Incidence of bleeding found by angiography according to age. Note the stark difference in incidence of bleeding observed in the 2 older age groups.
Kimbrell BJ, Velmahos GC, Chan LS, Demetriades D. Angiographic Embolization for Pelvic Fractures in Older Patients. Arch Surg. 2004;139(7):728-733. doi:10.1001/archsurg.139.7.728
Old age predicts reliably the presence of pelvic bleeding, requiring angiographic embolization (AE) among blunt trauma patients with major pelvic fractures.
Four-year prospective observational study (April 1, 1999, to May 31, 2003).
Academic level I trauma center practicing AE liberally.
Regardless of hemodynamic stability or the absence of a blood transfusion, patients with major pelvic fractures or significant pelvic hematomas on computed tomography were offered pelvic angiography with the intent to embolize.
Main Outcome Measure
Angiographically confirmed pelvic bleeding, resulting in AE.
Of 92 patients who underwent pelvic angiography, 55 (60%) had bleeding found on angiography and underwent AE. Patients 60 years and older had a higher likelihood than younger patients to have bleeding identified and to undergo AE (16 [94%] of 17 patients vs 39 [52%] of 75 patients; P<.001). An age of 60 years or older was the only independent predictor of the need for AE. Of patients in this age group, two thirds had normal vital signs on hospital admission. Bleeding was controlled by AE in all patients (100% efficacy).
Among blunt trauma patients with significant pelvic fractures, those 60 years and older have a high likelihood of active retroperitoneal bleeding. Angiographic embolization should be offered liberally to patients in this age group, regardless of presumed hemodynamic stability.
Pelvic fractures represent a major challenge to the trauma surgeon because of their close association with intraperitoneal organ injuries and significant retroperitoneal bleeding.1 High morbidity (40%-50%) and mortality (5%-30%) rates are caused in part by the inability to control pelvic retroperitoneal bleeding surgically. Angiographic embolization (AE) has become the main method to manage bleeding from pelvic fractures.2- 12
Despite the proved effectiveness of AE,6,11 its indications remain unclear. For this reason, angiography may be offered to more patients than those who are actively bleeding. A liberal policy of performing angiography in patients at risk for pelvic bleeding leads to unnecessary angiographic procedures. A restrictive policy results in delayed intervention. There is a need to identify those patients who will benefit from early AE and those who should not be subjected to the unnecessary risk of a negative angiographic result. In a previous study13 of patients undergoing pelvic and visceral angiography for trauma, the criteria for predicting populations at high risk for bleeding that requires AE were defined. An age older than 55 years was the most powerful predictive criterion found, increasing the odds of bleeding by 8-fold. Even so, the 95% confidence interval was wide (1.9-56.1), allowing a range of likelihood for bleeding of 64% to 94%. Such predictive power was deemed to be low and of moderate help for making clinical decisions. By focusing only on pelvic fractures and increasing the threshold of age, we attempt to define a population that will almost certainly experience significant bleeding and, therefore, benefit by an aggressive policy of early AE. We hypothesize that trauma patients 60 years or older with pelvic fractures who are sent for angiography always require AE.
Data from all blunt trauma patients with pelvic fractures who required angiography were entered into a prospective database from April 1, 1999, until May 31, 2003, at our level I urban trauma center. During this period, the same group of trauma surgeons took 24-hour in-house call for trauma. Although the ultimate decision about the need for angiography was left to the discretion of the individual attending surgeon, the group followed general guidelines. According to these guidelines, pelvic angiography was ordered if a patient (1) was hemodynamically labile in the absence of other sources of bleeding, (2) had specific patterns of pelvic fractures (vertical shear, open book, or "butterfly") even if hemodynamically stable, and (3) had a pelvic computed tomographic (CT) scan showing a large pelvic hematoma. Angiographic embolization was offered early and before any specific amount of blood was transfused. Angiographic embolization was performed after the secondary survey and a CT scan were completed. On many occasions, the angiography team was called before the CT scan was obtained, if the presence of a major pelvic fracture on a plain pelvic film combined with hemodynamic instability made the need for angiography obvious. However, the time required for the angiography team to assemble (approximately 1 hour) was used to obtain a CT scan. The techniques of AE are described elsewhere.6,11- 13 In short, we used coils for isolated bleeding sites or a gelatin foam slurry for diffuse bleeding. The entire trauma team escorted the patient to the angiography suite, performing aggressive monitoring and resuscitation throughout the procedure.
The following variables were collected: age, race, sex, initial Glasgow Coma Scale score, mechanism of injury, fracture pattern, Injury Severity Score, pelvic Abbreviated Injury Score, volume and rate of fluid given for resuscitation and blood given for transfusion, systolic blood pressure (SBP), diastolic blood pressure, heart rate, length of hospital stay, intensive care unit length of stay, mortality, morbidity, and angiographic findings and outcomes. An angiographic result was defined as negative if no bleeding was found. Bleeding control was defined based on angiographic evidence of cessation of contrast extravasation and clinical evidence of decrease in resuscitation (fluid and blood) needs.
To define the age threshold related to high risk of angiographically confirmed pelvic bleeding, we plotted the incidence of bleeding found on angiography against age. A bimodal distribution was found, but the highest incidence of bleeding was noted in patients 60 years or older (Figure 1). The 60th year was selected, and the patients were divided into 2 subgroups (those <60 years and those ≥60 years). Categorical values were compared between the 2 groups by χ2 or Fisher exact tests. Continuous variables were compared by a paired t test and a Wilcoxon rank sum 2-sample test. P<.05 was considered statistically significant.
In addition, a univariate analysis was used to compare patients with and without bleeding on angiography. Variables with P<.20 in this analysis were entered into a stepwise logistic regression analysis to identify independent predictors of bleeding.
Of 1017 blunt trauma patients with pelvic fractures, 332 had severe pelvic fractures (pelvic Abbreviated Injury Score >3) and 92 (9% of the entire population with pelvic fractures) were subjected to an angiographic examination secondary to pelvic bleeding during the study period. Of these 92 patients, 74 (80%) were involved in a motor vehicle crash and 18 (20%) fell from a height. The mean ± SD age of the patients with pelvic fractures was 37 ± 19 years (range, 1-96 years); and of the patients with pelvic fractures undergoing angiography, 42 ± 19 years (range, 5-81 years). Fifty-five (60%) of the 92 patients had angiographically confirmed bleeding and required AE. In all of them, AE controlled the bleeding, although 7 required repeat AE. Four complications related to the AE were recorded in 3 patients. Two of them had transient increases in serum creatinine levels that returned spontaneously to normal within a week. One developed a small pseudoaneurysm at the femoral artery puncture site, which was treated successfully by compression. Finally, an 80-year-old patient developed an intimal flap at the site of puncture, which required resection and replacement by an artificial graft. Mortality was 15% (14/92) in the entire group, 18% (10/55) among those undergoing AE, and 11% (4/37) among those not undergoing AE.
Of the 92 patients, 75 (82%) were younger than 60 years and 17 (18%) were 60 years or older. There were no differences between the 2 age groups with regard to Injury Severity Score, pelvic Abbreviated Injury Score, pelvic fracture pattern, SBP on hospital admission, indication for angiography, or fluid or blood requirements before angiography (Table 1). However, significantly more older patients required AE than younger patients. Remarkably, 62% (10/16) of older patients undergoing AE had "normal" vital signs (SBP >90 mm Hg and heart rate <100 beats/min) on admission compared with 41% (16/39) of younger patients undergoing AE (P = .23).
Only age and SBP were different between patients who had bleeding and underwent AE and those who did not (Table 2). On stepwise logistic regression analysis, an age of 60 years or older was the only independent predictor of the need for AE by increasing its odds by 15 times.
Angiographic embolization is the optimal method for controlling retroperitoneal bleeding from pelvic fractures. However, the decision to move a critically injured patient to the angiography suite is not to be made lightly. Access to angiography varies among institutions. Monitoring and resuscitation capabilities are usually not ideal during the procedure. For these reasons, many advocate that angiography should be reserved for those patients with "hard" signs of active retroperitoneal bleeding, such as persistent hypotension or an ongoing need for a blood transfusion. Studies14,15 have reported a blood transfusion threshold of 6 U before AE is considered. These same studies have reported mortality rates up to 33%. By deferring AE until hard clinical signs of bleeding appear, the opportunity to interrupt early the physiologic insult of shock and blood transfusion may be missed.5
In contrast to such restrictive policies, another treatment philosophy represents the liberal use of angiography. In this case, AE is offered early and before significant blood transfusions or major physiologic alterations occur.16 Patients with suggestive fracture patterns, significant pelvic hematomas on CT, or "soft" signs of bleeding, such as tachycardia and a drifting hemoglobin level, are subjected to diagnostic angiography and AE, if necessary. The downside of this practice relates to the incidence of unnecessary angiographic procedures. Many of the patients transferred to the angiography suite may have stopped bleeding spontaneously. By adopting a policy of liberal angiography at our institution, the effectiveness of bleeding control increased to 95%, but the rate of unnecessary angiographic procedures ranged from 36% to 43%.11,13
Therefore, there is a fine balance between risking delayed intervention and performing many unnecessary angiographic procedures. Ideally, groups of patients at high risk of bleeding and with a low likelihood of unnecessary angiography should be identified. In a previous analysis, several independent predictors for a positive angiographic result were identified. An age older than 55 years was the most powerful, increasing the odds of finding active bleeding on angiography by 8-fold. Even so, the presence of this criterion alone did not exclude the possibility of a negative angiographic result in approximately one third of the patients. The addition of other criteria decreased the rate of a negative angiographic result but restricted the population significantly.
In this study, we further explored age as a criterion for angiography. We focused exclusively on pelvic fractures and tested different age thresholds. An age of 60 years or older was associated with a 94% likelihood of a positive angiographic result among patients who were considered for AE according to our liberal guidelines. In the multivariate analysis, an age of 60 years or older was the sole independent factor predicting a positive angiographic result.
Based on the findings described, we advocate a liberal policy of AE for blunt trauma patients 60 years and older who have a significant pelvic fracture, regardless of the presence of hemodynamic stability or the need for a blood transfusion. There are many problems with the definition of hemodynamic stability in elderly patients. For example, a normal SBP of 110 mm Hg may indicate shock in an elderly patient with a daily SBP of 160 mm Hg. A normal heart rate may simply indicate the effect of β-blockade or the inability of the heart to mount a tachycardic response. Previous reports16 have shown that up to 64% of older patients who required intensive care unit admission and 48% who died after trauma had a normal blood pressure and heart rate.
Our study does not have a control group and, therefore, the natural history of hemorrhage found on angiography cannot be evaluated. It is unknown, for example, if certain hemorrhages stop spontaneously without causing significant clinical deterioration. However, given the physiologic and anatomical changes associated with this age group, we believe that spontaneous bleeding cessation would be unlikely. Older patients have nonconstricting atherosclerotic vessels and periosteum that is not densely adhered to the bone.17 They frequently receive anticoagulants at the time of trauma. Coexisting diseases may affect their physiologic reserve. Therefore, we believe that the 94% likelihood of a positive angiographic result makes a liberal policy justifiable. An additional source of bias is the absence of strict criteria for angiography in our study. We have used guidelines that allowed individual decision making according to type of injury and physiologic condition. We have not examined specifically how helical CT findings (size and location of the pelvic hematoma) relate to angiographic evidence of bleeding; this is the topic of an ongoing study. Further studies need to be performed to detect additional risk factors in different populations for the purpose of decreasing unnecessary angiographic procedures while offering timely intervention to those who need it.
Gregory J. Jurkovich, MD, Seattle, Wash: The presence of a pelvic fracture and retroperitoneal bleeding in a blunt trauma patient is more than disconcerting to the surgeon. For trauma surgeons, of course, as well as for the patient, this constellation is a harbinger of a complicated, demanding, and often unsatisfying hospital course. Dr Kimbrell and his colleagues at LA County and USC Medical Center have presented a provocative and informative analysis of their experience with this patient group. They examined the outcome of 92 patients with pelvic fractures who underwent pelvic angiography for presumed arterial bleeding. The patient group was collected over 4 years. We are not provided information on the total number of pelvic fracture patients seen over this entire time period. Hence, this is not a population study but rather an observational study of a biased population.
Some background information helps put their findings into the appropriate perspective. All pelvic fractures have some degree of bleeding. The generally accepted principles of management are that arterial bleeding is best controlled with angiography and embolization, whereas venous bleeding is best managed by bone alignment and stabilization. The challenge, of course, is how to determine if the blood loss is arterial vs venous. And there have been various predictor models that have been entertained or considered. The generally accepted predictors of arterial bleeding are the following: (1) more than 4 to 6 U of blood loss related to the pelvic bleeding, (2) the more complex posterior fractures, and (3) the location and size of the hematoma correlates with arterial bleeding, as Dr Blackmore and our Harborview group presented here last year. In that study, we looked at 760 admitted patients with pelvic fractures and a CT of the pelvis. Twenty-four percent had an angiogram, and 15% of the total population was positive for arterial bleeding, making our positive angiogram rate at 63%, similar to this study.
But if you look at an entire population of patients with pelvic fracture, only about 7% to 10% will have arterial bleeding. In the current study, the authors note that patients older than 60 years of age are most likely to have a positive angiogram. They make the argument that patients 60 and older with pelvic fractures should have angiography regardless of hemodynamic status, blood requirements, or fracture pattern.
I have to question this recommendation based on the following observations. (1) The data in this study were collected only on patients who underwent angiography. Hence, we cannot extrapolate these data to subjects who did not undergo angiography. What we need to know is whether age predicts arterial injury in all patients with pelvic fracture, including those who did and those who did not undergo angiography. Unfortunately, since most of the relevant subjects are excluded from the study presented, the generalization of the results is perilous. (2) The manuscript reflects a complication rate of 3.2% in patients and 4.3% overall. This is not a benign diagnostic test. And, (3) There is no evidence that this approach improves outcome. Their reported mortality in the elderly patients was 30% and consistent with other literature. Despite these reservations, I know I will return home and look more closely at our indications for AE in elderly patients with pelvic fractures. Undoubtedly this group has done a masterful job of selecting a patient group at high risk for pelvic arterial bleeding.
I have the following questions for the authors. (1) How many patients with pelvic fractures admitted over the same time period did not undergo angiography? (2) Was an active bleeding source always the indication for embolization, or was embolization done for suspicious areas or cutoffs without evidence of active bleeding? (3) What was the timing of this AE? In other words, were all of these 92 AEs done within 4 hours or so of ER [emergency room] arrival? (4) What was the incidence of repeat AE? I believe this is an underrecognized entity and we have to be aware that occasionally initial AE fails, and repeat embolization is necessary.
Finally, let me propose the following. A 72-year-old male is in a car crash at 12:30 AM. He sustains a closed head injury and a left femur fracture. He is hemodynamically normal and stable, but is intubated with good blood gases and minimal base deficit. A CT scan of the abdomen is remarkable only for a pubic ramus fracture and a partial posterior dislocation of the left acetabulum. There is a small retroperitoneal hematoma about this acetabulum, estimated at 200 mL. The patient's hematocrit has been stable at 35% for 2 hours, and he has received no blood. Would you request an angiogram on this patient, and would you get one at 2 AM? If you answer yes to these questions, should all non–trauma center patients who are over the age of 60 with pelvic fractures be referred to your trauma center for an angiogram?
Dr Velmahos: These are very insightful comments. Before I go straight to the answers, I want to clarify that we did not recommend angiography for all pelvic fractures in patients over the age of 60 years. We used the magic word "significant" pelvic fractures, and I do understand that it is kind of vague. The common knowledge about what constitutes a significant pelvic fracture at risk for bleeding and, therefore, a need for angiography includes the 3 patterns that we showed: pubic symphysis diastasis arbitrarily set as more than 2.5 cm; posterior pelvic fractures, particularly if they indicate vertical shear; and bilateral inferior and superior pubic rami fractures, the so-called butterfly fractures. However, beyond this established pattern of fractures, we have seen more and more elderly patients bleed from a what we would call "insignificant" pelvic fracture, just simple pubic rami fractures, noncomplex posterior fractures. Our study sends a message about patients who are older than 60 years of age. Be aggressive with angiographic exploration of such patients even in the absence of hemodynamic instability or very severe pelvic fracture patterns.
To your question, how many patients with pelvic fractures we had over that period of time, I do not have these data. What were the angiographic indications for embolization? Our interventional radiology team has great experience with this injury, so we rely on them to tell us whether the angiographic image shows bleeding, even in the absence of active extravasation. They do use indirect signs of bleeding, such as vessel cutoffs, pseudoaneurysms, abnormal tortuosity, and contrast paddling.
How many patients had repeat embolization? About 10% of the patients need repeat embolization. Seven patients in our study needed repeat embolization, and this is something that we should all remember: if one embolization fails, you have not burned any bridge yet. You have not lost the opportunity to embolize the patient again if the bleeding continues.
Finally, your last and most difficult question, what would I do with a 72-year-old man with a head injury and a femur fracture, a posterior acetabular dislocation, and a pubic ramus fracture, who is hemodynamically stable and with a stable hematocrit? Would I embolize this patient? I think that more and more I tend to ask for an angiography on such patients based on the facts that we just presented. Our experience has shown that many of these patients show just a small hematoma on the initial CT scan but then proceed to become hemodynamically unstable, require blood transfusions, and show evidence of pelvic bleeding. I tend to be more aggressive and preemptive on these older patients. We showed a lot of data but some things rely on the art of trauma surgery rather than hard evidence. The art is to not expect the whole clinical picture to blow up before you get the signal to intervene. The art is to predict what is going to happen and prevent the deterioration before it occurs. This is what we tried to do here.
John A. Weigelt, MD, Milwaukee, Wis: I have 2 questions. First, in your analysis, could you give us any information on comorbidities and especially drug use that would affect the coagulation profile of the elderly patient who may be taking any number of drugs that can affect their coagulation? Second, your use of embolization. Is it always selective, or are you also using nonselective, and when you use nonselective, is that done in collaboration with the orthopedic surgeons since nonselective embolization may affect their surgical approach to pelvic fracture stabilization?
Dr Velmahos: These are important questions. With regard to the comorbidity in these elderly patients, in our population for some strange reason we do not see too many elderly trauma patients on anticoagulation. Overall, about one third of our elderly population uses medication that could affect the clinical presentation. More than anticoagulants, I would emphasize that they use β-blockade, which may alter the hemodynamic signs. This is a very important point because the older person who comes with a blood pressure of 120 mm Hg can easily be hemodynamically unstable. A blood pressure of 120 mm Hg in a person who runs daily life at 170 mm Hg is a 50–mm Hg drop and, therefore, equal to 70 mm Hg in a person who runs daily life at 120 mm Hg.
With regard to selective or nonselective embolization: when we have one discrete point of bleeding, we use a coil. But more and more it seems that these are not just isolated points of bleed, but rather diffuse bleeds. Therefore, we tend to use more and more gelatin foam. As a matter of fact, in at least 50% of such patients, we embolize both internal iliac artery injuries. We have published these results in another article and are in the process of updating this information. Bilateral internal iliac artery embolization with Gelfoam is safe and very effective.
Accepted for publication February 23, 2004.
This paper was presented at the 111th Scientific Session of the Western Surgical Association; November 12, 2003; Tucson, Ariz; and is published after peer review and revision. The discussions that follow this article are based on the originally submitted manuscript and not the revised manuscript.
Correspondence: George C. Velmahos, MD, Department of Surgery, Los Angeles County and University of Southern California Medical Center, 1200 N State St, Room 9900, Los Angeles, CA 90033 (firstname.lastname@example.org).