscitation (sBP of 80 mm Hg, n = 7), severe hypotensive resuscitation (sBP of 65 mm Hg, n = 6), or no resuscitation (n = 4). The authors also included a sham group of animals that were instrumented and splenectomized, but that did not undergo hemorrhagic shock (n = 4). After the initial 8 hours of randomized pressure-targeted resuscitation, all animals were resuscitated to a sBP of 90 mm Hg for 16 hours.
Results: Animals that underwent severe hypotensive resuscitation were less likely to survive, compared with animals that underwent normotensive resuscitation. Mean arterial pressure (MAP) decreased with hemorrhage and increased appropriately with pressure-targeted resuscitation. Base excess (BE) and tissue oxygen saturation (StO2) decreased in all animals that underwent hemorrhagic shock. This decrease persisted only in animals that were pressure target resuscitated to a sBP of 65 mm Hg.
Conclusions: In this model of controlled hemorrhagic shock, initial severe hypotensive pressure-targeted resuscitation for 8 hours was associated with an increased mortality rate and led to a persistent base deficit (BD) and to decreased StO2, suggesting persistent metabolic stress and tissue hypoxia. However, mild hypotensive resuscitation did not lead to a persistent BD or to decreased StO2, suggesting less metabolic stress and less tissue hypoxia.
For battlefield-wounded soldiers with hemorrhagic shock in the United States Armed Forces, the current standard of care is initial hypotensive resuscitation to a weakly palpable radial artery pulse1–3 that correlates to a systolic blood pressure (sBP) of 80 mm Hg.4 In 1918, Cannon et al.5 initially proposed limiting fluid resuscitation for patients with hemorrhagic shock before controlling active bleeding: “restoration of blood pressure prior to control of active bleeding may result in loss of blood that is sorely needed.” However, clinical data supporting hypotensive resuscitation before surgical control of hemorrhagic shock did not become available until 1994.6 Since 1994, application of hypotensive resuscitation to civilian injuries in the United States has been limited. Many cite rapid transport times and early surgical intervention as reasons to initially fully resuscitate, because there is too little time for hypotensive resuscitation to have a positive effect on outcomes. However, the transport time for battlefield-wounded soldiers in Afghanistan, and to a lesser extent in Iraq, has allowed the implementation of early hypotensive resuscitation for hemorrhagic shock in those arenas.1
The critical missing elements in the doctrine of early hypotensive resuscitation are how low and how long are safe? Little is known regarding the physiologic effects of prolonged hypotensive resuscitation. In particular, questions remain regarding the safety of extended-duration low sBP. In theory, low sBP should prevent rebleeding at sites where a clot has formed in a low-flow state. In a porcine model of aortic injury, rebeeding occurred at a mean sBP of 94 mm Hg.7 Too much resuscitation may lead to “popping of the clot,” rebleeding, subsequent anemia, and coagulopathy. Alternatively, too little resuscitation might lead to tissue hypoxia and to increased metabolic stress, resulting in intracellular injury, increased extracellular fluid requirements, and possibly death of the organism.
The aim of our study was to address the first critical missing element in the doctrine of hypotensive resuscitation: how low is safe? We designed our study to compare the mortality rate and physiologic parameters in pigs that were initially resuscitated to normotensive (sBP of 90 mm Hg, n = 7), mild hypotensive (sBP of 80 mm Hg, n = 7), and severe hypotensive (sBP of 65 mm Hg, n = 6) pressure targets after hemorrhagic shock. We developed a prospective, comparative, randomized survival study of controlled hemorrhagic shock designed to mimic combat injuries and prolonged transport time before definitive surgical control (Table 1). We chose to target resuscitation endpoints in sBP because this is an easily obtainable measurement in the field and does not require additional calculation. We chose a period of 8 hours for hypotensive resuscitation because: 1) although current transport times in Iraq are shorter, previous experience in Afghanistan and Somalia has demonstrated delays in care of battlefield-wounded soldiers suffering from hemorrhagic shock due to battlefield conditions of up to 14 hours,1,8 2) recent published data9,10 and preliminary data from our laboratory suggest that longer hypotensive resuscitation time periods are more likely to reveal clinically relevant hemodynamic and metabolic changes, and 3) we wanted to identify a dose–response effect of different resuscitation endpoints on prolonged hypotensive resuscitation. We chose to use a controlled model of hemorrhagic shock to isolate the effects of prolonged hypotensive resuscitation on physiology without the confounding effects of rebleeding. We hypothesized that animals initially resuscitated to a severe hypotensive (vs. normotensive) target pressure would have an increased mortality rate and increased signs of ongoing tissue hypoxia and of metabolic stress.
Table 1. Current Battlefield Standard of Care versus Our Porcine Model
Resuscitation to a weakly palpable radial artery pulse
No hemorrhage (n = 4)
sBP 90 mm Hg (n = 7)
sBP 80 mm Hg (n = 7)
sBP 65 mm Hg (n = 6)
No resuscitation (n = 4)
Full resuscitation (16 hours)
sBP 90 mm Hg
UO 0.5 mL/kg/hr
Hgb 6.0 g/dL
Extubate (24 hours)
This was a prospective, comparative, randomized survival study of controlled hemorrhagic shock using a porcine model. The University of Minnesota Institutional Animal Care Committee approved our study protocol. The care and handling of animals were in accord with the National Institutes of Health guidelines for ethical animal research.
A total of 28 male Yorkshire-Landrace pigs (Covance Research Products, Kalamazoo, MI) weighing 15 to 25 kg were fasted 24 hours before surgery but were allowed water ad libitum.
Resuscitation protocols were enclosed inside identical envelopes and randomly chosen on the day of the pig’s protocol (Figure 1). The pigs were anesthetized with an intramuscular dose of ketamine (10 to 12 mg/kg, Abbott Labs, Chicago, IL) and althesin (2 mg/kg, Abbyvett Export Ltd., North Yorkshire, England). Anesthesia was maintained throughout the experiment by intravenous infusion of a combination of althesin (7 to 14 mg/kg/hr) for the first 8 hours after resuscitation and then propofol for the time period between 8 and 24 hours after resuscitation (AstraZeneca Pharmaceuticals, Wilmington, VA), along with 60% inhaled nitrous oxide. The pigs were orally intubated and ventilated (Adult Star 2000, Infrasonics Inc., San Diego, CA) with supplemental oxygen to maintain a partial pressure of arterial oxygen (PaO2) of 70 to 120 torr and a partial pressure of arterial carbon dioxide (PaCO2) of 35 to 45 torr.
We surgically exposed the right carotid artery and jugular vein and then placed a catheter in the carotid artery, to continuously measure blood pressure and to intermittently sample arterial blood. We also placed an introducer (7 French Avanti, Cordis Corp., Miami Lakes, FL) and a Swan-Ganz catheter (5 French, Edwards Lifesciences, Irvine, CA) in the right jugular vein to intermittently measure pulmonary artery occlusion pressure (PAOP) and thermodilution cardiac output and to intermittently sample mixed venous blood. We placed a tissue oxygenation saturation (StO2) monitor on the medial aspect of the left hind limb (Hutchinson Technology Inc., Hutchinson, MN).
Next, we performed a midline laparotomy and splenectomy to reduce autotransfusion. We also placed catheters in the urinary bladder (via a stab cystotomy) and in the inferior vena cava (IVC). The surgical preparation was followed by a stabilization period until the blood lactate levels via blood gas analyzer (Instrumentation Laboratory Co., Lexington, MA) were less than 2.0 mmol/L.
Hemorrhagic Shock Model
We induced hemorrhagic shock in 24 animals by removing blood via the IVC catheter to obtain a sBP of 48 to 58 mm Hg (typically, 35% total blood volume, or 30 mL/kg). Each pig was then maintained at this sBP for 45 minutes by removing or replacing blood as necessary. After this 45-minute period of sustained hemorrhagic shock, the animals were randomized to undergo normotensive resuscitation (sBP of 90 mm Hg, n = 7), mild hypotensive resuscitation (sBP of 80 mm Hg, n = 7), severe hypotensive resuscitation (sBP of 65 mm Hg, n = 6), or no resuscitation (n = 4). The animals that did not survive to the end of the shock period were excluded from the study (n = 2). The pigs were resuscitated in a stepwise fashion by infusing first Hextend (5 mL/kg/bolus) to a maximum of 500 mL and then lactated Ringer’s solution (10 mL/kg/bolus), until their randomized pressure-targeted resuscitation goal was met. We chose to use Hextend as the initial resuscitation fluid because this is the current practice in the U.S. Armed Forces. During the initial 8-hour resuscitation, the animals received additional fluid as necessary to maintain their randomized resuscitation goal. We also included a sham group of animals that were instrumented and splenectomized but not hemorrhaged (n = 4).
After the initial 8 hours of resuscitation, the surviving animals were resuscitated using Hextend and lactated Ringer’s (as a continuation of the protocol described above) to a target sBP of 90 mm Hg and a urine output of 0.5 mL/kg/hr. Additionally, we autotransfused whole blood (blood removed during the controlled hemorrhage and maintained in a citrate buffer at 4°C) to maintain a blood hemoglobin (Hgb) of greater than 6.0 g per dL.
For the duration of intubation, the pig’s core temperature was continuously monitored via the pulmonary artery catheter and maintained at 39°C by heating blankets and cold packs as necessary. Invasive monitors were removed and the animals were extubated 24 hours after the initiation of resuscitation and allowed food and water ad libitum. Then, the pigs were again sedated and intubated 48 hours after the initiation of resuscitation. At this final time point, the animals were euthanized.
We measured and calculated physiologic parameters at baseline (before hemorrhagic shock), after 45 minutes of shock, and 8 and then 24 hours after the initiation of resuscitation. Physiologic measurements included heart rate (HR), mean arterial pressure (MAP), PAOP, temperature, cardiac output, blood Hgb, StO2, systemic oxygen delivery (DO2), and systemic oxygen consumption (VO2). We analyzed arterial and venous blood via a blood gas analyzer (Instrumentation Laboratory Co.), to measure PaO2, PaCO2, mixed venous oxygen saturation, partial pressure of venous carbon dioxide (PvCO2), blood hematocrit, base excess (BE)/base deficit (BD), and blood lactate levels. All four animals that underwent hemorrhagic shock but were not resuscitated died 15 to 90 minutes after the end of the shock period.
For all statistical analyses, we used SPSS 13.0.0 for Macintosh (SPSS, Chicago, IL). We calculated and assessed means, standard deviations, and normality for all continuous variables, separately and by randomized group. We analyzed variables during resuscitation, beginning with the time point at the end of 45 minutes of shock, using analysis of variance with repeated measures for different time points. For post hoc multiple tests, we used a Wilcoxon signed rank test, with Bonferroni corrections for multiple testing. For comparisons between baseline and the end of 45 minutes of shock, we used a Wilcoxon signed-rank test. For comparisons for individual time points between subgroups, we used a Mann-Whitney U-test. For graphical representation of mortality between randomization groups, we created a Kaplan-Meier survival graph. We used Fisher’s exact test to compare overall mortality between randomization groups. A p-value of 0.05 was considered significant.
We used our previous experience with similar models to estimate the variance at 7%.11,12 We set the p-value < 0.05, confidence intervals (CIs) = 0.95, and assumed a margin of error (m) = 15%. Power calculations using a paired t-test yielded an estimated sample size of 12 animals per group, given: n = ((z * s)/m)2 = ((1.960(0.26))/0.15)2 = 12 animals. We planned for an interim analysis after 6 animals in each group. This interim analysis revealed that we had reached statistical significance after only six animals per group, and we decided to halt the study.
We found no significant differences in the volume of hemorrhage between any of the four randomized (90 mm Hg, 80 mm Hg, 65 mm Hg, and no resuscitation) groups. All control animals that underwent hemorrhagic shock but were not resuscitated died within 90 minutes after the shock period. In contrast, all sham animals that were instrumented but not hemorrhaged survived to the end of our study. The mortality rate was greater in animals randomized to severe hypotensive resuscitation, compared with normotensive resuscitation (Figure 2). The causes of death for the 10 animals that underwent hemorrhagic shock and resuscitation and eventually died were spontaneous arrhythmia, airway edema, abdominal compartment syndrome, brain dysfunction, and severe hypoglycemia (Table 2).
Table 2. Mortality Information for Animals
Time of Death
Cause of Death
aRandomized group (i.e., initial resuscitation to a target systolic blood pressure of 90, 80, or 65 mm Hg).
bThis pig was unresponsive with bilateral fixed and dilated pupils 4 hours after discontinuation of all forms of sedation.
cUpon extubation, this pig was unable to breathe due to massive tracheal edema and expired.
dThis pig developed severe abdominal distension, increased peak airway pressures, decreased tidal volumes, and bladder pressures in excess of 40 cm H2O.
eThis pig developed seizure activity that was unresponsive to anticonvulsive therapy and died.
fThis pig developed a spontaneous arrhythmia despite normal labs and expired.
gThis pig developed hypoglycemia unresponsive to intravenous dextrose and expired.
90 mm Hg
22 hours after resuscitation
80 mm Hg
26 hours after resuscitation
80 mm Hg
8 hours after resuscitation
Abdominal compartment syndromed
80 mm Hg
6 hours after resuscitation
80 mm Hg
2 hours after resuscitation
65 mm Hg
26 hours after resuscitation
65 mm Hg
17 hours after resuscitation
65 mm Hg
39 hours after resuscitation
65 mm Hg
5 hours after resuscitation
65 mm Hg
15 hours after resuscitation
When we subdivided the animals by pressure-targeted randomization groups and analyzed the data, we found that pigs in the three randomization subgroups (90, 80, and 65 mm Hg) showed changes in all physiologic parameters: HR (increased with hemorrhage and decreased with resuscitation, not shown), MAP, PAOP (decreased with hemorrhage and increased with resuscitation, not shown), cardiac output (decreased with hemorrhage and increased with resuscitation, not shown), Hgb (decreased with hemorrhage and decreased with resuscitation, not shown), DO2 (decreased with hemorrhage and increased with resuscitation, not shown), VO2 (unchanged and not shown), BE, blood lactate levels (increased with hemorrhage and decreased with resuscitation, not shown), and StO2. Those changes were consistent with hemorrhagic shock and resuscitation. Additionally, those same physiologic parameters were appropriate in the nonresuscitated and sham groups.
Mean arterial pressure decreased in all animals that underwent hemorrhagic shock. After 8 hours of pressure-targeted resuscitation, animals that were initially resuscitated to a target sBP of 80 or 65 mm Hg had lower MAP, compared with animals that were initially resuscitated to target sBP of 90 mm Hg. In all surviving randomization groups, MAP remained below baseline after 24 hours of resuscitation (8 hours of pressure-targeted resuscitation to sBP of 65, 80, or 90 mm Hg followed by 16 hours of resuscitation to a target sBP of 90 mm Hg). However, there was no difference between sham animals and those animals that were hemorrhaged and resuscitated (groups initially pressure target resuscitated to sBP of 65, 80, and 90 mm Hg; Figure 3).
Base excess decreased in all animals that underwent hemorrhagic shock. After 8 hours of pressure-targeted resuscitation, BE in animals that were initially resuscitated to a target sBP of 90 or 80 mm Hg returned to baseline levels and was similar to BE in animals in the sham group. However, animals that were initially resuscitated to a target sBP of 65 mm Hg persisted with a BD. In animals that survived 8 hours of initial resuscitation to a target sBP of 65 mm Hg, BE returned to baseline after 16 hours of normotensive resuscitation (a total of 24 hours of resuscitation: 8 hours of resuscitation to a target sBP of 65 mm Hg and then 16 hours of resuscitation to a target sBP of 90 mm Hg; Figure 4).
StO2 decreased in all animals that underwent hemorrhagic shock. After 8 hours of resuscitation, StO2 in animals that were initially resuscitated to a target sBP of 90 mm Hg or 80 mm Hg returned to baseline levels and was similar to StO2 in animals in the sham group. However, the decrease in StO2 persisted in animals that were initially resuscitated to a target sBP of 65 mm Hg. In animals randomized to an initial target sBP of 65 mm Hg that survived the initial 8-hour resuscitation period, StO2 was similar to that in animals initially resuscitated to a target sBP of 80 or 90 mm Hg 24 hours after the initiation of resuscitation. In all resuscitated animals, StO2 24 hours after resuscitation (i.e., after 8 hours of pressure-target resuscitation and then 16 hours of normotensive resuscitation to a target sBP of 90 mm Hg) was elevated, compared with StO2 in animals in the sham group (Figure 5).
Resuscitation volumes were smaller in animals that were initially resuscitated to a target sBP of 80 mm Hg (42 mL/kg), compared with resuscitation volumes in animals resuscitated to a target sBP of 90 mm Hg (157 mL/kg, p = 0.048). Resuscitation volumes were generally smaller in animals that were initially resuscitated to a target sBP of 65 mm Hg (69 mL/kg), compared with resuscitation volumes in animals initially resuscitated to a target sBP of 90 mm Hg as well. However, this trend was not statistically significant, as one of the animals initially resuscitated to a sBP of 65 mm Hg required a large volume of fluid (246.9 mL/kg). Total resuscitation volumes were generally larger in animals that survived to the end of the experiment than in animals that died early in the protocol (Table 3).
Table 3. Total Resuscitation Volumes for Animals
90 mm Hg
80 mm Hg*
65 mm Hg
aThese animals died before the end of the experiment.
*p = 0.048 when compared to animals randomized to initial target-pressure resuscitation to a systolic blood pressure (sBP) of 90 mm Hg.
No Hem = no hemorrhage; No Res = no resuscitation.
Resuscitation Volumes (mL/kg):
In our study, the volume of hemorrhage required to reach our shock goal sBP of 48–58 mm Hg were large, and illustrate both the severity and the controlled nature of our hemorrhagic shock protocol.13 Our finding that severe hypotensive resuscitation (vs. normotensive resuscitation) led to an increased mortality rate contradicts the finding of Bickell et al.6 However, our data are consistent with the recent findings of both Kentner et al.9 and Vaid et al.10 The varied duration of hypotensive resuscitation combined with the effects of the controlled nature of hemorrhagic shock model may account for these seemingly contradictory results. Bickell et al. collected data from patients with penetrating thoracoabdominal injuries in a setting in which the actual duration of hypotensive resuscitation was consistently less than 1 hour. The duration of hypotension was 3 hours in the studies by Vaid et al. and Kentner et al. Although short-duration hypotensive resuscitation seems safe and may improve outcomes due to reduced rebleeding, longer hypotensive periods may lead to hypoperfusion, cellular injury, organ dysfunction, and potentially to an increased mortality rate. Additionally, the controlled nature of our hemorrhagic shock model likely overestimates the increase in survival in animals initially resuscitated to a sBP of 90 mm Hg. Sondeen et al.13 has shown that pigs resuscitated to a similar goal sBP in an uncontrolled hemorrhage model often rebleed, leading to increased hemodynamic embarrassment and mortality.
Our finding that initial severe hypotensive resuscitation (sBP of 65 mm Hg) led to a persistent BD is consistent with the finding of Siegel et al.14 In our study, most animals that were initially resuscitated to a target sBP of 80 mm Hg for 8 hours did not have a persistent BD. However, two pigs in our 80 mm Hg group did have a persistent BD. This discrepancy suggests a continuum of individual susceptibility to the development of a persistent BD. In our study, all animals initially resuscitated to a target sBP of 65 mm Hg for 8 hours developed a persistent BD; a smaller percentage of animals initially resuscitated to a target sBP of 80 mm Hg for 8 hours developed a persistent BD; and no animals initially resuscitated to a target sBP of 90 mm Hg developed a persistent BD. Our study reveals little regarding the pathophysiologic consequences of this persistent BD in animals that were hypotensively resuscitated. However, Siegel et al.14 were able to correlate their findings of persistent metabolic changes in animals that were hypotensively resuscitated with end-organ histopathologic changes, including renal cellular and hepatocellular injury. Their histopathologic evidence, combined with our finding of an increased mortality rate with hypotensive resuscitation, suggests pathophysiologic consequences of persistent metabolic stress in animals initially resuscitated to a target sBP of 65 mm Hg for 8 hours.
Our finding that severe hypotensive resuscitation (sBP of 65 mm Hg) led to a persistent decrease in StO2 is consistent with our finding of a persistent BD. Furthermore, it underscores the metabolic derangement of this hypodynamic state. Our StO2 data concur with those of Watters et al.,15 and we agree that StO2 reflects both the severity of hemorrhagic shock and the adequacy, or inadequacy, of resuscitation.16 Interestingly, the increase in StO2 that we saw 24 hours after resuscitation in all surviving animals that underwent hemorrhagic shock and were resuscitated (but that we did not see in sham animals) is suggestive of a hyperdynamic state. However, neither DO2 nor cardiac output were elevated in these animals. We suspect this increase in StO2 represents a combination of peripheral vasodilation and decreased oxygen utilization in peripheral tissues.
Although our data suggest a trend toward decreased total resuscitation volumes in animals randomized to initial mild hypotensive resuscitation (sBP 80 mm Hg), there are at least two confounding factors that limit the usefulness of this data set. First, despite our efforts to standardize our animal’s breed, age, size, and diet, there remained a large degree of variability in each individual animal’s hemodynamic response to hemorrhage. This variability is likely due to a combination of preprotocol hydration status and intrinsic differences specific to each animal. Second, the data from later time points in our study are susceptible to survival bias. For example, survival bias may explain the decrease in total resuscitation volumes in animals initially resuscitated to a sBP of 65 mm Hg versus animals initially resuscitated to a sBP of 80 mm Hg. Additionally, physiologic parameters such as MAP, BE, and StO2 may appear to return to baseline at later time points in the study. Those animals that have more abnormal physiologic parameters at early time points in the study tend to die earlier in the study and do not affect the mean at later time points. However, individual subject data evaluation and statistical analysis of survivors 24 hours after resuscitation reveals similar changes in MAP, BE, and StO2, suggesting that survival bias plays a minimal role in these physiologic variables.
At the least, our findings (combined with those of Vaid et al.,10 Kentner et al.,9 and Siegel et al.14) suggest that the fluid-sparing strategy of severe hypotensive resuscitation, while preventing rebleeding in some settings, over longer intervals of resuscitation results in more profound metabolic stress (e.g., BD), tissue hypoxia (e.g., StO2), and possibly an increased mortality rate. We recommend additional properly powered uncontrolled hemorrhage animal trials to determine the safe limits of hypotensive resuscitation. Until such experiments are performed, it seems prudent to continue the standard of resuscitation to an endpoint of a weakly palpable radial artery pulse in field settings where transport to definitive care will be delayed.
In our model of controlled porcine hemorrhagic shock, severe hypotensive resuscitation led to increased mortality, persistent BD, and decreased StO2, suggesting hypoperfusion, persistent metabolic stress, and tissue hypoxia. Mild hypotensive resuscitation did not lead to persistent BD or to decreased StO2, suggesting less metabolic stress and less tissue hypoxia. The safe duration of hypotensive resuscitation is not clear. Until further information is available, it seems prudent to continue resuscitation of hypotensive trauma patients to the endpoint of a weakly palpable radial artery pulse in field settings where transport to definitive therapy will be delayed.
The authors thank Nancy Witowski for her support with data analysis. Additionally, they thank Anne Modrzynski, Natalie Aldrich, Josh Blomberg, and Patei Iyegha for their integral work managing the animals during the experiment. The authors acknowledge the assistance of John B. Holcomb, MD, COL, USA, for his thoughtful review of the manuscript.