Hospital Volume Impacts the Outcomes of Endovascular Repair of Thoracoabdominal Aortic Aneurysms

Clinical Research Hospital Volume Impacts the Outcomes of Endovascular Repair of Thoracoabdominal Aortic Aneurysms Satinderjit Locham, Farhan Hussain, Hanaa Dakour-Aridi, Andrew Barleben, John S. Lane, and Mahmoud Malas, San Diego, California

Background: Few centers in the United States have the expertise to manage patients with a thoracoabdominal aortic aneurysm (TAAA). The purpose of this study is to use a nationally representative vascular database to assess the role of hospital volume on outcomes in patients undergoing endovascular repair for TAAA. Methods: All patients undergoing complex endovascular repair (cEVAR) for TAAA were identified in the Vascular Quality Initiative (VQI) database (2012e2018). The total mean number of cases per year was identified at each center and were used to group into three quantiles containing an equal number of patients (Low [LVH], Medium [MVH], High [HVH]). Standard univariate and multivariable (logistic regression) analyses were performed to evaluate the patient’s characteristics and short-term outcomes. Results: A total of 2,115 patients from 118 centers (Low e 92, Medium e 19, High e 7) were identified in VQI from 2012 to 2018. The annual mean (S.D.) number of cases at HVH, MVH, LVH were 22.7 (4.7), 9.6 (3.0), 3.6 (1.4), respectively. The repair of Type III TAAA was slightly higher in HVH versus MVH versus LVH (22.5% vs. 21.0% vs. 15.1%), while Type I was more common among LVH versus MVH versus HVH (13.7% vs. 11.5% vs. 3.7%) (Both P < 0.001). Custom/modified devices were more likely to be used in HVH versus MVH versus LVH (67.9% vs. 27.6% vs. 27.2%) (P < 0.001). Additionally, HVH and MVH utilized fenestrated/ branched or chimney/snorkel options more frequently, whereas surgical bypasses were common in LVH for revascularization of visceral arteries. In univariate analysis, HVH were associated with significantly lower mortality (2.2% vs. 5.1% and 6.5%), failure to rescue [FTR] (3.5% vs. 11.6% and 12.1%) and any complication (24.6% vs. 27.1% and 31.2%) compared to LVH and MVH (All P < 0.001). After adjusting for potential confounders, both LVH and MVH were associated with 2e4 fold increase in the odds of mortality [OR (95% CI): 2.30 (1.20e4.41) and 2.14 (1.16e3.93)] and FTR [OR (95% CI): 4.42 (1.86e10.54) and 4.08 (1.73e9.62)] compared to HVH. Conclusions: Our study demonstrates significantly lower morbidity and mortality in high volume hospitals performing cEVAR for TAAA, despite operating on older patients with more

This work will be presented at the 2019 Western Society of Vascular Surgery Annual Meeting, Hawaii. S.L., M.M., contributed to the conceptual framework, research question, critical revision and design of the study. S.L., F.H., contributed to the literature review and writing the manuscript. S.L., contributed to the acquisition, analysis, or interpretation of data for the work. S.L., F.H.,A.B., M.M., contributed to the final approval of the version to be published. Conflicts of interests: None of the authors have any relevant financial disclosures.

Correspondence to: Mahmoud Malas, MD, MHS, FACcS, Professor In Residence Chief of Vascular and Endovascular Surgery, Vice Chair of Surgery for Clinical Research, University of California San Diego Health, 9300 Campus Point Drive, La Jolla, CA, USA; E-mails: [email protected] or [email protected] Ann Vasc Surg 2019; -: 1–10 https://doi.org/10.1016/j.avsg.2019.09.018 Ó 2019 Elsevier Inc. All rights reserved. Manuscript received: July 30, 2019; manuscript accepted: September 9, 2019; published online: - - -

Division of Vascular and Endovascular Surgery, Department of Surgery, University of California San Diego, San Diego, CA.

1

2 Locham et al.

Annals of Vascular Surgery

complex TAAA types. This is likely due to better rescue phenomenon in addition to more experienced operators. Complex endovascular repair of TAAA can be performed safely in high volume aortic centers of excellence.

INTRODUCTION Since the introduction of the endovascular approach in vascular surgery in the early 1990s, the use of endografts in managing aortic diseases has become widely accepted. In a systematic review, Piper et al.1 studied the relationship of hospital volume and outcomes following major surgical procedures, including abdominal aortic aneurysm (AAA), and reported lower mortality for elective and ruptured AAA repair in high-volume compared to low volume hospitals. The Society of Vascular Surgery (SVS) guidelines recommends that elective EVAR should be performed in hospitals that perform at least 10 EVAR cases per year and have mortality and conversion rates of less than 2%. However, the quality of evidence for the above recommendation is poor (Level 2, C).2 Traditionally, patients with thoracoabdominal aortic aneurysms (TAAA) receive an open repair; however, advancement in endovascular technology provides an appealing and less invasive alternative to open repair. In recent studies, our group demonstrated improved outcomes in favor of endovascular repair of TAAA.3,4 Overall hospitalization cost was also significantly lower in endovascular versus open repair, despite the increased cost of endovascular grafts. This was attributed to the higher number of complications and longer length of hospital stay in open repair.3 Very few studies in the literature have looked at the volume-outcome relationship following repair of thoracic aortic aneurysms (TAA) and TAAA. Patel et al.5 found no difference in mortality (3.9% vs. 5.5%, P ¼ 0.43) between low versus high volume hospitals undergoing thoracic endovascular aneurysm repair (TEVAR). Cowan and colleagues looked at open repair of TAAA and found a statistically significant difference in mortality between low versus high volume hospitals.6 In a more recent study from California, Weiss et al.7 found no difference in mortality between low and high volume institutes performing the open repair of TAAA. While the proportion of patients with TAAA receiving open aortic repair (OAR) has remained stable, the number of endovascular repairs has increased over time.8 However, there is no study evaluating the impact of hospital volume and outcomes following endovascular repair of TAAA.9

Thus, the purpose of this study is to use a vascular specific, nationally representative vascular database and evaluate short-term outcomes of patients undergoing endovascular repair for TAAA between different hospital volumes (low, medium, high).

METHODS Database A retrospective study of prospectively collected Vascular Quality Initiative (VQI) TEVAR and complex aortic aneurysm data from 2012 to 2018 was performed. VQI is a large international database with over 500 centers across the United States and Canada.10 It captures patient-specific data following major vascular surgical procedures on demographics, comorbidities, and perioperative characteristics (procedural variables, postoperative outcomes). More information regarding VQI can be found at www.vqi.org. The VQI research advisor committee approved this research. The Institutional Review Board approved this study, and the need for patient consent was waived because of de-identified data. Study Cohort All patients undergoing endovascular repair of thoracoabdominal aortic aneurysms (Type I-IV) were included. The different types of TAAA were identified using specific proximal and distal zones of the disease, as shown in Appendix I. Patients outside of these zones or presenting with rupture or other aortic emergencies, such as penetrating ulcer, hematoma, and aortic thrombus were excluded from the study. There were 118 centers performing the endovascular repair of TAAA. To create hospital volumes, first, the total number of cases per year per center was determined. Then the mean case number at each center was used to create three quantile volumes (Low [LVH], medium [MVH], and high [HVH]). Patient Characteristics All available patient demographics and comorbidities in the VQI database were included and studied. Obesity is defined as a body mass index greater than

Volume

-, -

2019

30 kg/m2. Some of the procedural characteristics included were contrast volume, crystalloid volume, fluoroscopy time, procedure time, intraoperative transfusion, use of IVUS/TEE, estimated blood loss, conversion to open, and use of spinal drain (none, preoperative, postoperative). Technical success was defined in VQI as successful implantation of the device without any corrective intervention, patency of the aortic endovascular stent graft, absence of device deformations, absence of inadvertent covering, and successful withdrawal of the delivery system without the need for unanticipated corrected intervention related to the withdrawal. We also identified whether visceral arteries (celiac, SMA, or renals) were occluded or required the use of surgical bypass versus endovascular bypasses, such as chimney/snorkel option or the use of fenestrated/ branched endograft. We have included a summary of how these variables were defined and grouped in Appendix II. Outcomes The primary endpoint of the study was 30-day mortality. Secondary outcomes included any cardiac complications (MI, dysrrhythmia, CHF), neuro (stroke/TIA, Spinal cord ischemia [transient or present at discharge]), respiratory failure (pneumonia, required intubation or both), renal (acute renal failure or renal ischemia), bowel ischemia, arm/leg embolism, and failure to rescue (FTR). FTR is mortality within 30-days following major complications, including the need for transfusion, neuro, bowel ischemia, arm/leg embolism, acute renal failure, renal ischemia, dialysis, MI, dysrrhythmia, CHF, stroke, or respiratory failure. Statistical Analysis Categorical and continuous covariates were studied using Student’s t-test, fishers exact, median, and chi-square, as appropriate. Multivariable logistic regression models for primary (30-day mortality) and secondary outcomes (FTR, major complication, bowel ischemia, cardiac, pulmonary, neuro, and renal failure) were created and clustered by centers. The covariates were included based on clinical relevance and statistical significance in the univariate analysis. All models were tested using HosmerLemeshow goodness-of-fit test and area under the receiver-operating curve. A P-value of less than 0.05 was defined as statistically significant. Statistical analysis was performed at our institution using STATA-15 software (StataCorp LLC, College Station, TX).

Hospital volume and outcomes of TAAA 3

RESULTS A total of 2,115 patients were identified undergoing endovascular repair of TAAA. One hundred eighteen centers were performing TAAA repair: 92 low [34.1%], 19 medium [32.9%] and 7 high [33.1%] volume hospitals. The mean number of cases (S.D.) performed at low (LVH), medium (MVH), and high (MVH) were 3.62 (1.38), 9.63 (3.01), 22.73 (4.66), respectively. Patients at HVH were more likely to treat Type III TAAA; however, Type I TAAA repair was more common among LVH (Table I). Compared to LVH and MVH, patients undergoing TAAA repair at HVH were slightly older (mean age: 72 vs. 69 and 71) and had higher prevalence of multiple comorbidities, including coronary artery disease (32.6% vs. 22.5% and 21.7%), chronic obstructive pulmonary disease (37.9% vs. 31.4% and 20.9%), smoking history (84.1% vs. 76.4% and 75.7%), history of prior aneurysm repair (33.8% vs. 17.6% and 26.8%), aortic surgery (34.9% vs. 21.2% and 28.5%), and history of genetic disorders (5.4% vs. 3.1% and 3.0%) (All P < 0.05). Patients at HVH were also more likely to be on preoperative medications, including aspirin, P2Y12 inhibitor, and anticoagulation (Table I). In regards to aneurysm and operative characteristics, patients at LVH were more likely to undergo TAAA repair for dissection and were symptomatic at presentation compared to HVH (Table II). A significantly higher number of elective cases were performed at HVH versus LVH versus MVH (84.7% vs. 75.5% vs. 71.8%) for aneurysmal disease as indication (81.3% vs. 59.5% vs. 63.5%) (Both P < 0.001). Total contrast volume was higher between LVH and MVH versus HVH (mean (S.D.): 129.29 (74.96) and 136.63 (78.55) versus 113.51 (77.11) milliliters, P < 0.001). However, patients at HVH had longer operative and fluoroscopy time (Table II). The spinal drain was more likely to be placed preoperative (56.1% vs. 43.0% vs. 37.3%, P < 0.001) in HVH versus MVH versus LVH. In contrast, postoperatively spinal drain placement was higher among LVH and MVH versus HVH (4.4% and 3.7% vs. 1.9%, P < 0.001). Custom/ modified devices were more likely to be used in HVH versus MVH versus LVH (67.9% vs. 27.6% vs. 27.2%) (P < 0.001) (Table II). Additionally, HVH and MVH were more likely to use fenestrated/branched endografts or chimney/scallop approach versus surgical bypass, which was higher in LVH for revascularization of the visceral arteries (Table III). The overall technical success was also higher in HVH versus MVH versus LVH (98.5% vs. 97.1% vs. 95.9%, P ¼ 0.04).

4 Locham et al.

Annals of Vascular Surgery

Table I. Patient characteristics undergoing thoracoabdominal aortic aneurysm repair at low (LVH) versus medium (MVH) versus high (HVH) volume hospitals Patient characteristics

Mean number of cases, SD Total Centers (N ¼ 118) TAAA Type I Type II Type III Type IV Demographics Age, median (IQR) Male Race White Black Others Transferred from Home Insurance Medicaid/Medicare Commercial Other Comorbidities BMI, median (IQR) History of cerebrovascular disease None Stroke/asymp Stroke/Symp Hx CAD Hx CHF Hx COPD Hx DM Hx Dialysis HTN Smoker Hx PCI/CABG Hx CEA/CAS Aneurysm repair PVI/Bypass Major amputation Anemia Normal/Mild Moderate Severe Previous aortic surgery Genetic history (Marfan/Ehlers/ Loeys/Others) Preoperative medications Aspirin P2Y12 inhibitor

Low, 721 (34.1%)

3.62 (1.38)

Medium, 695 (32.9%)

9.63 (3.01)

N ¼ 92

N ¼ 19

99 303 109 210

80 282 146 187

(13.7) (42.0) (15.1) (29.1)

(11.5) (40.6) (21.0) (26.9)

High, 699 (33.1%)

22.73 (4.66) N¼7 26 301 157 215

(3.7) (43.1) (22.5) (30.8)

69 (60,76) 481 (66.7)

71 (63,77) 461 (66.3)

72 (65,78) 471 (67.4)

539 124 58 548

457 163 75 488

596 51 52 551

(74.8) (17.2) (8.0) (76.0)

(65.8) (23.5) (10.8) (70.4)

P-value

(85.3) (7.3) (7.4) (78.8)

<0.001 0.64 0.001 0.28 <0.001 0.92 <0.001

0.001 <0.001

408 (56.6) 249 (34.5) 64 (8.9) 27.57 (24.34, 32.04)

396 (57.1) 231 (33.3) 67 (9.7) 27.45 (23.83, 31.88)

331 (47.4) 231 (33.1) 137 (19.6) 27.19 (23.92, 30.81)

657 41 23 162 99 226 113 22 643 551 176 13 127 76 2

(91.1) (5.7) (3.2) (22.5) (13.7) (31.4) (15.7) (3.1) (89.2) (76.4) (24.4) (1.8) (17.6) (10.5) (0.3)

623 46 25 151 91 215 105 26 642 526 165 25 186 61 3

(89.8) (6.6) (3.6) (21.7) (13.1) (20.9) (15.1) (3.7) (92.4) (75.7) (23.7) (3.6) (26.8) (8.8) (0.4)

634 40 25 228 80 265 102 16 643 588 185 23 236 72 5

(90.7) (5.7) (3.6) (32.6) (11.4) (37.9) (14.6) (2.3) (92.1) (84.1) (26.5) (3.3) (33.8) (10.3) (0.7)

511 176 20 153

(71.4) (24.6) (4.1) (21.2)

460 195 36 198

(66.6) (28.2) (5.2) (28.5)

530 149 19 244

(75.9) (21.4) (2.7) (34.9)

0.46 0.92

<0.001 0.41 0.008 0.86 0.29 0.06 <0.001 0.47 0.1 <0.001 0.48 0.48 0.003

<0.001

22 (3.1)

21 (3.0)

38 (5.4)

0.03

397 (55.1) 81 (11.3)

384 (55.3) 69 (9.9)

453 (64.9) 75 (10.7)

<0.001 0.72 (Continued)

Volume

-, -

2019

Hospital volume and outcomes of TAAA 5

Table I. Continued Patient characteristics

Statin Beta blocker ACE inhibitor Anticoagulation

Low, 721 (34.1%)

395 481 325 63

(54.9) (66.8) (45.1) (8.8)

Medium, 695 (32.9%)

392 496 300 74

(56.4) (71.4) (43.2) (10.7)

High, 699 (33.1%)

473 451 280 109

(67.8) (64.8) (40.2) (15.6)

P-value

<0.001 0.03 0.17 <0.001

Table II. Operative characteristics in patients undergoing thoracoabdominal aortic aneurysm repair at low (LVH) versus medium (MVH) versus high (HVH) volume hospitals Operative characteristics

Indication Dissection Aneurysm Presentation Asymptomatic Symptomatic Urgency Elective Urgent/Emergent Contrast, mean (S.D.) Crystalloid in liters, mean (S.D.) EBL, mean (S.D.) Fluro time, mean (S.D.) Total procedure time, mean (S.D.) Intraoperative transfusion IVUS/TEE use Conversion to open repair Spinal drain placement None Preop Postop Custom/Modified Devices Technical success Staged branched procedure

LVH, 721 (34.1%)

MVH, 695 (32.9%)

HVH, 699 (33.1%)

292 (40.5) 429 (59.5)

254 (36.6) 441 (63.5)

131 (18.7) 568 (81.3)

422 (58.8) 296 (41.2)

388 (55.8) 307 (44.2)

519 (74.4) 179 (25.6)

P-value

<0.001 <0.001 <0.001 544 177 129.29 2.2

(75.5) (24.6) (74.96) (1.6)

498 196 136.63 2.3

(71.8) (28.2) (78.55) (1.5)

591 107 113.51 2.5

(84.7) (15.3) (77.11) (1.4)

<0.001 <0.001

387.9 (652.6) 45.2 (43.1) 210.0 (135.6)

501.6 (786.6) 57.6 (48.4) 236.2 (148.1)

454.8 (768.0) 62.9 (41.3) 237.1 (114.0)

0.02 <0.001 0.0001

145 (20.2)

214 (30.9)

213 (30.5)

<0.001

332 (46.1) 4 (0.6)

359 (51.9) 9 (1.3)

262 (37.5) 4 (0.6)

<0.001 0.21

420 269 32 193

370 299 26 191

294 392 13 469

<0.001 (58.3) (37.3) (4.4) (27.2)

540 (95.9) 80 (11.4)

In univariate analysis, 30-day mortality was significantly higher among patients undergoing TAAA repair at LVH and MVH compared to HVH (5.1% and 6.5% vs. 2.2%, P ¼ 0.002) (Table IV). After stratifying by the types of TAAA, 30-day mortality was lower among HVH performing complex Type II (2.5% vs. 5.0% and 7.4%, P ¼ 0.047) and Type III (0.8% vs. 7.1% and 7.8%) endovascular repair of TAAAs compared to LVH and MVH (Figure 1). No significant difference was seen between hospital volume and Types I & IV TAAA. Additionally, patients at LVH and MVH were also at increased risk

(53.2) (43.0) (3.7) (27.6)

571 (97.1) 77 (11.2)

(42.1) (56.1) (1.9) (67.9)

527 (98.5) 80 (11.6)

<0.001 0.04 0.98

of developing stroke/TIA (4.3% and 3.2% vs. 1.7%, P < 0.02) and respiratory complications (9.3% and 10.7% vs. 6.6%, P < 0.02) compared to HVH. No significant difference was seen in renal failure, cardiac failure, pulmonary failure, bowel ischemia, and neuro complications (Table IV). Failure to rescue [FTR] was also significantly lower among HVH versus MVH versus LVH (3.5% vs. 12.1% vs. 11.6%, P ¼ 0.001). In multivariable analysis, after adjusting for potential confounders, compared to HVH, LVH had nearly a two-four fold increased risk of 30-day

6 Locham et al.

Annals of Vascular Surgery

Table III. Revascularization of visceral artery across different hospital volumes [low (LVH) versus medium (MVH) versus high (HVH)] Revascularization of the visceral vessels

Celiac artery None Occluded/Covered Chimney/Scallop Fenestration/ branched Surgical bypass Superior mesenteric artery None Occluded/Covered Chimney/Scallop Fenestration/ branched Surgical bypass Left renal artery None Occluded/Covered Chimney/Scallop Fenestration/ branched Surgical bypass Right renal artery None Occluded/Covered Chimney/Scallop Fenestration/ branched Surgical bypass

Low

415 37 27 38

(76.4) (6.8) (5.0) (7.0)

26 (4.8)

353 4 82 88

(63.0) (0.7) (14.6) (15.7)

33 (5.9) 275 12 24 124

(60.4) (2.6) (5.3) (27.3)

20 (4.4) 280 9 27 120

(61.0) (2.0) (5.9) (26.1)

23 (5.0)

mortality (OR (95% CI): 2.30 (1.20e4.41), P ¼ 0.01), FTR (OR (95% CI): 4.42 (1.86e10.54), P ¼ 0.001), and bowel ischemia (OR (95% CI): 2.35 (1.19e4.67), P ¼ 0.01). Additionally, compared to HVH, MVH had increased odds of 30day mortality (OR (95% CI): 2.14 (1.16e4.41), P ¼ 0.02), FTR (OR (95% CI): 4.08 (1.73e9.62), P ¼ 0.001) and any complication (OR (95% CI): 1.42 (1.02e1.99), P ¼ 0.04) (Figure 2).

DISCUSSION In a recent systematic review on 237,074 patients undergoing AAA repair from the European population, Phillips et al.11 reported an inverse volumeoutcome relationship between hospital volume and mortality. Similarly, using Medicare Beneficiaries data from the United States, Zettervall and colleagues demonstrated decreased mortality with increased hospital volume for both endovascular (Quintile 1 [0e9 EVARs]: 1.9%; quintile 5 [49e

Medium

High

337 66 33 92

195 18 64 342

(62.5) (12.2) (6.1) (17.1)

11 (2.0)

277 4 89 149

(51.7) (0.8) (16.6) (27.8)

17 (3.2) 240 27 49 158

(49.0) (5.5) (10.0) (32.2)

16 (3.3) 255 17 60 153

(50.7) (3.4) (11.9) (30.4)

18 (3.6)

P-value

(31.3) (2.9) (10.3) (54.9)

<0.001

4 (0.6)

148 0 45 415

(24.2) (0) (7.4) (67.8)

<0.001

4 (0.7) 165 25 19 388

(27.4) (4.2) (3.2) (64.5)

<0.001

5 (0.8) 158 12 25 405

(26.2) (2.0) (4.1) (67.1)

<0.001

4 (0.7)

198 EVARs]: 1.4%; P < 0.01) and open (Quintile 1 [0e5 open repairs]: 6.3%; quintile 5 [14e62 open repairs]: 3.8%; P < 0.01) repair of AAA.12 The role of hospital volume and surgical outcomes for the repair of abdominal aneurysm has been well studied; however, the evidence is lacking on a volumeoutcome relationship for the repair of thoracoabdominal aortic aneurysm (TAAA). TAAA requires complex repair and is associated with higher mortality in both intact3 (10%) and ruptured4 (34%) cases. Traditionally, TAAA has been managed via an open surgical approach; however, there is a gradual shift toward the use of endovascular approaches in managing these complex aneurysms.8 Prior studies by our group on outcomes and cost of TAAA and complex AAA repair favor the use of endovascular over open repair.3,4,13 Very few studies have looked at the impact of hospital volume and overall experience for managing the endovascular repair of TAAA. In our study, we utilized a large national database and looked at 118 centers performing the endovascular repair of intact

Volume

-, -

2019

Hospital volume and outcomes of TAAA 7

Table IV. Short-term outcomes in patients undergoing thoracoabdominal aortic aneurysm repair at low (LVH) versus medium (MVH) versus high (HVH) volume hospitals Postoperative outcomes

30-day mortality Postoperative complications Any complication Transfusion Bowel ischemia Renal failure Neurological complication Spinal ischemia Stroke/TIA Cardiac complication Pulmonary Dialysis Arm/Leg embolism Failure to rescue Discharge to Home Total LOS, median (IQR) Total ICU, median (IQR)

LVH, 721 (34.1%)

29 (5.1%)

MVH, 695 (32.9%)

36 (6.5%)

HVH, 699 (33.1%)

12 (2.2%)

P-value

0.002

195 215 24 67 60

(27.1) (30.1) (3.3) (18.9) (8.3)

216 274 23 59 54

(31.2) (39.6) (3.3) (19.7) (7.8)

172 296 11 52 45

(24.6) (42.4) (1.6) (15.4) (6.4)

0.02 <0.001 0.07 0.31 0.38

35 31 76 67 28 1 26 513 5

(4.9) (4.3) (10.5) (9.3) (4.0) (0.1) (11.6) (76.3) (3, 8)

40 22 101 74 33 1 32 510 5

(5.8) (3.2) (14.6) (10.7) (5.0) (0.2) (12.1) (79.3) (3, 9)

35 12 87 46 18 2 9 516 5

(5.0) (1.7) (12.5) (6.6) (2.6) (0.3) (3.5) (76.9) (3, 8)

0.71 0.02 0.07 0.02 0.09 0.77 0.001 0.39 0.06

3 (2, 5)

Fig. 1. 30-day mortality following endovascular repair of Type I-IV Thoracoabdominal aortic aneurysm between low versus medium versus high volume hospitals (*P < 0.05).

TAAA. Our study found that patients undergoing repair at HVH had significantly lower rates of 30day morbidities, mortality, and failure to rescue, compared to LVH and MVH. A prior study by Cowan and colleagues looked at effect of hospital volume following open repair of TAAA.6 The mortality rates were nearly halved in HVH (15.0%) versus LVH (27.4%); however, this study period was prior to the introduction of endovascular approach for TAAA repair (1988e1998). In our study, overall 30-day mortality following

3 (2, 6)

3 (2, 5)

0.69

elective endovascular repair of TAAA was only 3.0%. In spite of HVH treating higher risk patients with more advanced TAAA types, the operative mortality rate was significantly lower compared to MVH and LVH (2.2% vs. 6.5% and 5.1%, P ¼ 0.002). After adjusting for potential confounders, both LVH and MVH had nearly two-fold increase odds of mortality at 30-days. Similar to our study, a recent study from Germany, Geisbusch et al.9 analyzed patients undergoing open and endovascular TAAA repair and reported twice as low inhospital mortality in HVH versus LVH (13.9% vs 32.1%, P < 0.01). Additionally, they also found an increase in annual caseloads to be significantly associated with decreased mortality. In contrast, using a California Office of State Health Policy and Development patient discharge database, Weiss et al.7 found no significant difference in the odds ratio of mortality (OR: 0.37, P ¼ 0.077) at high-volume hospitals compared to low-volume hospitals for open repair of TAAA. The inconsistency in the results above may be due to the variability in defining the hospital volumes, different types of TAAA, surgical approach (open versus endovascular), and indication (ruptured versus elective). In our study, we only looked at endovascular repair, excluded any ruptured cases, and were able to identify and adjust for the different extent of TAAA. Whereas, the aforementioned

8 Locham et al.

Fig. 2. Logistic regression analysis of major complications in low (LVH) versus high (HVH) and medium (MVH) versus HVH for endovascular treatment of thoracoabdominal aortic aneurysms (TAAA). Adjusted for hospital volume (low [LVH], medium [MVH], high [HVH] e Reference), TAAA types, age, gender, race, coronary artery disease, pulmonary disease, hypertension, prior aneurysm repair, aortic surgery, genetic history, preoperative medications (aspirin, statin, beta-blockers), urgency, indication, estimated blood loss, and intraoperative transfusion.

studies are largely on open repair and included both intact and ruptured cases and were not able to differentiate between different types of TAAA. Type-II and Type-III TAAA are far more complex than Type-I TAAA and are associated with an increased risk of developing perioperative complications. Compared to LVH and MVH, HVH had significantly lower 30-day mortality if performing the endovascular repair of complex Type II and III TAAA (Figure 1). Furthermore, there is marked variability in the number of cases used to define hospital volumes in the surgical literature on TAAA. In Cowen et al.,6 high volume hospitals were defined as having a median number of 12 cases per year. Whereas, Weiss et al.7 and Geisbusch et al.9 defined high volume as  9 and  13 cases, respectively. Compared to these studies, the mean number of cases included in HVH (>22) of our study was much higher. We believe our definition of high volume might be more clinically and statistically relevant while containing more appropriate and vascular specific data. We included only endovascular repairs and utilized the period (2012e2018) when suitable endovascular options were readily available compared to the gold-standard open repair. Cook Fenestrated endograft received FDA approval in 2012 to treat complex AAA, but it could only be used for para-renal aneurysms and is not applicable in patients with more proximal aneurysms.

Annals of Vascular Surgery

However, hospitals with a higher volume of TAAA cases are probably more likely to obtain an Investigational device exemption to be able to modify the fenestrated or standard endografts for these complex endovascular cases. These high-volume centers are more likely to have access to industry-sponsored and physician-sponsored IDE trials. In our study, we found that HVH was more likely to use custom/ modified devices and utilized fenestrated/branched options more frequently for managing TAAA compared to both LVH and MVH (Table III). Failure to rescue (FTR) (i.e., preventing death following major complication) is relatively a new concept that has increasingly been described in the surgical literature and is dependent on the hospital characteristics. In our study, FTR was significantly lower among HVH (3.5%) compared to LVH (11.6%) and MVH (12.1%) (P ¼ 0.001). Ilonzo et al.14 found similar results in the rate of FTR compared to hospital volume in endovascular aneurysm repair (EVAR) of AAA, showing HVH had a lower FTR rate than LVH (0.7% vs. 1.69%, P < 0.001). Waits et al.15 looked at FTR in relation to mortality and found that the FTR rate was greater in higher mortality tertile hospitals compared to lower mortality tertile hospitals. Using data from a single large tertiary vascular center, Hicks et al.16 reported a tenfold decrease in operative mortality in the tertiary vascular center compared to medicarederived predictions. This may be attributed to the overall surgical and hospital experience in managing patients with complex disease. HVHs have a well-structured approach/protocols and are better equipped to deal with a complication. In our study, HVH was far more likely to place a spinal drain preoperatively than MVH and LVH (56.1% vs. 43.02% and 37.3%, P < 0.001). In contrast, spinal drains were more commonly utilized late postoperatively in LVH versus MVH versus HVH (4.4% vs. 3.7% vs. 1.9%, P < 0.001). Although we are not able to identify the exact reason for the increase in the utilization of the spinal drain postoperatively, one can assume this may be due to spinal cord ischemia (SCI); a serious complication which occurs in the range of 10.4e31.0% of patients undergoing endovascular repair of TAAA.17,18 Literature suggests that the adoption of standardized protocol aiming at prevention, early diagnosis, and treatment of SCI led to frequent regression of SCI symptoms (100% vs. 46.2%, P ¼ 0.017).17 In a review by Nancy Epstein, it was found that utilization of a protocol, including cerebrospinal fluid drain, helped reduced SCI risk from 20% to 2.3% in patients who underwent thoracic and thoracoabdominal aneurysm repair, as well as TEVAR.19

Volume

-, -

2019

Compared to HVHs, LVHs may lack resources and experience to help reduce the number of perioperative complications and mortality. Ghaferi et al.20 identified teaching hospitals with hospital size greater than 200 beds, average daily census greater than 50% capacity, increased nurse-to-patient ratios, and high hospital technology to be associated with lower FTR. In a recent study by Hicks and colleagues, academic hospital type was the single most significant factor of reduced mortality following AAA repair.21 The same group reported additional intrinsic patient factors and overall hospital experience with general surgery and AAA to be associated with lower in-hospital mortality.22 Nurse-to-patient ratio is extremely important when it comes to managing patients with complex disease. Vascular surgery is one of the unique surgical specialties, which is continuously evolving and incorporates new technologies. For example, the use of fenestrated/branched endografts and endoanchors in complex aortic pathologies offers suitable and less invasive options to patients that were previously not good candidates for any treatment. HVH are commonly the sites that offer these innovative approaches and should be the primary sites when it comes to managing complex vascular pathologies, including the repair of thoracoabdominal aortic aneurysm. The study has some limitations related to the methodology and study database. As described above, there is marked variability in defining hospital volumes across the surgical literature. However, we believe our methodology of creating three equal quintiles based on the mean number of cases performed at each center was appropriate and sufficient to address our study question. Additionally, surgical experience may also influence the results, which was not studied in this paper. Finally, although we were able to identify the types of TAAA using the proximal and distal extent of the disease based on the different zones as shown in Appendix I; but this is open to critique and may be labeled incorrectly depending on the knowledge and experience to classify these different zones accurately. Like any retrospective study using a large database, there is also the potential of selection bias in addition to errors in data entry.

CONCLUSION Using a large national VQI database, our study reports significantly lower mortality, morbidity, and FTR in patients undergoing endovascular repair of TAAA at HVH compared to LVH and MVH. Complex

Hospital volume and outcomes of TAAA 9

endovascular repair of TAAA is extremely challenging, and HVH may have the expertise and the tools to help reduce morbidity and mortality. Patients with TAAA should strictly be managed in high volume Aortic Centers of Excellence.

No special funding was obtained to complete this study.

REFERENCES 1. Pieper D, Mathes T, Neugebauer E, et al. State of evidence on the relationship between high-volume hospitals and outcomes in surgery: a systematic review of systematic reviews. J Am Coll Surg 2013;216:1015e1025.e18. 2. Chaikof EL, Dalman RL, Eskandari MK, et al. The society for vascular surgery practice guidelines on the care of patients with an abdominal aortic aneurysm. J Vasc Surg 2018;67: 2e77.e2. 3. Locham S, Dakour-Aridi H, Nejim B, et al. Outcomes and cost of open versus endovascular repair of intact thoracoabdominal aortic aneurysm. J Vasc Surg 2018;68:948e955.e1. 4. Locham SS, Grimm JC, Arhuidese IJ, et al. Perioperative outcomes of open versus endovascular repair for ruptured thoracoabdominal aneurysms. Ann Vasc Surg 2017;44: 128e35. 5. Patel VI, Mukhopadhyay S, Ergul E, et al. Impact of hospital volume and type on outcomes of open and endovascular repair of descending thoracic aneurysms in the United States medicare population. J Vasc Surg 2013;58:346e54. 6. Cowan JA Jr, Dimick JB, Henke PK, et al. Surgical treatment of intact thoracoabdominal aortic aneurysms in the United States: hospital and surgeon volume-related outcomes. J Vasc Surg 2003;37:1169e74. 7. Weiss A, Anderson JA, Green A, et al. Hospital volume of thoracoabdominal aneurysm repair does not affect mortality in California. Vasc endovascular Surg 2014;48:378e82. 8. Liao JM, Bakaeen FG, Cornwell LD, et al. Nationwide trends and regional/hospital variations in open versus endovascular repair of thoracoabdominal aortic aneurysms. J Thorac Cardiovasc Surg 2012;144:612e6. 9. Geisb€ usch S, Kuehnl A, Salvermoser M, et al. Hospital incidence, treatment, and in hospital mortality following open and endovascular surgery for thoraco-abdominal aortic aneurysms in Germany from 2005 to 2014: secondary data analysis of the nationwide German DRG microdata. Eur J Vasc Endovascular Surg 2019;69:2004. 10. Cronenwett JL, Kraiss LW, Cambria RP. The society for vascular surgery vascular quality initiative. J Vasc Surg 2012;55:1529e37. 11. Phillips P, Poku E, Essat M, et al. Procedure volume and the association with short-term mortality following abdominal aortic aneurysm repair in european populations: a systematic review. Eur J Vasc Endovascular Surg 2017;53:77e88. 12. Zettervall SL, Schermerhorn ML, Soden PA, et al. The effect of surgeon and hospital volume on mortality after open and endovascular repair of abdominal aortic aneurysms. J Vasc Surg 2017;65:626e34. 13. Locham S, Faateh M, Dakour-Aridi H, et al. Octogenarians undergoing open repair have higher mortality compared with fenestrated endovascular repair of intact abdominal aortic aneurysms involving the visceral vessels. Ann Vasc Surg 2018;51:192e9.

10 Locham et al.

14. Ilonzo N, Egorova NN, McKinsey JF, et al. Failure to rescue trends in elective abdominal aortic aneurysm repair between 1995 and 2011. J Vasc Surg 2014;60:1473e80. 15. Waits SA, Sheetz KH, Campbell DA, et al. Failure to rescue and mortality following repair of abdominal aortic aneurysm. J Vasc Surg 2014;59:909e914.e1. 16. Hicks CW, Black JH, Arhuidese I, et al. Mortality variability after endovascular versus open abdominal aortic aneurysm repair in a large tertiary vascular center using a medicarederived risk prediction model. J Vasc Surg 2015;61:291e7. 17. Dias N, Sonesson B, Kristmundsson T, et al. Short-term outcome of spinal cord ischemia after endovascular repair of thoracoabdominal aortic aneurysms. Eur J Vasc Endovascular Surg 2015;49:403e9. 18. Katsargyris A, Oikonomou K, Kouvelos G, et al. Spinal cord ischemia after endovascular repair of thoracoabdominal

Annals of Vascular Surgery

19.

20.

21.

22.

aortic aneurysms with fenestrated and branched stent grafts. J Vasc Surg 2015;62:1450e6. Epstein NE. Cerebrospinal fluid drains reduce risk of spinal cord injury for thoracic/thoracoabdominal aneurysm surgery: a review. Surg Neurol Int 2018;9:48. Ghaferi AA, Osborne NH, Birkmeyer JD, et al. Hospital characteristics associated with failure to rescue from complications after pancreatectomy. J Am Coll Surg 2010;211:325e30. Hicks CW, Wick EC, Canner JK, et al. Hospital-level factors associated with mortality after endovascular and open abdominal aortic aneurysm repair. JAMA Surg 2015;150: 632e6. Hicks CW, Canner JK, Arhuidese I, et al. Comprehensive assessment of factors associated with in-hospital mortality after elective abdominal aortic aneurysm repair. JAMA Surg 2016;151:838e45.

Volume

-, -

2019

Hospital volume and outcomes of TAAA 10.e1

Appendix Appendix I. Classification of Thoracoabdominal aortic aneurysms based on the proximal and distal extent of the disease. Proximal Zone 3 4 5 6 7 8 9

Distal Zone 3 4 5 6 7 8 9 10 11

Excluded: Zones 0e2.

Type 1

Type 2

Type 3

Type 4

Begins in the proximal half of the descending thoracic aorta (equivalent to ‘‘above T6 disc space’’), involves the remainder of the descending thoracic aorta and transcends the diaphragmatic boundary, involving the upper abdominal aorta without extending below the renal arteries. (3 + 6), (3 + 7), (4 + 6), (4 + 7). Begins in the proximal half of the descending thoracic aorta (‘‘above T6 disc space’’), involves most or all of the descending thoracic aorta and most or all of the abdominal aorta, extends below the renal arteries. (3 + 8), (3 + 9), (3 + 10), (3 + 11), (4 + 8), (4 + 9), (4 + 10), (4 + 11) Involves the distal half (equivalent to ‘‘below T6 disc space’’) of the descending thoracic aorta and involving varying segments of the abdominal aorta. (5 + 6), (5 + 7), (5 + 8), (5 + 9), (5 + 10), (5 + 11) Includes most or all of the entire abdominal aorta, including the renal arteries. These definitions adapted by Crawford, eliminate the Type V category. (6 + 7), (6 + 8), (6 + 9), (6 + 10), (6 + 11), (7 + 8), (7 + 9), (7 + 10), (7 + 11)

Appendix II: Grouping of options for revascularization of visceral arteries None: None Occluded/Covered Includes: 1 ¼ Purposely covered, 2 ¼ Unintentionally covered, 3 ¼ Occluded - coil, 4 ¼ Occluded - plug, 5 ¼ Occluded - open, 6 ¼ Stent, 7 ¼ Stent-graft 1. Purposely covered ¼ intentionally covered by a stent graft without embolization; 2. Unintentionally covered ¼ covered by a stent graft without planning to do so can be due to device malfunction or mal-deployment;

3. Occluded-coil ¼ occlusion of branch vessel using coil embolization methods; 4. Occluded-plug ¼ occlusion of branch vessel using plug devices, such as the Amplatzer; 5. Occluded-open ¼ occlusion of branch by open surgical technique; Chimney/Scallop Includes: 8 ¼ Chimney, 9 ¼ Scallop, 10 ¼ Stented Scallop 8. Chimney ¼ branch vessel stent or stent graft placed in parallel stent configuration alongside an aortic stent graft. Includes ‘‘chimney, snorkel, periscope, and sandwich’’ configurations;

10.e2 Locham et al.

9. Scallop ¼ opening in the grafted portion of the aortic stent graft at the proximal or distal edge of the aortic graft (such that graft material surrounds only a portion of the opening) with no stent/stent graft through the scallop; 10. Stented Scallop ¼ Scallop WITH a stent/stent graft into the branch vessel through the scallop; Fenestration/Branch Includes: 11 ¼ Fenestration, 12 ¼ Stented-fen, 13 ¼ Fen branch, 14 ¼ Side-arm Branch 11. Fenestration ¼ Opening in the grafted portion of the aortic stent graft with graft material on

Annals of Vascular Surgery

all sides, NO stent/stent graft placed through the opening; 12. Stented-fen ¼ fenestration with a bare metal stent through the graft opening; 13. Fen-branch ¼ fenestration with a covered stent through the fenestration; 14. Side-arm branch ¼ directional graft branch (can be internal or external) with a bridging stent graft placed into the branch vessel; Surgical Bypass Includes: Surgical Bypass ¼ Bypass Graft or Transposition, a Socalled Debranching Procedure