The mortality rates in patients with pelvic fractures range from 9% to 20%.10,14,18 but have been reported to be as high as 42% to 50% in patients who present with pelvic fracture and hemodynamic instability.6,11 In 2 published series, 1 of the most important determinants of patient outcome was the patient's state of hemodynamic stability at the time of presentation to the hospital. Mortality rates were only 3%11 and 8%6 in patients who were stable on arrival versus 42%11 and 50%6 in patients who were deemed unstable. These marked differences in mortality represent a wide spectrum of injury patterns ranging from simple, uncomplicated fractures resulting from low impact mechanisms, to complicated fractures from high energy mechanisms which often cause multisystem injury and significant hemorrhage. In some early series, 50% to 60% of deaths associated with pelvic fracture were the result of massive hemorrhage.3,14,16 In addition, patients who died from hemorrhage invariably did so within the first 24 hours of their injury.6,11 Thus, successful treatment of these patients will depend heavily on the expedient identification, localization, and control of life threatening hemorrhage and will often require a joint collaborative effort by the trauma surgeon, orthopaedist, and interventional radiologist.
At the author's institution, an aggressive multidisciplinary approach is practiced that includes early angiographic evaluation in patients who present with complicated pelvic fracture and hemodynamic instability.12 Angiographic transcatheter embolization is performed in those patients with identifiable arterial injuries. Using this approach, it has previously been shown that 95% of patients who underwent angiogram had identifiable arterial bleeding sites and that successful embolization with subsequent control of hemorrhage was achieved in 87% of these patients.10 These findings suggest that pelvic arterial injuries play an important role when life threatening hemorrhage complicates pelvic fractures. Margolies et al8 reported 3 patients with obturator artery injuries from pubic rami fractures and Quinby15 showed that avulsion of the superior gluteal artery is often seen in children with disruption of the posterior pelvis further suggesting that different arterial injuries may be associated with different fracture patterns.
However, despite the increasing use of angiography and percutaneous transcatheter embolization for pelvic bleeding during the past decade, there is little in the literature that defines the precise nature of these arterial injuries. Nor has there been any previous attempt to correlate these injuries to anatomy of the pelvic fracture.
The present study was performed to evaluate further the author's experience with angiography and therapeutic embolization in patients with complicated pelvic fractures. More specifically, an attempt was made to characterize angiographically the source of pelvic bleeding in hemodynamically unstable patients with pelvic fractures, and to determine if specific vessel injuries could be correlated to pelvic fracture anatomy, mechanism of injury, hemodynamic stability, and patient survival.
MATERIALS AND METHODS
The angiographic findings, pelvic radiographs, and hemodynamic and resuscitative data of patients who presented with extensive pelvic fractures and hemodynamic instability or large retroperitoneal hematoma to the Kings County Trauma Center between January 1979 and August 1995 were retrospectively reviewed. Complete records and data were available for 39 patients. All patients were initially treated according to a defined protocol with established criteria for angiography and represent only a subpopulation of more than 500 patients with pelvic fractures presenting to the institution during that time.
Detailed description of the treatment algorithm for the acute treatment of the unstable patient with pelvic fracture from blunt trauma has been previously described.10 In brief, the initial workup included portable cervical spine, chest, and pelvic radiographs at the time of initial resuscitation followed by urgent open supraumbilical diagnostic tap and lavage to assess for intraabdominal bleeding and the need for immediate laparotomy. Angiography was performed in patients with pelvic fractures requiring 4 or more units of blood transfusion within 24 hours, 6 or more units within 48 hours, unstable patients with negative or borderline tap and lavage, or postoperatively in patients with significant pelvic retroperitoneal hematoma discovered at the time of celiotomy.
Angiographic catheterization was performed through the right groin except when the area had lacerations or a degloving injury at the puncture site. When massive hematoma prevented palpation of the femoral pulse, the arterial puncture was made with fluoroscopic guidance using the middle third of the acetabular roof as a landmark for the Seldinger needle or, with the assistance a needle Doppler flow probe (the Smart Needle, Peripheral Systems Group, Mountain View, CA). After guideview placement, an arterial catheter sheath was introduced and sutured to the skin. The side port of the sheath allowed continuous arterial pressure monitoring, and provided easy and safe exchange of varying size catheters.
Analog cut film was used in the earlier angiograms but during the course of the study, digital subtraction equipment replaced the older angiography suites. Digital subtraction is used exclusively with no deterioration in diagnostic information and in addition, has considerably shortened the examination time.
An aortic pigtail catheter was introduced first and the catheter placed above the origin of the inferior mesenteric artery. Filming was done to include the region of the fifth lumbar vertebra and the ischial tuberosities. The aortogram was examined for massive extravasation, large vessel injuries, or vessel spasm. If the thoracic aorta and abdominal viscera also required evaluation, the pigtail catheter was appropriately positioned and the studies performed. Selective hypogastric arteriography was then performed if the aortogram did not show evidence of bleeding. To perform selective catheterization of both the ipsilateral and contralateral hypogastric arteries and their branches, the pigtail catheter was exchanged for a curved end hole catheter.
Embolization was performed for obvious sites of extravasation and to permanently occlude injured vessels that were thrombosed. All particulate embolization in this series was performed using gelfoam cakes soaked in contrast medium. Cubes ranging from 1 to 3 cc were injected with a tuberculin syringe in increments of 0.1 to 0.2 cc until flow was shown to slow fluoroscopically and before reflux occurred. Steel/Dacron mini and microcoils were used for large vessel hemorrhage and for coil blockade, which consisted of selective placement of coils in intact vessels distal to an injury before embolization of the injured vessel with gelfoam. Pelvic angiogram was routinely repeated after embolization to show complete control of hemorrhage.
The angiographic findings were recorded and categorized as either isolated vessel injuries or multiple vessel injuries. All arterial vessel injuries were evaluated for their location, type of injury, and their relationship to the fracture site. All attempts were made to be as specific as possible in localizing the site of injury. In an attempt to investigate the relationship between the specific vessels injured and the resultant blood loss, a vessel score was designed. Each bleeding site was graded by the size of the vessel injured (Table 1). The vessel score for each patient was obtained by summing the scores for each bleeding site.
All pelvic radiographs were reviewed by the same attending orthopaedic trauma surgeon who was kept blinded to the angiographic and hemodynamic data. Pelvic fractures were graded and classified according to the Young and Burgess20 and the Bucholz4 systems for pelvic fractures. Additionally, the location of the posterior injuries was recorded. Posterior pelvic ring injuries were documented as either sacroiliac subluxation, sacroiliac dislocation, sacroiliac fracture dislocation, sacral fracture, or iliac fracture.
Patient medical records were reviewed with respect to age, gender, mechanism of injury, survival, and the presence of associated injuries. Injury severity scores were calculated. Patterns of organ injury were determined on the basis of admission examination, radiographic diagnosis, and operative findings or postmortem findings, if performed. Bladder and urethral injuries were identified by cystogram and urethrogram, respectively.
For the purpose of this study, hemodynamic instability was defined as a systolic blood pressure less than or equal to 90 mm Hg. Vital signs were recorded from within the field, on arrival to the institution, and at hourly intervals for the first 24 hours or until hemodynamic stability was thought to be achieved by the attending trauma surgeon. Blood transfusion requirements were also recorded for the immediate resuscitative period and until hemodynamic stability was achieved.
Statistical comparison of each parameter was performed with respect to its relationship to survival, mechanism of injury, angiographic findings, vessel score, associated injuries, and pelvic fracture classification. All data were tabulated using the nomenclature of the mean ± the standard deviation. Chi square analysis and logistic regressions were performed where appropriate.
The study population consisted of 26 males and 13 females ranging from 10 to 80 years of age (mean, 35 years). Of the 39 patients, 10 patients were struck by automobiles, 15 patients fell from heights, 4 were involved in motor vehicle accidents, 5 were involved in motorcycle accidents, and 5 patients suffered crush injuries; 4 were crushed between vehicles and 1 was crushed by a falling pipe. Mean injury severity score was 38 ± 16 with a range from 12 to 91.
Twenty seven patients (69%) were hypotensive with systolic blood pressures less than or equal to 90 mm Hg in the field or on arrival to the trauma center. Seven of the patients had an indication for emergent laparotomy and went directly to surgery. Two patients underwent splenectomy, 1 underwent hepatorrhaphy, 1 underwent oophorectomy for an avulsed ovarian artery, 1 had repair of an intraperitoneal bladder rupture, 1 underwent a revascularization procedure for a lacerated femoral artery, and 1 underwent laparotomy that revealed bleeding from the pelvis through a rent in the retroperitoneum. All 7 patients subsequently had an angiogram performed postoperatively either for continued hemodynamic instability (n = 3) or for a large retroperitoneal hematoma found at laparotomy (n = 4). The remaining 17 patients who were hypotensive underwent angiogram directly. Of these, 2 patients required emergent surgery after the angiogram: 1 patient had a splenorrhaphy, and 1 patient had external fixation of his pelvic fracture for continued ongoing blood losses.
Twenty-five patients survived giving an overall mortality rate of 37%. Twelve of the 14 patients who did not survive were hypotensive on presentation to the emergency room. Two patients who were normotensive died as a result of irreversible head injuries. The presence of hypotension on admission significantly increased mortality. Forty-four percent of the patients who were hypotensive (12/27) died, whereas only 17% (2/12) of the patients who were normotensive died (p < 0.05). The cause of death in the hypotensive group was attributed to continued blood loss in 3 patients, head injury in 2, and multisystem organ failure or sepsis in 3. One elderly patient died acutely from a myocardial infarction on the second day of injury. Two patients died suddenly several weeks postinjury from nonspecified causes and a recurrent retroperitoneal hemorrhage developed in 1 patient 10 days after injury when the patient was started on heparin for a pulmonary embolism. Of note, the 3 patients who died from acute hemorrhage did so within the first 24 hours of injury (Table 2).
There were 15 lateral compression (LC) type fracture patterns, 7 anteroposterior compression (APC) type, and 7 vertical shear (VS) patterns. The distribution and grading of these fracture patterns along with their associated mortality is shown in Table 3.
Angiography was completed in all but 1 patient who died from massive hemorrhage in the angiography suite. Of the 38 patients who completed angiogram, all but 3 had pelvic arterial injuries. These 3 patients, however, were found to be bleeding from splenic lacerations during the procedure and underwent successful transcatheter embolization of their spleens. Fifteen patients (39%) had an isolated pelvic arterial injury and 20 patients (53%) were found to have 2 or more arterial vessels injured. In several of the patients with multiple bleeding sites (n = 10), the bleeding sites were bilateral. It was also not uncommon to find a bleeding lumbar artery in addition to a pelvic bleeder in patients who had vertebral fractures accompanying their pelvic fracture. In the current series, 6 patients sustained lumbar vertebral body fractures (n = 3) or transverse process fractures (n = 3) from a fall from a height. All 6 patients were found to be bleeding from a lumbar artery at the level of their vertebral fracture; 3 from the third lumbar artery, 2 from the fourth, and 1 from the second. In all 6, bleeding was controlled by embolization. Although most of the injuries presented as free extravasation of contrast on angiogram, 1 patient was found on angiogram to have an arteriovenous fistula and 2 were found to have pseudoaneurysms. Eight patients had an injury to the internal iliac artery and 1 had an injury at the level of the common iliac artery. It was far more common, however, to find extravasation from 1 or more of the branches of the internal iliac artery than from the internal iliac artery itself. As shown in Table 4, the 2 most frequently injured branches of the internal iliac artery were the pudendal artery and the superior gluteal artery.
All patients had a posterior component to their pelvic fracture. Table 5 shows the frequency of posterior vascular injuries as it related to the stability of their fracture. Sixty-seven percent of patients who were deemed to have an unstable posterior component to their pelvic fracture had an accompanying posterior vascular injury compared with only 33% in whom the posterior pelvic fracture was deemed stable. Table 6 shows the incidence of anterior vascular injuries as they relate to the pelvic fractures categorized by mechanism of injury. Lateral compression forces, which are associated with anterior pubic rami fractures, had a 60% incidence of injury to an anterior branch of the internal iliac artery whereas the other fracture patterns not associated with fractures of the pubic rami had an incidence of 35%. Of note, there was no difference in the frequency of pudendal or obturator artery injuries in patients with lateral compression fractures. There were no obturator artery injuries found in patients who had pelvic fracture patterns other than a lateral compression type. The overall incidence of pudendal injuries in this group was 35%.
The average number of blood transfusions for the study population was 12 ± 11 units. One patient who succumbed to ongoing hemorrhage received a total of 62 units. Twenty-three percent (8/39) of the patients received 4 or less units of blood up to the time of achieving hemodynamic stability and an additional 33.3% (13/39) received between 5 and 10 units of total transfusion. Twelve patients (31%) had transfusions between 12 and 20 units and the remaining 5 patients (12.8%) received more than 20 units of blood (a maximum of 36 excluding the patient with 62 units of blood). There was no correlation between the number of total units of blood transfused to the pelvic fracture type or to the vascular injury. When vascular injuries were graded according to the system described in Table 1, no correlation was found between the vessel score and the number of units transfused. Likewise, there was no correlation between fracture type and vessel score.
As expected, injury severity scores correlated indirectly with survival (p = 0.0003). There was no correlation, however, between survival and total blood loss or vessel score. When the total number of units of blood transfused was combined with the injury severity scores, there was a direct correlation between these variables (p =.0002).
Rapid control of retroperitoneal hemorrhage in patients with unstable pelvic fractures is essential to prevent early death, reduce the complications associated with massive transfusions, and to improve overall outcome. Previous attempts to control retroperitoneal bleeding by surgical ligation of the internal iliac artery has proven to be futile because of the extensive collateral circulation within the pelvis.3,5 There has been increasing acceptance of the use of angiography with transcatheter embolization in the treatment of pelvic bleeding and recently, several studies have shown its efficacy under these circumstances.2,8-12,17 The present study was performed to better define the angiographic abnormalities in patients with unstable pelvic fractures and to determine how and if these findings correlate to fracture pattern and hemodynamic stability. For the purpose of this study, only a selective group of patients was evaluated who met predetermined criteria to undergo angiogram.
In the 38 patients in whom the angiogram was completed, all but 3 patients had pelvic arterial vessel injuries. As one would expect, these injuries involved the internal iliac artery and/or 1 of its branches. It was not surprising that patients with unstable posterior pelvic fractures had a high incidence of injury to a posterior branch of the iliac artery (67%) because these fracture patterns are known to be associated with a higher rate of bleeding. Of interest, however, was that ⅓ of patients who were deemed to have a stable posterior component to their fracture, and, hence, less likely to be accompanied by major hemorrhage, were also found to have an injured posterior vascular injury. Similarly, ⅔ (60%) of patients with a pubic rami fracture had an injury to an anterior branch of the iliac artery but an additional 35% of the patients without a pubic rami fracture were also found to have an injury to an anterior vessel. In addition, the presence of an anterior vessel injury did not preclude the presence of a posterior vascular injury. Only 20% of patients studied were found to have a single pelvic arterial bleeding site on angiogram. Fifty-three percent of patients had at least 2 bleeding sites. Furthermore, in those patients with multiple arterial injuries, these injuries were often bilateral and were not contained to the side of the fracture.
In this study, the most frequently injured arterial vessel associated with a posterior fracture was the superior gluteal artery, which is consistent with the findings of Quinby15 who reported that avulsion of the superior gluteal artery was frequently seen in children with posterior pelvic disruption.14 As reported by Margolies et al,9 the present study also found a large number of obturator artery injuries in patients with pelvic rami fractures yet, in the present study the most frequently injured vessel accompanying this fracture was the pudendal artery.
The mortality rate of 37% in the present study is somewhat high compared with some reports in the literature. However, this mortality rate probably represents the highly selective nature of the study population. The study group represents only a subpopulation of patients with pelvic fracture who met specific criteria of hemodynamic instability or increased transfusion requirements to undergo angiogram. This is further shown by the large number of patients studied (27/39) who were found to be hypotensive on arrival to the trauma center. The significantly higher mortality rate in the patients presenting with hypotension (44%) compared with those who did not (17%), is similar to the findings of Gilliland et al6 and Mucha and Farnell11 who reported mortality rates of 50% and 42% versus 8% and 3% respectively in a similar patient population. The correlation between survival and injury severity scores suggests that once pelvic bleeding is controlled, ultimate outcome is determined by the presence of associated injuries.
The total transfusion requirements in the present patient population was very variable and did not correlate well to survival or to any specific pelvic fracture pattern. It was common, however, for most of the patients studied to have 1 or more other injuries, such as long bone fractures, intraabdominal solid organ injuries, and/or hemothoraces, that may have contributed to the variability in the amount of transfused blood seen among patients with similar pelvic fracture patterns. Given these variables and the diversity within the pelvic fracture classification used, one would need a significantly larger study population to stratify these factors and determine with greater confidence whether a correlation between these variables does or does not exist. In a larger series, the vessel injury score as described may also prove more accurate, particularly in patients with isolated pelvic trauma.
The average number of units transfused was 11 with 56% of all patients receiving 10 units of blood or less and 9 patients receiving less than 4 units. Success of embolization was shown by the absence of further extravasation from the bleeding vessel at the completion of angiography. Fifty-eight percent of patients who were hypotensive or arrival, had normalization of their vital signs after completion of angiography. Eighteen patients (47%) required no further transfusion postprocedure and the few who did, received only 2 to 4 units to correct a low hematocrit or during stabilization of their fractures. In the 3 patients who did not have pelvic arterial bleeding on angiogram, the procedure was still useful in identifying splenic hemorrhage and, in each, splenic hemorrhage was successfully controlled by angiographic embolization.
Thus, it is not uncommon for the patient with a pelvic fracture to have 1 or more arterial sources contributing to their total blood loss. If outcome is to be improved, it is essential that all bleeding sources be identified and controlled in an expedient manner. Pelvic bleeding is multifactorial and potential bleeding sites include fracture edges, veins, and arteries. One of the limitations of this study is the lack of use of external fixation. These patients were treated before the institution of the current protocol. Of the 39 patients, 19 would receive an external fixator by the current standards. Only 3 of the patients were treated with external fixation. One of the patients was never unstable and 2 were unstable on presentation. Of the 2 hemodynamically unstable patients, 1 remained unstable after angiographic treatment of a pudendal artery bleed and stabilized only after the placement of an external fixator in the operating room. The other stabilized to a systolic blood pressure of 90 after external fixation, oophorectomy, and repair of a liver laceration. Continued blood requirements were quelled only after effective embolization of superior gluteal and internal iliac branch bleeding sites. The current protocol includes emergent external pelvic fixation in the emergency room for all APC2, APC3, LC3, and VS injuries. This is followed by angiography if, after other bleeding sites are controlled, the patient is unstable or continues to require transfusions.
It is thought that for selected patients angiography is both diagnostic and therapeutic. This technique offers the advantage of localizing multiple bleeding sites and occluding them at their level of injury. It also has the advantage of leaving the retroperitoneum undisturbed thereby reducing the risk of releasing the tamponade and the risk of infecting pelvic hematoma. Early control of hemorrhage reduces transfusion requirements and its associated complications and, in addition, allows expeditious treatment of the patients other injuries. It is thought that angiography and transcatheter embolization continues to play an important role in the multidisciplinary approach to the patient who is unstable and who has pelvic fractures.
1. Agnew SG: Hemodynamically unstable pelvic fractures. Orthop Clin North Am 25:715-721, 1994.
2. Ayella RJ, DuPriest RW, Khaneja SC, et al: Transcatheter embolization of autologous clot in the management of bleeding associated with fractures of the pelvis. Surg Gynecol Obstet 147:849-852, 1978.
3. Braunstein PW, Skudder PA, McCarroll JR: Concealed hemorrhage due to pelvic fracture. J Trauma 4:832-838, 1964.
4. Bucholz R: The pathologic anatomy of Malgaigne fracture dislocations of the pelvis. J Bone Joint Surg 63A:400-404, 1981.
5. Chait A, Moltz A, Nelson JH: The collateral arterial circulation in the pelvis: An angiographic study. AJR 102:392-400, 1968.
6. Gilliland MD, Ward RE, Barton RM, Miller PW, Duke JH: Factors affecting mortality in pelvic fractures. J Trauma 22:691-693, 1982.
7. Hauser CW, Perry JF: Control of massive hemorrhage from pelvic fractures by hypogastric artery ligation. Surg Gynecol Obstet 121:313-315, 1965.
8. Mansour MA, Moore FA, Moore EE: Hypogastric arterial embolization in pelvic fracture hemorrhage: Case report. J Trauma 30:1417-1418, 1990.
9. Margolies MN, Ring EJ, Waltman AG, Kerr WS, Baum S: Arteriography in the management of hemorrhage from pelvic fractures. N Engl J Med 287:317-321, 1972.
10. Moreno C, Moore EE, Rosenberger A, Cleveland HC: Hemorrhage associated with major pelvic fracture: A multispecialty challenge. J Trauma 26:987-994, 1986.
11. Mucha P, Farnell MB: Analysis of pelvic fracture management. J Trauma 24:379-386, 1984.
12. Panetta T, Sclafani SJ, Goldstein AS, Phillips TF, Shaftan GW: Percutaneous transcatheter embolization for massive bleeding from pelvic fractures. J Trauma 25:1021-1029, 1985.
13. Peltier LF: Complications associated with fractures of the pelvis. J Bone Joint Surg 47A:1060-1069, 1965.
14. Perry JF, McCleelan RJ: Autopsy findings in 127 patients following fatal traffic accidents. Surg Gynecol Obstet 119:586-590, 1964.
15. Quinby WC: Fractures of the pelvis and associated injuries in children. J Pediatr Surg 1:353-364, 1966.
16. Reimer BL, Butterfield SL, Diamond DL, et al: Acute mortality associated with injuries to the pelvic ring: The role of early patient mobilization and external fixation. J Trauma 35:671-677, 1993.
17. Reyolds BM, Balsano NA, Reynolds FX: Pelvic fractures. J Trauma 13:1011-1014, 1973.
18. Rothenberg DA, Fischer RP, Perry Jr JF: Major vascular injuries secondary to pelvic fractures: An unsolved clinical problem. Am J Surg 136:660-662, 1978.
19. Stock JR, Harris WH, Athanasoulis CA: The role of diagnostic and therapeutic angiography in trauma to the pelvis. Clin Orthop 151:31-40, 1980.
20. Young JWR, Burgess AR, Brumback RJ, et al: Pelvic fractures; Value of plain radiography in early assessment and management. Musculoskel Radiol 160:445-451, 1986.