Modified Rapid Deployment Hemostat Bandage Reduces Blood Loss and Mortality in Coagulopathic Pigs with Severe Liver Injury : Journal of Trauma and Acute Care Surgery

Secondary Logo

Journal Logo


Modified Rapid Deployment Hemostat Bandage Reduces Blood Loss and Mortality in Coagulopathic Pigs with Severe Liver Injury

Jewelewicz, Dory D. MD; Cohn, Stephen M. MD, FACS; Crookes, B. A. MD; Proctor, K. G. PhD

Author Information
The Journal of Trauma: Injury, Infection, and Critical Care 55(2):p 275-281, August 2003. | DOI: 10.1097/01.TA.0000079375.69610.89
  • Free


Liver injury is common after blunt trauma. Most of these injuries are managed conservatively and do not require operative intervention. However, mortality in the higher grade injuries requiring surgery (grades V and VI) ranges from 50% to 100%. 1,2 Exsanguination is the primary cause of death in these patients. Hemostasis in these patients is hindered by coagulopathy, which results from hypothermia, dilution of clotting factors, thrombocytopenia, and acidosis. Topical hemostatic agents are frequently used as adjuncts in the control of intraoperative bleeding. Although a number of different materials in a variety of forms have been used with fair success, each has potential drawbacks, making a more effective topical hemostatic agent desirable. Currently available products include oxidized cellulose (Surgicel), absorbable gelatin sponge (Gelfoam), microfibrillar collagen powder (Avitene), fibrin sealant (Tisseel), fibrin sealant dressing (TachoComb), fibrin glue (Bolheal), and topical thrombin. No single product has emerged as dominant.

A new polysaccharide product, fully acetylated poly-N-acetyl glucosamine (p-GlcNAc) (Marine Polymer Technologies, Danvers, MA) has been recently developed. P-GlcNAc is a polysaccharide polymer produced by a fermentation process and isolated from controlled, aseptic, microalgal cultures grown on a defined culture medium. 3 Gel formulations of this substance have been evaluated as hemostatic agents and found to be effective. 4,5 The membrane formulation of p-GlcNAc, applied to one surface of a standard gauze bandage, has been termed the Rapid Deployment Hemostat (RDH) bandage.

The mechanism of action of p-GlcNAc is thought to work outside of the coagulation cascade. In this respect, it should maintain efficacy in the setting of coagulopathy. Patients with severe liver injuries often present with or quickly develop hypothermia and coagulopathy, complicating hemostasis. We hypothesized that the addition of the RDH bandage to gauze packing for hemorrhage control, despite a coagulopathy, would result in reduced blood loss compared with standard treatment with gauze packing alone. This study was performed to evaluate the effectiveness of the RDH bandage for hemorrhage control after severe liver injury in the presence of a hypothermic and dilutional coagulopathy.


These experimental protocols were approved by the University of Miami School of Medicine Animal Care and Use Committee and met National Institutes of Health guidelines for animal use. Two series of experiments were conducted for this study. Each involved a traumatic liver injury in the setting of a hypothermic, hemodilutional coagulopathy. Both protocols were similar up until the time of the liver injury; details of each are provided below. Data are provided as means ± SE.

Series 1

Female and male Yorkshire random-bred pigs (n = 14, 33.4 ± 1.8 kg) were fasted overnight. Animals were then sedated with intramuscular ketamine (10 mg/kg). General anesthesia was then maintained by continuous intravenous infusion of ketamine and fentanyl. Animals were orotracheally intubated and mechanically ventilated. Bilateral external jugular vein and left internal carotid artery introducers were placed via cervical cutdowns. A pulmonary artery catheter was placed via the right jugular vein. Measured parameters included heart rate, mean arterial pressure (MAP), Sao2, end-tidal carbon dioxide, hemoglobin, hematocrit, platelet count, pH, Pco2, base deficit, lactate, central venous pressure, pulmonary artery pressure, wedge pressure, Svo2, cardiac output, and core temperatures. All blood samples were obtained from the arterial line. Animals were allowed an equilibration period of 15 minutes. Over the ensuing 20 minutes, three sets of baseline laboratory values were drawn. Baseline inclusion parameters were MAP ≥ 70 mm Hg and a normal coagulation profile. Coagulation status was determined by thromboelastography. The thromboelastograph (TEG) consists of a cuvette into which blood is placed, and a pin is attached to a torsion wire and suspended in the sample. The cup oscillates, and as clot is formed, the pin becomes coupled to the motion of the cuvette. The torsion wire then generates a signal that is recorded as a tracing. Several measurements are determined by the thromboelastograph: reaction time, r, is the interval between the start of recording and initial fibrin formation; clot formation time, k, is the interval measured from the r time to a fixed level of clot firmness, the point at which the amplitude of the tracing reaches 20 mm; the alpha angle is the slope of the TEG tracing and demonstrates the rate of clot formation; and the maximum amplitude, MA, is the greatest amplitude of the TEG tracing and is a reflection of clot strength. The above factors are combined into a coagulation index (CI), which corresponds to a coagulation state.

Creation of Coagulopathy

A hypothermic, hemodilutional coagulopathy was induced in all animals. All animals underwent a 45% of estimated blood volume, isovolemic, hypothermic exchange transfusion with 6% hetastarch (Hextend) cooled to 4°C; 45% blood volume was estimated by using the following equation: 2 × body weight (kg)/3 × 45. The exchange transfusion was accomplished over three 10-minute intervals. Fifty percent of the hemorrhage proceeded during the first 10 minutes, followed by 25% of the total hemorrhage volume over each of the following 10-minute periods. Blood was removed from the arterial line and simultaneously replaced with refrigerated hetastarch through the jugular venous line. While the hemorrhage was taking place, a laparotomy was performed and ice packs were placed in the paracolic gutters to facilitate a drop in core temperature. The ice packs were removed when a desired core temperature of 33°C was achieved. Arterial blood samples were then drawn and hypocoagulopathy documented by thromboelastography.


A reproducible, severe liver injury (grade IV) was induced by the crushing and avulsing of the left, lateral hepatic lobe and its major vessels. The same surgeon produced the injury in each animal. Immediately after the injury, a Pringle maneuver was held and pressure placed against the injury surface with a laparotomy pad for 5 minutes. After 5 minutes, pressure was released and the presence of brisk bleeding was ensured. An envelope was then opened, revealing either an RDH bandage or standard gauze bandage. Although all participants were blinded up to this point, differences in texture and appearance between bandages were apparent once the envelopes were opened. The Pringle maneuver was then reinstituted and pressure held against the injury surface with the bandage for 10 minutes. After 10 minutes, the abdomen was packed with laparotomy pads, with care taken to ensure the study bandage remained against the injury surface, and the abdomen closed with towel clips. Animals were then observed for 1 hour, with blood samples drawn every 10 minutes. After 1 hour, the abdomen was opened and all packs removed. A small amount (10 mL or less) of lactated Ringer’s (LR) solution was used to dampen the bandage lying against the injury surface, which was then slowly removed, taking care not to disrupt any formed clot. The absence or presence of bleeding was then noted. Quantification of the bleeding was not attempted. Complete absence of bleeding was documented as negative, whereas any active bleeding was documented as positive.

Series 2

Female and male Yorkshire random-bred pigs (n = 10, 40.6 ± 2.1 kg) were fasted overnight. The protocol was followed as described above up to the point at which the abdomen was packed and closed. After the abdomen was closed, each animal was resuscitated with one unit of whole porcine blood and as much LR solution as required to maintain an MAP ≥ 70 mm Hg. Arterial blood samples were drawn every 10 minutes for the first hour after the liver injury or until time of death.

After 1 hour, the abdomen was opened and all laparotomy pads were removed. A small amount (10 mL or less) of LR solution was used to dampen the bandage lying against the injury surface, which was then slowly removed, taking care not to disrupt any formed clot. The abdomen was then closed and the animal followed for 2 more hours or until time of death. The animals continued to receive resuscitation as needed. Arterial blood samples were drawn every 20 minutes for the remainder of the experiment. After 2 hours or at time of death, the abdomen was opened. The presence or absence of bleeding at the injury surface was noted. The Pringle maneuver was held, ensuring cessation of bleeding, and the abdomen was wiped clean of blood using preweighed laparotomy pads. Blood loss was calculated by weighing these laparotomy pads. Animals were then killed by infusion of a concentrated potassium solution.

Statistical Analysis

Statistical analysis was performed using a commercially available statistical software package (StatView). Groups were compared by analysis of variance. Differences were considered statistically significant with a value of p < 0.05.


Documentation of Coagulopathy

Data from both series were pooled for statistical analysis. At baseline, the CI of the animals was 4.4 ± 1.2, corresponding to a normal coagulation state. After isovolemic hemodilution and hypothermia, the CI was −2.8 ± 1.48, corresponding to a hypocoagulable state (p < 0.001) (Table 1).

Table 1:

Series 1

Fourteen animals were used in this series, seven in each of the study groups. There were no statistically significant differences in baseline physiologic parameters between the groups at baseline (Table 1). There were no significant differences in hemorrhage volume, weight of resected liver, or physiologic parameters at the time of injury. Hemoglobin and hematocrit dropped by approximately 50% in both groups. Induction of the coagulopathic state did not cause any significant change in MAP, but did cause an increase in lactate that was similar in both groups (Table 3). There were no significant differences in physiologic parameters between groups at 1 hour postinjury (Table 4). Six of seven animals in the control group had active bleeding when the packing was removed at 1 hour postinjury. Only one of seven animals in the RDH group had active bleeding at this time point (p < 0.05) (Fig. 1).

Table 3:
Parameters at Time of Liver Injury*
Table 4:
Parameters at 1 H Postinjury*
Fig. 1:
Active bleeding at 1 hour postinjury (p < 0.05).

Series 2

Ten animals were used in this study, five in each of the two study groups. There were no statistically significant differences in baseline physiologic parameters between the groups at baseline (Table 5). There were no significant differences in hemorrhage volume, weight of resected liver, or physiologic parameters at the time of injury (Table 6). Induction of the coagulopathic state did not cause any significant change in MAP, but did cause an increase in lactate (similar in both groups). Hemoglobin and hematocrit dropped by approximately 50% in both groups. At 1 hour postinjury, there were significant differences in MAP and lactate between groups (Table 7).

Table 5:
Baseline Physiologic Parameters*
Table 6:
Parameters at Time of Liver Injury*
Table 7:
Parameters at 1 H Postinjury


The length of the study protocol from the time of the liver injury to the end of the experiment was 3 hours. No animals in the control group survived this entire length, resulting in a 0% survival rate at 3 hours. The mean survival time was 76 ± 20 minutes. Three of five animals died before unpacking the abdomen. In the treatment group, four of five animals survived for the length of the experiment, yielding an 80% survival rate at 3 hours (p < 0.05). The mean survival time was 174 ± 11 minutes (p < 0.01).

Blood Loss

Total blood loss in the control group was 3,740 ± 1,004 mL. Taking into account weight and survival time, this amounted to 1.19 ± 0.13 mL/kg/min. Total blood loss in the treatment group amounted to 1,514 ± 547 mL, or 0.26 ± 0.13 mL/kg/min (p < 0.001).

Resuscitation Fluids

Total resuscitation fluid in the control group was 5,420 ± 1,860 mL. Taking into account weight and survival, this corresponds to 1.57 ± 0.28 mL/kg/min. Total fluid requirement in the treatment group was 3,500 ± 1,400 mL, or 0.58 ± 0.27 mL/kg/min (p < 0.05).


In this study, two series of experiments were conducted to evaluate the hemostatic effectiveness of the RDH bandage in a porcine model of severe liver injury in the setting of hypocoagulopathy. The addition of the RDH bandage to gauze packing was compared with gauze packing alone in each series. The first series judged the ability of the bandage to achieve complete hemostasis at 1 hour after a severe liver injury. Hemostasis was defined as the complete absence of any arterial or venous bleeding. Animals did not receive fluid resuscitation in this model and blood loss was not quantified. We found that the addition of the RDH bandage to gauze packing more effectively achieved hemostasis than gauze packing alone. The second series of experiments evaluated the RDH bandage’s ability to decrease blood loss, decrease fluid requirements, and improve survival at 3 hours in a resuscitated model of severe liver injury with coagulopathy. We found that, when compared with gauze packing alone, the addition of the RDH bandage decreased blood loss, decreased fluid requirements, and improved survival.

Liver injury is common after blunt trauma. Although the trend is toward conservative, nonoperative management of the majority of these injuries, higher grade injuries still account for substantial morbidity and mortality. Rapid and effective achievement of hemostasis is vital to the survival of this patient population. Hoyt et al. documented that 82% of all early civilian trauma deaths were attributable to uncontrolled hemorrhage, and 50% of these deaths were from severe liver injuries. 6 Achieving hemostasis is made more difficult by the coagulopathy that commonly occurs in these patients. Newer and more effective methods for obtaining rapid hemostasis would be a significant addition to the surgeon’s armamentarium.

The RDH bandage is composed of a membrane formulation of a new polysaccharide product, fully acetylated p-GlcNAc, placed on a gauze backing. It has recently been approved by the U.S. Food and Drug Administration for the treatment of extremity trauma. P-GlcNAc is a pure compound, free of proteinaceous debris and contaminants. 3 It is a polymer and can be formulated into membranes, fibers, sponges, or gels that can be directly applied to wound surfaces. The membrane formulation is easy to handle and coapts to the wound surface. It is fully biodegradable and can be left in place on a bleeding surface to provide continued hemostasis after wound closure. This product is easy to use, does not require special storage conditions, and does not require premixing of reagents or the use of donated blood products.

The proposed mechanism of action of p-GlcNAc is based on in vitro and in vivo studies and suggests that it functions primarily outside of the clotting cascade. When treating patients with coagulopathies, this gives the RDH bandage an advantage over other hemostatic agents that rely on activation of the clotting cascade. Immediately on contact with blood, p-GlcNAc activates the formation of red cell aggregates at the surface of the wound dressing that is in contact with blood. These aggregates may act as nitric oxide (NO) sinks, decreasing the local concentration of NO, causing a localized vasoconstriction. In addition, the decreased concentration of NO may induce the exposed endothelium of injured arteries to secrete endothelin-1, which causes further vasoconstriction. When in contact with blood, p-GlcNAc stimulates platelet aggregation and activation, leading to secretion of thromboxane. Thromboxane adds a further stimulus for local vasoconstriction. The vasoconstriction induced by the above mechanisms causes cessation of bleeding, leading to wound closure and subsequent clot formation.

P-GlcNAc has previously been tested in animal models of hemorrhage and compared with existing hemostatic products. Cole et al. used a swine splenic incision and splenic capsular stripping hemorrhage model to compare the hemostatic effectiveness of p-GlcNAc, oxidized cellulose, and absorbable collagen. 7 They concluded that p-GlcNAc was the most effective hemostatic agent. The authors note that this model produced small wounds that are not representative of potential injuries faced in actual operating room settings. In addition, the animals were not coagulopathic. Chan et al. used a similar model to compare p-GlcNAc to fibrin sealant;8 in a second series of experiments, the animals were heparinized and p-GlcNAc was compared with absorbable collagen. Again, p-GlcNAc was found to be the most effective hemostatic agent in this study.

In the current study, a severe liver injury was induced in the setting of a hemodilutional, hypothermic coagulopathy. We compared the addition of the RDH bandage to standard gauze packing as measures to control hemorrhage. No other measures such as suture ligation, electrocautery, or other topical hemostatic agents were used. In series 1 of the experiment, all animals survived to the endpoint and were similar in terms of physiologic parameters at all time points. Six of seven animals in the control group had active bleeding present at 1 hour postinjury, whereas only one of seven animals had any evidence of bleeding in the RDH bandage group (p < 0.05). The addition of the RDH bandage was more effective than gauze packing alone in this model of uncontrolled hemorrhage. However, it is important to note that animals were not resuscitated in this series, and both groups were hypotensive at the experimental endpoint. Although it is significant that the control group continued to bleed despite this hypotension, pointing to the severity of the injury and coagulopathy, resuscitation and improvement of the MAP may have led to rebleeding in the RDH treatment group. The question of rebleeding was addressed in series 2 of this study. The addition of the RDH bandage led to decreased total blood loss, decreased fluid requirements, and improved survival. We feel this is because of the rapid achievement of hemostasis, which is sustained despite resuscitation to preinjury MAP. The continued bleeding in the control group led to further deterioration of MAP and lactate and ultimately a 100% mortality rate at 3 hours.

This animal study compares the use of a new hemostatic bandage plus gauze packing to gauze packing alone in a reproducible model of severe liver injury combined with a hemodilutional, hypothermic coagulopathy. We feel this injury model is an improvement over previous models in which p-GlcNAc agents have been evaluated, and more closely resembles what is actually seen in the operating room. However, the results must be viewed within the context of the limitations of the study. The study was performed in swine, which were anesthetized and ventilated; compensatory physiologic responses may have been altered by this condition. Anatomic differences between swine and humans may make correlation of our results to an actual operating room setting difficult; physiologic differences may potentially play a role as well. In addition, we used a relatively small sample size in both series of the experiment. The RDH bandage was tested only against gauze packing alone, so comparisons to other topical hemostatic agents could not be addressed. Most important, however, was the inability to blind the investigators to the type of bandage being applied to the injury surface. This may have introduced bias when holding pressure against the injury surface and when packing the abdomen. The fact that three of five animals in the control group died during the first hour after injury, despite abdominal packing, was a surprise to us. However, we feel this is attributable to the severity of the injury and coagulopathy, rather than a bias in favor of the RDH bandage, and lends further support to the efficacy of the RDH bandage in achieving hemostasis.

Future studies should be directed toward comparing the RDH bandage with other currently used topical hemostatic agents. In addition, the use of the RDH bandage to achieve hemostasis rapidly without abdominal packing should also be explored. If needed, this new p-GlcNAc formulation could be placed on a Vicryl or other biodegradable backing, and thus could be left in the abdomen; this may obviate the need for a second-look operation. In summary, the addition of the RDH bandage to gauze packing more rapidly achieved hemostasis, leading to decreased fluid requirements and increased survival as compared with gauze packing alone in this porcine model of severe liver injury in the setting of coagulopathy. The RDH bandage may have broad applications for rapid control of hemorrhage in coagulopathic patients.


Dr. Wendy L. Wahl (Ann Arbor, Michigan): First, I would like to congratulate Dr. Jewelewicz on an excellent presentation. Doctors Timberlake and Cherry, members and guests, the treatment of blunt liver injury in this era is mostly nonoperative. Yet, when patients with hepatic injury require operation, it is often in the face of hypothermia and coagulopathy.

To this end, Dr. Jewelewicz and colleagues approach this vexing problem looking at blunt liver injury in swine, which are made both cold and coagulopathic. In this model, they have added either just liver packing or liver packing with a modified RDH or rapid deployment hemostat bandage.

The bandage is coated with a polysaccharide of acetylated glucosamine. Poly-N-acetyl glucosamine has been studied in the treatment of gastric varices and blunt splenic injuries.

It is also U.S. Food and Drug Administration approved for application on extremity puncture sites. Pigs with a blunt liver injury treated with the bandage were more likely to stop bleeding in one set of experiments than in a second set of studies that pigs treated with the RDH bandage and lactated Ringer’s and blood resuscitation were more likely to survive a 3-hour point, require less fluid resuscitation, and have higher mean arterial pressures and lower lactate levels.

I have one comment and a few questions about this potentially miraculous product. The numbers in each group were small; do you plan on adding to your preliminary series, because you could easily have statistical differences that would go away with one or two more animals?

Also, it is hypothesized that the RDH bandage works outside the coagulation cascade, lowering nitric oxide levels with subsequent vasoconstriction and stimulating thromboxane secretion with vasoconstriction. Have you conducted studies to confirm this mechanism, either in vitro or in vivo? Is the increased mean arterial pressure from this vasoconstriction effect of no scavenging and thromboxane secretion?

Also, the experiment concluded at 3 hours. What is the natural time course of this product, how long does the vasoconstriction last, and when it wears off, will there be rebleeding?

What is the cost of the RDH bandage? Finally, have there been any allergic responses to this product, as it is a derivative of algae? I would like to thank the Association for the privilege of discussing this article.

Dr. David G. Burris (Bethesda, Maryland): Recognizing that you have to stack models like this to make them bleed sometimes, and that the data that you showed were by driving the blood pressure up, what happened to the no-resuscitation group, just out of curiosity?

What was their mortality? What was their lactate levels, their vital signs, and so forth? We recognize that more of them were bleeding in the nontreated group than in the treated group, but were they going to survive anyway?

Dr. Reuven Rabinovici (New Haven, Connecticut): Critical to your conclusion is that your injury, or your insult, is reproducible. I wonder whether you have any data—did you quantify the blood loss secondary to your injury?

Dr. Kimball I. Maull (Birmingham, Alabama): In 1995, Dr. Scott Frame, whose memorial lecture was just given, published a report in the World Journal of Surgery using absorbable mesh packing for the management of severe liver injuries in the model used in this study. I have one question for the author, and that is, does he attribute the effectiveness of this product simply to the ability of it to cause coagulation, or is there some effect from tamponade, because Frame’s study, which was also coauthored by Dr. Enderson, our current president, showed that simply tamponade without any additional coagulation agent could effect good control of liver bleeding from deep lacerations. Thank you.

Dr. Dory D. Jewelewicz (closing): Dr. Wahl and others, thank you very much for those excellent questions. In terms of the sample size, we don’t plan on adding more animals; however, there is currently a clinical trial going on at Ryder Trauma Center using the bandage.

There were several questions relating to the mechanism. It appears that the polymer, the polysaccharide itself, has a direct action on the endothelial cells, causing them to release endothelin-1, which has a direct vasoconstrictive effect. In addition, it appears that it directly activates platelets, causing thromboxane release, which is a secondary stimulus for vasoconstriction, and there is another function, which is an indirect function.

It appears that the membrane acts as a scaffold for the aggregation of red blood cells, and these red blood cells then act as a nitric oxide sink, causing a decrease in the local concentration of nitric oxide, which serves as an additional stimulus for vasoconstriction.

In terms of the time course of the vasoconstriction, I’m not sure that that information is known. However, it is reversible, and when the action is over, the vasoconstriction is reversed. However, at that time, ideally, an adequate fibrin clot has been formed.

In terms of the costs, it appears to be cost effective, although the exact cost of the bandage has not been established. It should be approximately $8 per bandage. In this study and other studies previously conducted with this polysaccharide, there does not appear to have been any allergic reactions.

There were also some questions in terms of the protocol. Looking at series 1, these are the animals that did not receive any resuscitation.

There was actually a 100% survival at 1 hour in both groups, which is interesting. However, their mean arterial pressure was approximately in the high 30s to low 40s in both groups, so it’s unlikely that they would have survived for the 3-hour time period. There was also an increase in their lactate levels, which was similar in both groups, the treatment group and the control group.

In terms of quantifying the blood loss, in the second series of animals, we had preweighed laparotomy pads, and that’s how we weighed the blood loss from them. That’s how we came up with the total blood loss. Thank you very much.


Table 2:
Baseline Parameters*


1. Pachter HL, Liang HG, Hofstetter SR. Liver and biliary tract trauma. In: Feliciano DV, Moore EE, Mattox KL, eds. Trauma. 3rd ed. Stamford, CT: Appleton & Lange; 1996: 487–523.
2. Moore EE, Cogbill TH, Jurkovich GJ, et al. Organ injury scaling: spleen and liver (1994 revision). J Trauma. 1995; 38: 323–324.
3. Vournakis J, Pariser ER, Finkielsztein S, et al., inventors; Poly-N-acetyl Glucosamine. U.S. patent 5,623,064. April 22, 1997.
4. Kulling D, Woo S, Demcheva MV, et al. Poly-N-acetyl glucosamine: will it replace cyanoacrylate for the treatment of gastric varices? A pilot study in a rabbit model. Endoscopy. 1998; 30: 41–42.
5. Kulling D, Vournakis JN, Woo S, et al. Endoscopic injection of bleeding esophageal varices with a poly-N-acetyl glucosamine gel formulation in the canine portal hypertension model. Gastrointest Endosc. 1999; 49: 764–771.
6. Hoyt DB, Bulger EM, Knudson MM, et al. Death in the operating room: an analysis of a multi-center experience. J Trauma. 1994; 37: 426–432.
7. Cole DJ, Connolly RJ, Chan MW, et al. A pilot study evaluating the efficacy of a fully acetylated poly-N-acetyl glucosamine membrane formulation as a topical hemostatic agent. Surgery. 1999; 126: 510–517.
8. Chan MW, Schwaitzberg SD, Demcheva M, et al. Comparison of poly-N-acetyl glucosamine (p-Glc-NAc) with absorbable collagen (Actifoam) and fibrin sealant (Bolheal) for achieving rapid hemostasis in a swine model of splenic hemorrhage. J Trauma. 2000; 48: 454–458.

Hemostasis; Blunt abdominal trauma; RDH bandage; Severe liver injury

© 2003 Lippincott Williams & Wilkins, Inc.