Is There a Selection Bias in Applying Endovascular Aneurysm Repair for Rupture?

Article Outline

• Abstract
• Introduction
• Methods
• Results
• Discussion
• Conclusions
• References
• Copyright

Accumulating data suggest that endovascular repair (EVAR) of ruptured abdominal aortic aneurysms (RAAAs) leads to reduced mortality, but concern exists that this may reflect selection bias. We reviewed our overall rupture experience early after our protocol was instituted to explore this question. We instituted a defined protocol for RAAA with emphasis on EVAR in July 2002, which included device availability (consignment), preoperative training, 24-hr access to our surgical endosuite and ability to operate imaging in an emergency, and immediate availability of a transbrachial balloon cutdown cart for all cases. Charts of all RAAA patients who arrived in the operating room alive since institution of our protocol were reviewed. Computed tomographic (CT) scans were re-reviewed to assess potentially suitable anatomic candidates. From July 2002 to May 2006, a total of 52 RAAAs were treated at our institution: 15 pararenal RAAAs, all treated by open repair (PR-OPEN), and 37 infrarenal RAAAs, 20 treated by open repair (IR-OPEN) and 17 treated by EVAR (IR-EVAR, 32% of all ruptures). Mortality rates in the three groups were 47%, 75%, and 35% (p < 0.02 vs. IR-OPEN), respectively. Although mortality was significantly lower in the EVAR group, overall mortality was 53% (28/52). On re-review of the operative notes and CT scans, it is estimated that more than half of those cases repaired using open techniques could have been repaired using EVAR based on anatomic criteria alone. The most common reason for open repair was hemodynamic instability preoperatively; only a minority of cases were excluded from EVAR based on unfavorable anatomy after CT scan review in the emergency room. In conclusion, during our early experience EVAR for rupture was associated with significantly reduced mortality. However, our overall mortality was no different from historical values, and this fact along with the extremely high mortality seen in the IR-OPEN group suggest that we are simply selecting patients with the greatest chance of survival to undergo EVAR. It also appears that many patients who are anatomically suitable for EVAR are undergoing open operation because of hemodynamic instability. If EVAR for rupture truly decreases mortality in all patients, a much more aggressive attitude toward EVAR may be required to lower the overall mortality rate.

Back to Article Outline

Introduction 

In 1966, Lawrence and colleagues¹ at the University of Iowa reported that the mortality rate following operative repair of ruptured abdominal aortic aneurysms (RAAAs) was 52%. Although dramatic advances have been made in understanding the etiology and natural history of aneurysmal disease, diagnosis, methods of operative repair, and intensive care unit management, the mortality rate after open surgical repair in the United States remains stubbornly unchanged, in 2006 again being reported as 52% in a large review of the past decade's experience² and 47% in a survey of results since 2000 in one-third of the U.S. population.³ RAAA ranks as the twelfth leading cause of death among adults greater than 65 years of age in the United States and is responsible for greater than 10,000 deaths per year.⁴

Elective endovascular aneurysm repair (EVAR) has been repeatedly associated with decreased morbidity and mortality compared to open repair.², ³, ⁵, ⁶, ⁷, ⁸ Accordingly, it has been hypothesized by many that endovascular repair for rupture will have similar beneficial effects. The first report of EVAR for the treatment of RAAA was published by Yusuf et al. in 1994,⁹ and over the last 12 years many U.S. investigators have reported apparently excellent results on EVAR for RAAA. The obvious question that has been repeatedly raised, however, is that of selection bias: Are we treating the same disease? For example, a recent large report from Columbia/Cornell, describing outcomes after EVAR for RAAA in an area covering approximately one-third of the United States, found that only 59% of repairs were performed the same day as admission.³ These patients, therefore, obviously differ from the typical “true” rupture, who arrives hypotensive in the emergency room and either dies or has a clamp applied to the aorta within 20 or 30 min.

We began treating ruptured aneurysms using endovascular technology approximately 5 years ago, under the auspices of a predefined multidisciplinary protocol. While we felt that we followed an aggressive decision-making process, it became apparent that the majority of our patients were still being treated with open surgery. We believe this is due to selection bias and undertook this review to answer this question.

Back to Article Outline

Methods 

We instituted a defined protocol for the treatment of RAAA with emphasis on endovascular repair beginning July 2002. This was a multidisciplinary effort instituted as part of the creation of aortic and vascular teams and included input from and education of members of the Departments of Emergency Medicine and Radiology, operating room (OR) management, anesthesia, nursing, cardiothoracic surgery, and intensive care medicine. The protocol is activated when a patient with known or suspected rupture is admitted to the emergency room or via intrahospital transfer; emphasis includes permissive hypotension with expeditious computed tomographic (CT) scanning, early notification of the radiology technician, utilization of a vascular call team (nursing, technicians, and anesthesiologists), aggressive use of regional/local anesthetic for EVAR, and a transbrachial balloon cutdown cart immediately available for the surgeon to use without nursing assistance. We have a large supply of both Zenith (Cook, Bloomington, IN) and Excluder (Gore, Flagstaff, AZ) endovascular grafts on consignment readily available to the operating team at all times. Additional advantages in our institution include a dedicated OR endosuite with a fixed Philips (Hamburg, Germany) imaging system; because our elective cases, both open and endovascular, are performed in this suite, the operating team is intimately familiar with the location of supplies, grafts, and how to position the patient and locate the supporting equipment necessary to perform the operation requested. All cases are managed and all decisions made by a vascular surgeon, and all cases are performed in our OR endosuite. In practice, the decision to perform open versus surgical repair is based on the patient's condition at the time of arrival (as determined by mental status, vital signs, and laboratory findings) as well as anatomic suitability for EVAR based on thin-cut (2 mm) CT of the patient's abdomen and pelvis.

We performed a retrospective review of patients undergoing repair of RAAA from July 2002 to January 2006. Inpatient and outpatient charts were reviewed, and CT scans of patients undergoing open repair were re-reviewed to assess suitability for EVAR, in retrospect, by an attending surgeon. Patients with aneurysms involving the origin of the renal or visceral vessels were excluded, and the remaining patients were thus classified as having a pararenal open repair (PR-OPEN), an infrarenal open repair (IR-OPEN), or an EVAR.

Patients were considered to have RAAA if they were found to have rupture at exploration or CT scan findings of hematoma or contrast extravasation outside the abdominal aortic wall. We defined an “unstable” patient as one having two or greater of the following: systolic blood pressure <120 mm Hg, heart rate ≥100, and respiratory rate ≥20.

CT scans were retrospectively reviewed by a single reviewer (K. A. I.) blinded as to the treatment actually received. Judgment as to whether EVAR was possible was made subjectively without firm criteria, with, of course, the knowledge that the patient had a rupture and was thus at much higher risk for mortality after conventional surgery.

Results were assessed by analysis of variance and chi-squared testing (with Fisher's exact testing for small cell sizes) for continuous and categorical variables, respectively. Unless otherwise noted, results are expressed as mean ± standard deviation.

Back to Article Outline

Results 

Between July 2002 and May 2005, a total of 57 patients presented with rupture. Of these, five were clearly supravisceral and were excluded from this series, leaving us with a total of 52 patients with RAAA who were conceivably candidates for EVAR to examine.

The patients were divided into three groups: 15 (29%) underwent open repair of a pararenal aneurysm (PR-OPEN), 20 (38%) underwent open repair of what were described as infrarenal aneurysms (IR-OPEN), and 17 (33%) underwent EVAR. Table I shows their initial presenting demographics. While 66% and 75% of patients undergoing PR and IR open repairs, respectively, were unstable, only 47% of those undergoing EVAR were. In addition, 11 (65%) of those undergoing EVAR were described as having a “contained” rupture versus only six (15%) and four (5%) of the PR- and IR-OPEN patients, respectively (p = 0.16 for EVAR vs. PR, p = 0.006 for EVAR vs. IR).

Table I. Presenting demographics

	PR-OPEN	IR-OPEN	EVAR	p
All patients	15	20	17
Men	14	13	10	NS
Women	1	7	7
Time from ER arrival to OR (min)	68.8 ± 61.8	95 ± 64	115 ± 115	NS
Unstable	10 (66%)	15 (75%)	8 (47%)	0.08 vs. all open
Rupture described as “contained”	6	4	11	<0.006 vs. IR-OPEN, 0.16 vs. PR-OPEN
Mean systolic blood pressure (mm Hg) on arrival to ER	120 ± 31	113 ± 37	131 ± 35	NS
Lowest systolic blood pressure (mm Hg) on arrival to ER	70	60	71
CT performed at our hospital?	0	5 (25%)	17 (100%)

Perioperative results are described in Table II. Operative time was modestly lower in the EVAR group and blood loss significantly less than in either open group. Six of the 17 patients undergoing EVAR for rupture died within 30 days or having never left the hospital, for a mortality rate of 35%. In contrast, 15 of the 20 IR-OPEN patients died during this time frame, for a mortality rate of 75% (p = 0.015). Length of stay (LOS) among all patients was less in the open surgical group compared to the EVAR group, largely due to the greater overall mortality of the open group. When only survivors were compared, LOS in patients undergoing EVAR, 10.7 ± 14.6 (range 1-44) days, was significantly shorter than that in IR-OPEN patients (25.8 ± 20, range 4-50, days; p < 0.06).

Table II. Perioperative results

If we delve deeper into mortality rates, however, several interesting observations can be made (Table III). First, as described above, the mortality rate of the EVAR group, at 35%, is significantly less than that of the IR-OPEN group (75%) but not the PR-OPEN group (47%). If mortality after all open repairs (63%) is compared to that after all EVARs (47%), EVAR results in a significant survival benefit (p < 0.06). However, if we compare mortality in the PR-OPEN group with all IR ruptures, irrespective of type of repair, there is no difference: 47% vs. 57%. Finally, our overall mortality rate—28 of 52—is 54%, a number no different from those reported in 1966¹ and 2006.³

Table III. Mortality rates detailed

Is there a selection bias? Ten of the 17 patients who eventually underwent EVAR were described in the original hospital record as having a “contained leak.” The in-hospital survival for these patients (91%) was markedly better (p = 0.002) than that for those without the designation (17%). EVAR patients (47%) were more likely to be characterized as stable than open repair patients (74%) (p = 0.06). When evaluating the mortality of patients in terms of stability, no differences in survival rates were found between groups in any category (Table III). No patient in this study treated with EVAR developed abdominal compartment syndrome, further suggesting that the severity of disease was not typical of the average patient with rupture. Finally, a retrospective review of CT scans, admittedly a very subjective task, suggested that about half of the IR-OPEN patients could conceivably have been candidates for EVAR.

Back to Article Outline

Discussion 

In our experience, EVAR of ruptured aneurysms is associated with a lower mortality rate than open repair, as well as decreased blood loss, decreased transfusion requirement, shorter operative time, and shorter LOS among survivors. However, it is apparent that we are operating on healthier patients. Patients undergoing EVAR are less likely to be hemodynamically unstable on arrival, have higher systolic blood pressures, have lower heart rates, are more likely to be described as having “contained leak” (in fact, the subcategory of patients with contained leak undergoing EVAR had the lowest mortality rate of all, at 9%), and have a zero rate of postrepair abdominal compartment syndrome. Moreover, on re-review of CT scans, it was estimated that approximately 50% of patients undergoing open repair could have received EVAR. Finally, our overall mortality during this period, 54%, is no different from the historical benchmark of 50% or our local survival rate before this protocol was initiated (53%).

This selection bias toward operative repair with EVAR in the hemodynamically stable RAAA patient is supported by the evidence of Greco et al.³ in a review of discharge data sets from four states. It was found that only 59% of RAAAs undergoing EVAR were performed the same day of admission vs. 72% with the open technique. Furthermore, the mortality for open repair when the operation was performed on the same day of admission (45.7%) was found to be significantly lower than if performed on a following day (51.7%, p = 0.001). The mortality rates were not statistically significantly different between the EVARs performed on the same day of admission and those performed on a subsequent day (38.6% vs. 33.3%, p = 0.42). The results from the same analysis show an overall in-hospital mortality of 39.3% for EVAR vs. 47.7% for open repair of RAAA (p = 0.005).

Early reports estimated that 40% of patients presenting with rupture would be amenable to EVAR;¹⁰ current estimated rates vary widely from 45% to 83% of ruptures.⁵, ⁶, ⁷, ⁸, ¹¹, ¹² A retrospective review of operative notes and CT scans of patients undergoing open repair at our institution suggests that 50% receiving open repair were anatomic candidates for EVAR. Including all patients with RAAA presenting to our emergency department, only 52% presented with anatomic findings on CT scan suitable for infrarenal EVAR. Considering only patients with infrarenal abdominal aneurysm anatomy, 75% were found to be anatomically amenable for EVAR on retrospective review.

We interpret these results as evidence of a strong selection bias by our group, favoring EVAR only in more hemodynamically stable patients with a better chance of survival regardless of technique used. In other words, we are using hemodynamic stability, not anatomic suitability, as the criterion for open versus endovascular repair. We found that we had a difficult time considering and planning EVAR of RAAA in the setting of frank hemodynamic instability, especially when a CT scan was not directly available for review. In retrospect, this attitude may be an error (and has already changed with increasing experience at our institution). Despite the immediate availability of our transbrachial balloon control cart, this technique has only been used three times. It may be that a more aggressive attitude toward using balloon control in hemodynamically unstable patients would allow a greater percentage of them to undergo EVAR. Indeed, of the 17 unstable patients with infrarenal anatomy, 11 were repaired with the open technique.

The second most common anatomic exclusion criterion in this series was a neck that was too large (i.e., a juxta- or pararenal aneurysm). When considered using criteria that are applied to conventional, elective repair, most readily available devices can treat neck diameters no larger than 32 mm or so. While this number may not change, we can “put the neck in a different place” by changing our attitude toward the renal arteries. For example, covering one or even both renal arteries may be an acceptable price to pay if we really can document a true difference between techniques. Alternatively, while EVAR in the setting of a large cuff may not be expected to be durable,¹³ it may, in the short term, take care of the immediate problem, allowing more leisurely, safer, and definitive correction. Finally, while the use of fenestrated and branched endovascular devices is in its infancy and currently time-consuming, it is anticipated that as technology improves, this option may also be of use in rupture.

We also hypothesized that a delay in obtaining a high-quality CT scan would negatively impact overall survival. In fact, when comparing times between admission to the emergency department and transfer into the OR, the time difference of open versus EVAR is found to be only 20 min longer for those undergoing EVAR (95 ± 64 open, 115 ± 115 EVAR, p = not significant). Boyle and associates¹⁴ noted that the time of death after arrival to the hospital among 21 RAAA patients electing not to undergo operation was 435 min (range 15 min-6 days). Lloyd et al.¹⁵ reported a median total time to death from onset of symptoms of 998 min (range 61 min-6 days) in 56 patients expiring due to RAAA without operative intervention. The patients selected for EVAR in our study were all taken to the OR after a CT scan was available for review (admittedly not the practice of the Montefiore group, probably those with the most experience⁷). Of the patients undergoing EVAR in our study, 58% had a CT scan of the abdomen and pelvis performed after arriving at Strong Memorial Hospital and before going to the OR, while the remaining 42% arrived with a CT scan from an outside facility of reasonable quality to select the patient for EVAR. None of the patients in this study selected to undergo EVAR for RAAA expired prior to starting the procedure in the OR. While EVAR can be performed without CT scanning, our data suggest that this delay is not a common cause of adverse outcome, whenever the eventual method of repair.

These results have led us to modify our practice in essentially four ways. First, we believe we have a tendency to make the decision too quickly, i.e., while driving in, knowing that the patient is hemodynamically unstable, the faculty member listening to the judgment of a vascular fellow or senior resident and attempting to move things along as quickly as possible. Even if the vascular fellow or senior resident is technically excellent, he or she has not accumulated the amount of experience a faculty member has. Our data do not show a significant danger in waiting for an expeditious CT scan, so we would recommend waiting until the attending surgeon has had a chance to personally review the CT for suitability before making this decision, even if the operation is delayed. Second, we advocate very aggressive use of transbrachial balloon control in the unstable patient and, if this is needed, obtaining an angiogram before declaring the patient anatomically unsuitable. Third, we are increasingly quick to call in one of our partners; we believe that two independent viewpoints offer much better decision-making capability in emergency and provide significant psychological support for “pushing the envelope” in order to save lives. Fourth, we believe that coverage of one or both renal arteries should be at least considered. While few data exist, if we can indeed double survival by fixing a ruptured aneurysm by endovascular means, renal failure in some patients may be an acceptable option.

Back to Article Outline

Conclusions 

Although EVAR of RAAAs is associated with increased survival, in our experience overall survival has not changed despite using the technique in one-third to one-half of our ruptures. We believe we are selecting healthier patients to undergo EVAR, i.e., making our decision based on hemodynamic stability rather than anatomic criteria.

It is unlikely that a true randomized trial will ever be performed to answer this question. To combat this bias, we advocate (and are instituting) more aggressive consideration of EVAR and more aggressive use of transbrachial balloon control in the unstable patient, performance of these cases by more than one attending surgeon, and a very aggressive anatomical approach, including potential renal artery coverage. We believe that the best way to lower the 50% operative mortality rate that has held constant for 40 years is to employ a more aggressive approach to the use of EVAR in the repair of RAAA, especially in patients who are hemodynamically unstable.

Back to Article Outline

References 

1. Lawrence MS, Crosby VG, Ehrenhaft JL. Ruptured abdominal aortic aneurysm. Ann Thorac Surg. 1966;2:159–165
   • View In Article
   • MEDLINE
   • CrossRef
2. Dillavou ED, Muluk SC, Makaroun MS. A decade of change in abdominal aneurysm repair in the United States: have we improved outcomes equally between men and women?. J Vasc Surg. 2006;43:230–238
   • View In Article
3. Greco G, Egorova N, Anderson PL, et al. Outcomes of endovascular treatment of ruptured abdominal aortic aneurysms. J Vasc Surg. 2006;43:479–486
   • View In Article
4. http://www.cdc.gov/nchs/dvs/LCWK2_2002.pdf.
   • View In Article
5. Alsac JM, Desgranges P, Kobeiter H, Becquemin JP. Emergency endovascular repair for ruptured abdominal aortic aneurysms: feasibility and comparison of early results with conventional open repair. Eur J Vasc Endovasc Surg. 2005;30:632–639
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (256 KB)
   • CrossRef
6. Gerassimidis TS, Papazoglu KO, Kamparoudis AG, et al. Endovascular management of ruptured abdominal aortic aneurysms: 6-year experience from a Greek center. J Vasc Surg. 2005;42:615–623
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (227 KB)
   • CrossRef
7. Veith FJ, Ohki T, Lipsitz EC, Suggs ED, Cynamon J. Treatment of ruptured abdominal aortic aneurysms with stent grafts: a new gold standard?. Semin Vasc Surg. 2003;16:171–175
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (186 KB)
   • CrossRef
8. Piffaretti G, Caronno R, Tozzi M, Lomazzi C, Rivolta N, Castelli P. Endovascular versus open repair of ruptured abdominal aortic aneurysms. Expert Rev Cardiovasc Ther. 2006;4:839–852
   • View In Article
   • CrossRef
9. Yusuf SW, Whitaker SC, Chuter TAM, Wenham PW, Hopkinson BR. Emergency endovascular repair of the leaking aortic aneurysm. Lancet. 1994;344:1645
   • View In Article
   • CrossRef
10. Reichart M, Geelkerken RH, Huisman AB, van Det RJ, de Smit P, Volker EP. Ruptured abdominal aortic aneurysm: endovascular repair is feasible in 40% of patients. Eur J Vasc Endovasc Surg. 2003;26:479–486
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (128 KB)
    • CrossRef
11. Lee WA, Hirneise CM, Tayyarah M, Huber TS, Seeger JM. Impact of endovascular repair on early outcomes of ruptured abdominal aortic aneurysms. J Vasc Surg. 2004;40:211–215
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (87 KB)
    • CrossRef
12. Brandt M, Walluscheck KP, Jahnke T, Graw K, Cremer J, Muller-Hullsbeck S. Endovascular repair of ruptured abdominal aneurysm: feasibility and impact on early outcome. J Vasc Radiol. 2005;16:1309–1312
    • View In Article
13. Mohan IV, Harris PL, Van Marrewijk CJ, Laheij RJ, How TV. Factors and forces influencing stent-graft migration after endovascular aortic aneurysm repair. J Endovasc Ther. 2002;9:748–755
    • View In Article
    • MEDLINE
    • CrossRef
14. Boyle JR, Gibbs PJ, Kruger A, Sherman CP, Raptis S, Phillips MJ. Existing delays following the presentation of ruptured abdominal aortic aneurysm allow sufficient time to assess patients for endovascular repair. Eur J Vasc Endovasc Surg. 2005;29:505–509
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (133 KB)
    • CrossRef
15. Lloyd GM, Brown MJ, Norwood MGA, Deb R, Fishwick G, Bell PRF. Feasibility of preoperative computer tomography in patients with ruptured abdominal aortic aneurysm: a time-to-death study in patients without operation. J Vasc Surg. 2004;39:788–791
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (92 KB)
    • CrossRef