Survival and Reintervention Risk by Patient Age and Preoperative Abdominal Aortic Aneurysm Diameter after Endovascular Aneurysm Repair

Accepted Manuscript Survival and Re-intervention Risk by Patient Age and Preoperative Abdominal Aortic Aneurysm (AAA) Diameter Following Endovascular Aneurysm Repair (EVAR) Robert J. Hye, Afra U. Janarious, Priscilla H. Chan, Guy Cafri, Robert W. Chang, Thomas F. Rehring, Nicolas A. Nelken, Bradley B. Hill PII:

S0890-5096(18)30554-5

DOI:

10.1016/j.avsg.2018.05.053

Reference:

AVSG 3953

To appear in:

Annals of Vascular Surgery

Received Date: 19 January 2018 Revised Date:

1 May 2018

Accepted Date: 10 May 2018

Please cite this article as: Hye RJ, Janarious AU, Chan PH, Cafri G, Chang RW, Rehring TF, Nelken NA, Hill BB, Survival and Re-intervention Risk by Patient Age and Preoperative Abdominal Aortic Aneurysm (AAA) Diameter Following Endovascular Aneurysm Repair (EVAR), Annals of Vascular Surgery (2018), doi: 10.1016/j.avsg.2018.05.053. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

1

Survival and Re-intervention Risk by Patient Age and Preoperative Abdominal Aortic

2

Aneurysm (AAA) Diameter Following Endovascular Aneurysm Repair (EVAR)

RI PT

Robert J. Hye Department of Surgery Southern California Permanente Medical Group San Diego, CA USA

Guy Cafri Surgical Outcomes and Analysis Kaiser Permanente San Diego, CA USA

M AN U

Priscilla H. Chan Surgical Outcomes and Analysis Kaiser Permanente San Diego, CA USA

SC

Afra U. Janarious Department of Surgery Southern California Permanente Medical Group San Diego, CA USA

TE D

Robert W. Chang Department of Surgery The Permanente Medical Group South San Francisco, CA, USA,

EP

Thomas F. Rehring Department of Vascular Surgery Colorado Permanente Medical Group Denver, CO, USA

AC C

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Nicolas A. Nelken Department of Vascular Therapy Hawaii Permanente Group Honolulu, HI, USA Bradley B. Hill - Corresponding author Department of Vascular Surgery The Permanente Medical Group 710 Lawrence Expressway, Suite 290 Santa Clara, CA 95051 (e-mail: [email protected]) 1-408-851-2314 1

ACCEPTED MANUSCRIPT

Presented at the 30th Annual Meeting of the Western Vascular Society, Wailea, Maui, Hawaii, September 19-22, 2015.

EP

TE D

M AN U

SC

RI PT

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

AC C

45 46 47 48 49 50

2

ACCEPTED MANUSCRIPT

Abstract:

52

Background: Endovascular Aneurysm Repair (EVAR) has become the standard of care for

53

abdominal aortic aneurysm (AAA) but questions remain regarding benefit in high risk and

54

elderly patients. The purpose of this study was to examine the effect of age, preoperative AAA

55

diameter, and their interaction on survival and re-intervention rates following EVAR.

56

Study Design: Our integrated health system’s AAA endograft registry was used to identify

57

patients who underwent elective EVAR between 2010-2014.Of interest was the effect of patient

58

age at time of surgery (≤80 vs >80 years-old), preoperative AAA diameter (≤5.5cm vs >5.5cm),

59

and their interaction. Primary endpoints were all-cause mortality and re-intervention. Between-

60

within mixed effects Cox models with propensity score weights were fit.

61

Results: Of 1967 patients undergoing EVAR, unadjusted rates for survival at 4 years after EVAR

62

was 76.1%, and re-intervention-free rate was 86.0%. For mortality, there was insufficient

63

evidence for an interaction between age and AAA size (p=0.309). Patient age >80 was associated

64

with 2.53-fold higher mortality risk (HR=2.53 95%CI 1.73-3.70, p<0.001), while AAA >5.5cm

65

was associated with 1.75-fold higher mortality risk (HR=1.75, 95%CI 1.26-2.45, p=0.001). For

66

re-intervention risk, there were no significant interactions or main effects for age or AAA

67

diameter.

68

Conclusion: Age and AAA diameter are independent predictors of reduced survival after EVAR,

69

but the effect is not amplified when both are present. Age >80 years or AAA size >5.5 cm did

70

not increase risk of re-intervention. No specific AAA size, patient age or combination thereof

71

were identified that would contraindicate AAA repair

AC C

EP

TE D

M AN U

SC

RI PT

51

72 3

ACCEPTED MANUSCRIPT

Introduction

74

When to repair abdominal aortic aneurysm (AAA) grows increasingly complicated as life

75

expectancy advances into the ninth and tenth decades. Between 2005 and 2008, 25% of elective

76

AAA repairs reported to the American College of Surgeons National Surgical Quality

77

Improvement Program (ACS NSQIP) were performed in patients aged at least 80 years.(1)

78

Compared to open surgical repair (OSR), endovascular aneurysm repair (EVAR) offers the

79

advantages of regional anesthesia, shorter hospital stay, lower peri-operative mortality, and

80

similar survival benefit despite higher rates of re-intervention and graft-related complications.(2-

81

4) Octogenarians undergoing EVAR and OSR have similar mid- and long-term survival, but

82

EVAR patients experience less peri-operative mortality,(1) reduced blood loss, shorter intensive

83

care unit stay, shorter hospital stay, and higher likelihood of discharge.(5) Thus, EVAR has

84

become preferable to OSR in elderly patients.

85

Despite the acceptably low peri-operative mortality with EVAR, ranging from 0-6% in those

86

over 80 years,(6-10) octogenarians have longer procedures, greater blood loss, and larger

87

aneurysms than their younger counterparts.(9, 11) Overall survival ranges from 60-90% at five

88

years,(6, 7, 9, 10) which is less than for younger patients, but is considered acceptable for good

89

risk elderly patients. Importantly, freedom from aneurysm related mortality is high (92.9%) in

90

this population, even five years post-operatively.(10) Therefore, most vascular surgeons have not

91

considered advanced age to be a contraindication to AAA repair in otherwise healthy patients.

92

Many clinical trials that established a threshold for elective AAA repair excluded elderly patients

93

and those with significant medical comorbidities. The Aneurysm Detection and Management

94

(ADAM) trial established a threshold diameter for open repair of asymptomatic AAA at 5.5 cm,

AC C

EP

TE D

M AN U

SC

RI PT

73

4

ACCEPTED MANUSCRIPT

yet the trial only included patients younger than 80 years of age.(4, 12) Early trials also found no

96

survival advantage from OSR for AAA with diameter <5.5 cm, but older patients, those with

97

significant comorbidities or life expectancy shorter than five years were excluded.(4, 13) Similar

98

selection biases limit the generalizability of trials demonstrating no survival advantage to EVAR

99

over surveillance for small aneurysms.(14, 15) The Comparison of surveillance versus Aortic

100

Endografting for Small Aneurysm Repair trial, (CAESAR) excluded patients 80 and over,(14)

101

and while the Positive Impact of endoVascular Options for Treating Aneurysm earLy study,

102

(PIVOTAL) allowed patients up to age 90, all patients with renal disease were excluded.(15)

103

While intuitively, it seems an increased size threshold for AAA repair in the elderly is justified, it

104

may select for a patient group with even lower long-term survival.

105

Some investigators have identify age specific outcome differences that might help guide

106

management of elderly patients with AAA. A large meta-analysis of over 25,000 AAA EVAR

107

cases found higher but acceptable perioperative and mid-term mortality rates and no difference in

108

secondary intervention rates when comparing the EVAR outcomes of octogenarians to younger

109

patients.(16)

110

Another important consideration unique to EVAR is the need for long-term surveillance, re-

111

intervention for endoleaks, graft limb occlusions or other complications, which increase patient

112

morbidity and expense to the health care system. Some literature suggests that re-intervention is

113

more likely in larger aneurysms and in elderly patients.(17-19) Larger AAA size has also been

114

associated with decreased survival, increased rate of type II endoleak and device migration after

115

EVAR.(17, 18, 20, 21)

AC C

EP

TE D

M AN U

SC

RI PT

95

5

ACCEPTED MANUSCRIPT

Registries offer a robust trove of data that partially fills the information gap for populations

117

excluded from landmark clinical trials. For example, the EUROSTAR registry shows patients

118

with larger aneurysm diameters (>6.5 cm) tend to be older and have more cardiovascular and

119

pulmonary disease than patients with smaller aneurysms. Registry data also indicate larger

120

aneurysm patients have increased overall and peri-operative mortality, rates of type I endoleak,

121

and higher future aneurysm rupture rates than patients with smaller AAA (≤6.4 cm) at time of

122

repair.(18, 21, 22) Recent data from the University of Alabama also showed age-matched

123

cohorts had better survival with small rather than medium or large aneurysms at the time of

124

EVAR.(20) Aneurysm diameter greater than 6.0 cm at time of repair has been identified as an

125

independent predictor of worse survival.(23)

126

Because of the increasing numbers of elderly patients and data showing decreased long-term

127

survival after EVAR in patients over 80 and in those with large AAA, we evaluated the outcomes

128

for patients in a large multi-center endograft registry. We hypothesized that age over 80 years

129

and aneurysm size over 5.5cm conferred a higher risk for death and that the presence of both

130

might amplify the risk. Additionally, we examined the effect of age and aneurysm size on overall

131

sample, age-matched groups, and aneurysm-size-matched groups in both mortality and re-

132

intervention risk.

133

Methods

134

Data of patients undergoing EVAR were collected and entered in the Kaiser Permanente

135

Endovascular Stent Graft Registry (KPSGR), a multi-center registry capturing data from

136

Northern California, Southern California, Hawaii, Colorado, Northwest and Mid-Atlantic regions

137

of the United States. All EVAR cases are captured electronically with 100% participation rate.

AC C

EP

TE D

M AN U

SC

RI PT

116

6

ACCEPTED MANUSCRIPT

Methods for validation and data collection for the registry have been described previously.(24)

139

Institutional Review Board (IRB) approval was obtained from each participating region. The

140

KPSGR leverages the electronic medical record system of a large, integrated health system

141

serving over 10 million members, to identify records and import data for patients undergoing

142

EVAR for AAA repair. Vascular surgeons perform the procedures, operating surgeons complete

143

a templated operative note capturing relevant procedural details that are entered into the registry.

144

Registry staff review charts to identify and enter missing data and adjudicate complications or re-

145

interventions flagged through automated screening. Patient outcomes are reported in the registry

146

until their Kaiser Permanente membership is terminated by either death or change of health

147

insurance.(24)

148

Inclusion-exclusion criteria

149

Patients who underwent elective EVAR between January 2010 and June 2014 in all centers from

150

participating regions were included in the study. Cases with missing implant information, or

151

infrequently used implants, were excluded (i.e., Cook ZenithFen, Medtronic AneuRx, Medtronic

152

Talent, Trivascular Ovation; N=142, 6.6%). Exclusions were necessary to ensure a balanced

153

analysis because not all exposure groups of interest included patients with these devices.

154

Exposure variables

155

The exposure variables of interest were (1) patient age (≤80 versus >80 years) and (2) pre-

156

operation AAA size (≤5.5cm versus >5.5cm). Based on these factors, we formed four groups: (1)

157

≤80 years old with AAA size ≤5.5 cm, (2) ≤80 years old with AAA size >5.5 cm, (3) >80 years

158

old with AAA size ≤5.5 cm, (4) >80 years old with AAA size >5.5 cm. Pre-op AAA diameter

159

measurement was provided to the registry by vascular surgeons from each center using maximal

AC C

EP

TE D

M AN U

SC

RI PT

138

7

ACCEPTED MANUSCRIPT

short axis on 2D imaging or maximal orthogonal view on complex imaging. We selected the

161

largest of anterior-posterior or transverse measurements as the pre-op diameter. All patients

162

included in the study underwent elective AAA EVAR. The indications for intervention in patients

163

whose AAA diameter was < 5.5 cm included saccular AAA, iliac aneurysms needing repair along

164

with concomitant AAA EVAR, females with fusiform AAA < 5.5 cm, AAAs deemed "rapidly

165

expanding," e.g., > 5 mm AAA diameter growth in < 6 months, and other factors that caused the

166

vascular surgeon to recommend elective AAA, either as the primary procedure or as an adjunctive,

167

concomitant procedure. To evaluate sensitivity, we repeated the same analysis using aneurysm

168

size cut-offs of 6.0 cm and 6.5 cm. Results were not different from the analyses using 5.5 cm as

169

the aneurysm cut-off diameter (refer to results section).

170

Outcomes

171

Outcomes for analysis included all-cause mortality following EVAR and re-intervention free

172

survival. Mortality data was obtained from the Social Security Administration, thus capturing all

173

patients, regardless of their insurance at the time of death.. Re-intervention was defined as any of

174

three events: conversion to open repair, graft revision, or secondary intervention. Graft revision

175

was defined as a procedure that involved placement of a new endograft component to ensure

176

integrity such a proximal cuff or iliac extension, and/or implantation or explantation of an EVAR

177

graft component to repair primary implant. Secondary interventions were defined as procedures

178

necessary for maintenance of integrity of the endovascular AAA repair and endograft such as coil

179

embolization for treating endoleaks and balloon angioplasty/stenting for limb kinking or

180

thrombosis. The Registry surveillance program captured re-interventions for all patients who

181

continued their insurance coverage in the integrated healthcare system.

182

Confounders

AC C

EP

TE D

M AN U

SC

RI PT

160

8

ACCEPTED MANUSCRIPT

Potential confounders included patient gender, race, diabetes, body mass index (BMI),

184

concurrent surgery for iliac aneurysm, implant type (Cook Zenith, EndoLogix AFX or IntuiTrak,

185

Gore Excluder, Medtronic Endurant), or renal insufficiency/peripheral vascular

186

disease/arteriosclerotic cardiovascular disease within one year pre-operation (based on Elixhauser

187

comorbidity diagnostics codes).(25) Chart review was performed to include the following

188

variables: smoking status (active smoker, never smoked, quit), peripheral artery disease (PAD)

189

history of myocardial infarction (MI), coronary artery bypass graft (CABG), coronary artery

190

disease (CAD), coronary stents, congestive heart failure (CHF), and cardiac chest pain or angina.

191

Statistical Analysis

192

To evaluate the association of risk groups and hazard of mortality and re-intervention, between-

193

within mixed effects Cox models with normal cross-classified random effects and propensity

194

score weights were fit. In the mixed effect model, we included random intercepts for surgeons

195

and hospitals, three fixed effect indicators for the four risk groups (≤80 years-old with AAA

196

≤5.5cm group as the reference group) and six proportions corresponding to the average amount

197

these risk groups occurred within each surgeon and hospital.

198

To ensure comparability of groups in observational data from registry source, we used propensity

199

score weighting as a technique to create strata of patients with similar risks. Specifically,

200

propensity score weighting was performed to reduce the effect of potential confounders (pre-

201

defined covariates) on estimating the average treatment effect. Propensity score weights were

202

obtained by marginal mean weighting through stratification.(26) To account for missing values,

203

multiple imputations were performed using a fully conditional specification(27) to create 50

204

versions of the analytic dataset. Each dataset was separately analyzed using the same model and

AC C

EP

TE D

M AN U

SC

RI PT

183

9

ACCEPTED MANUSCRIPT

the parameter estimates averaged over imputations and variance calculated using Rubin’s

206

rule.(28) The imputation model included all variables including treatment, confounders and

207

outcomes.

208

To evaluate whether age amplified the effect of aneurysm size on outcomes, the interaction of

209

age and aneurysm size was examined. Interaction was calculated by determining the difference in

210

age effects within each level of AAA size, and then taking the difference of the resulting

211

quantities, i.e. estimated coefficient of ≤80 years-old with AAA >5.5cm and estimated coefficient

212

of >80 years-old with AAA ≤5.5cm subtracted from estimated coefficient of >80 years-old with

213

AAA >5.5cm. When the interaction was not significantly different from zero, then there was no

214

evidence that age or aneurysm size amplifies risk of the other. In other words, the aneurysm size

215

effect between the two age-matched groups (i.e. age ≤80 years-old vs >80 years-old) were not

216

significantly different from each other. Likewise, the age effect between the aneurysm size

217

matched groups (i.e. AAA ≤5.5cm vs AAA >5.5cm) were not significantly different from each

218

other. As a result, the main effect of age and aneurysm size were reported independently. The

219

main effects for age and aneurysm size were calculated by averaging parameter estimates (using

220

inverse variance weights) over the factor that was not of interest. Hazard ratios (HR) along with

221

95% confidence intervals (CI) are reported. Statistical analyses were performed using R version

222

3.1.2 (R Foundation for Statistical Computing, Vienna, Austria); alpha=0.05 was used as the

223

threshold for statistical significance.

224

Results

225

A total of 1967 patients who underwent EVAR from 28 medical centers and 103 surgeons

226

between January 2010 and June 2014 met inclusion criteria: 778(40%) were ≤80 years old with

AC C

EP

TE D

M AN U

SC

RI PT

205

10

ACCEPTED MANUSCRIPT

AAA ≤5.5cm, 731(37%) were ≤80 years with AAA >5.5cm, 193(10%) >80 years with AAA

228

≤5.5cm and 265(13%) were >80 years with AAA >5.5cm. Thirty-day operative mortality in the

229

four groups was 0.9%, 1.1%, 1.6% and 2.3% respectively; it was 1.0% in those ≤80 and 2.0% in

230

those >80. The four-year post-operative unadjusted survival in the four groups was 87.4%,

231

75.8%, 65.6%, 48.9% respectively. The median follow up time for re-intervention was 587 days

232

(1.6 years, range from 1 day - 4.5 years). A total of 238(12.1%) were deceased (N=33 underwent

233

re-intervention before death) and 102(5.2%) were lost to follow up. Patients who died without

234

undergoing re-intervention (N=193, 9.8%) had a median follow up time of 491 days (1.3 years,

235

range 6 days - 4.5 years). Those who were lost to follow up (N=102, 5.2%, and 12 of them

236

deceased) had a median follow up time of 511 days (1.4 years, range 1 day to 4.1 years). During

237

the study period, 189 (9.6%) cases underwent a subsequent re-intervention, range from 0 to 1337

238

days (3.7 years), at a median time of 97 days (3.2 months, IQR 17 days to 419 days). 90% of the

239

re-intervention occurred within 2 years post-operative (90-percentile = 730.8 days).

240

Patient demographics, implant type, and medical comorbidities are shown in Table 1. The mean

241

age was 74.4 (standard deviation[SD] = 7.9) years, and mean AAA diameter was 5.7 (SD = 1.0)

242

cm. Those over 80 years-old comprised 23% (458/1967) of all EVAR patients, and 20%

243

(193/971) of those who had aneurysms ≤5.5cm were over 80 years-old. Compared with the

244

younger group, patients of over 80 years-old had lower proportions of males (75% versus 87%),

245

active smokers (9% versus 28%), and diabetics (21% versus 29%) but a higher proportions of

246

renal insufficiency (35% versus 22%). Comorbidities including PAD, history of MI, history of

247

CABG, CHF with reduced ejection fraction, and anginal chest pain were similarly distributed

248

between both age groups. In all groups, the graft implanted most frequently was the Gore

AC C

EP

TE D

M AN U

SC

RI PT

227

11

ACCEPTED MANUSCRIPT

Excluder (39%) followed by the Medtronic Endurant (24%), Cook Zenith (19%) and EndoLogix

250

AFX or IntuiTrak (18%) (Table 1).

251

Demographic, medical, and surgical implant differences among groups were significant with

252

43% of mean standardized differences among risk groups exceeding 0.1 (maximum 0.57 and

253

mean 0.12). After the propensity score weighting, group imbalance was reduced: 20% of the

254

mean standardized differences among risk groups were over 0.1 (max=0.33 and mean=0.06).

255

All-cause mortality

256

When compared with patients ≤80 years old and AAA ≤5.5cm, the hazard of mortality was 3.87-

257

fold higher (HR=3.87, 95%CI 2.47-6.08, p<0.001) in patients >80 years old with AAA >5.5cm,

258

2.98-fold higher (HR=2.98, 95%CI 1.82-4.90, p<0.001) in patients >80 years old with AAA

259

≤5.5cm, and 1.94-fold higher (HR=1.94, 95%CI 1.32-2.86, p<0.001) in patients ≤80 years old

260

with AAA >5.5cm (Table 2, 4). The interaction between aneurysm size and patient age was not

261

significant, indicating neither age nor aneurysm size amplifies risk of the other (Z=1.02,

262

p=0.309). Among patients with aneurysm size ≤5.5cm, there was 2.98-fold higher risk (HR=2.98,

263

95%CI 1.82-4.90) for over 80 years of age compared to the younger counterpart. Furthermore,

264

among patients with aneurysm size >5.5cm, there was 2.00-fold high risk (HR=2.00, 95%CI

265

1.10-3.61) for over 80 years of age compared to the younger counterpart. The average effect for

266

age over 80 was a 2.53-fold higher risk (HR=2.53, 95%CI 1.73-3.70, p<0.001) in mortality than

267

age ≤80. For aneurysm size effect among patients under 80 years of age, there was 1.94-fold

268

higher risk (HR=1.94, 95%CI 1.32-2.86) for patients with aneurysm size > 5.5cm compared to

269

the smaller aneurysm size counterpart. Furthermore, among patients over 80 years of age, there

270

was no significant difference in mortality risk between those having >5.5cm aneurysm and the

AC C

EP

TE D

M AN U

SC

RI PT

249

12

ACCEPTED MANUSCRIPT

smaller aneurysm size counterpart (HR=1.30, 95%CI 0.66-2.54). The average effect for pre-

272

operation aneurysm size over 5.5 cm was a 1.75-fold higher risk (HR=1.75, 95%CI 1.26-2.45,

273

p=0.001) in mortality than those with aneurysm size 5.5 cm and smaller. Figure 1 shows adjusted

274

survival curves based on the model for the comparison of post-surgery mortality among the four

275

risk groups.

276

We repeated the analysis using 6.0cm and 6.5cm as AAA size cutoff and found similar results.

277

Both results showed non-significant interaction between aneurysm size and patient age, with a

278

significant average age effect and a significant aneurysm size effect (data not shown).

279

Re-intervention

280

Compared with patients ≤80 and AAA ≤5.5cm, there were no statistically significant differences

281

in the hazard ratios for re-intervention among the patient exposure groups (Table 3, 4). The

282

interaction between aneurysm size and patient age was not significant (Z=0.85, p=0.398). There

283

was no significant difference in re-intervention risk for patients over 80 years of age compared to

284

the younger counterpart in size matched groups, with aneurysm size ≤5.5cm (p=0.490), those

285

with aneurysm size >5.5cm (p=0.603), and overall (p=0.859). Furthermore, there was no

286

significant difference in re-intervention risk for patients with aneurysm size >5.5cm compared to

287

the smaller aneurysm size counterpart in age matched groups, ≤80 years of age (p=0.087), >80

288

years of age (p=0.878), and overall (p=0.135). Figure 2 shows adjusted survival curves based on

289

the model for the comparison of re-intervention among four risk groups.

290

We repeated the analysis using 6.0cm and 6.5cm as AAA cutoff and found similar results, i.e. no

291

significant interaction, age effect, and aneurysm size effect was found.

AC C

EP

TE D

M AN U

SC

RI PT

271

292 13

ACCEPTED MANUSCRIPT

Discussion

294

Analysis of the KPSGR showed that age and aneurysm size influence mid-term mortality

295

following elective EVAR but not re-intervention outcomes. These findings are independent of

296

patient sex, race, medical co-morbidity, and the device implanted. Neither age nor aneurysm size

297

amplified the risk conferred by the other. For mortality after EVAR, age over 80 years conferred

298

a 2.53-fold risk compared to age of ≤80 years; pre-operative aneurysm diameter exceeding 5.5

299

cm conferred a 1.75-fold higher risk compared to aneurysm diameter of ≤5.5 cm. The highest

300

risk group was patients >80 years with AAA >5.5cm, whose risk of mortality was 3.87-fold that

301

of the younger group with smaller AAA. Neither age nor aneurysm size conferred a higher risk

302

of re-intervention.

303

The landmark trials that established size criteria for AAA repair oftentimes do not address the

304

“real world” experiences of practicing vascular surgeons. For example, in the ADAM trial,

305

patients undergoing treatment were younger than 80 years of age and underwent only open

306

repair, so the results are not directly comparable to EVAR or outcomes in older patients.(4)

307

Furthermore, registry data worldwide show that about half of endovascular repairs are now

308

performed on patients with aneurysms smaller than 5.5 cm.(17-20) Patients with aneurysms

309

≥5.5cm at time of EVAR have been shown to be older, have worse surgical risk profiles, and

310

higher post-operative rates of type I endoleak, complications, and mortality.(18, 19)

311

Others have examined the effect of AAA diameter on mortality after EVAR with findings similar

312

to ours. A recent University of Alabama review showed superior outcomes in long-term mortality

313

and secondary interventions for EVAR when performed on aneurysms smaller than 5.0 cm

314

compared to medium (5.0-5.9 cm) and large (≥6.0cm) AAA.(20) Age adjusted, all-cause ten-year

AC C

EP

TE D

M AN U

SC

RI PT

293

14

ACCEPTED MANUSCRIPT

mortality was 28% for small AAA versus 36.9% and 50.2% in medium and large aneurysms.(20)

316

The AneuRx device trial data showed five-year overall mortality of 31% in small versus 32% and

317

49% in medium and large aneurysms.(19) In addition to the impact on all-cause mortality,

318

increased AAA size in the EUROSTAR trial was associated with significantly reduced freedom

319

from aneurysm related death at four years: 97%, 95% and 88% in small, medium and large AAA

320

respectively.(18) This has been cited as a basis for lowering the threshold for AAA repair, but

321

when subjected to the rigor of a prospective randomized study, both the CAESAR and PIVOTAL

322

trials showed no difference in long-term outcomes for patients with AAA between 4.0 and 5.5

323

cm in diameter whether managed with EVAR or surveillance.(14, 15)

324

The ideal size threshold for repair in elderly and high-risk populations remains undetermined.

325

The decrease in peri-operative morbidity and mortality in the EVAR era does not change the fact

326

that rupture risk for AAA ≤5.5 cm is low, a strategy of surveillance is safe, and long-term

327

outcomes appear unimproved by early repair.(3, 14, 15, 29) The current study does not resolve

328

the issue but provides additional evidence that when controlling for other variables, larger AAA

329

size alone reduces survival after EVAR.

330

Data regarding the effect of age and AAA size on re-intervention rates is inconsistent. Keith, et al

331

found that AAA size increased the risk of secondary interventions by 4.74-fold for large

332

aneurysms and 2.32-fold for medium aneurysms compared to the smallest aneurysm group.(19,

333

20) The risk of all-type endoleak was not significantly different between groups but type I

334

endoleaks were significantly more frequent in the largest diameter group (14.8% for large vs

335

7.0% medium and 5.1% small, P=0.001).(20) Similarly, EUROSTAR data showed reduced

336

freedom from type I and III endoleaks in large AAA (≥6.5cm) compared to small (4.0-5.4cm)

337

and medium AAA (5.5-6.4cm) but no difference in rates of other complications such as device

AC C

EP

TE D

M AN U

SC

RI PT

315

15

ACCEPTED MANUSCRIPT

migration, kinking, and limb stenosis.(18) Han, et al found no difference in reintervention rates

339

between octogenarians and younger patients in their large meta-analysis of over 25,000 AAA

340

EVAR cases with variable follow-up from 12-60 months.(16) Others have not found an increased

341

re-intervention rate in AAA exceeding 5.5cm.(18, 22, 30) This KPSGR analysis found that

342

neither age nor AAA size influenced re-intervention rates but the effect of confounding variables

343

was reduced with propensity score weighting, which was not used in the previously described

344

studies. Additionally, this series is more contemporary (operations performed between 2010 and

345

2014) and surgeons had the advantage of access to a greater variety and sizes of endografts.

346

Finally, knowledge gained from EVAR experience in large aneurysms may have reduced the

347

number of technical problems requiring re-intervention.

348

Reduced peri-operative complication rates with EVAR have expanded the age criteria for AAA

349

repair and increased examination of outcomes in the elderly. ACS NSQIP results showed 30-day

350

mortality for octogenarians and nonagenarians undergoing elective EVAR was 2% and 3.8%,

351

respectively, compared to 0.5-1.2% for patients younger than 80.(23) In this series, 30-day

352

mortality was comparable at 1.0% for those ≤80 and 2.0% for those >80. Unadjusted survival at

353

4 years was 48.9% in those >80 years with an AAA >5.5 cm and 65.6% with an AAA ≤5.5 cm.

354

Age over 80 alone, while controlling for other variables, increased the risk of late mortality 2.5

355

fold.

356

A goal of this study was to analyze the KPSGR database and identify any potential threshold of

357

age, AAA size or their combination at which repair of the AAA by EVAR might be unjustifiable.

358

Although when controlling for confounders, age and AAA size were independently confirmed to

359

reduce survival after EVAR, a threshold was not identified. The EVAR 2 trial examined this

360

question in patients unfit for OSR with AAA >5.5cm and found no survival benefit to EVAR

AC C

EP

TE D

M AN U

SC

RI PT

338

16

ACCEPTED MANUSCRIPT

over surveillance.(31) One might speculate that a strategy to repair AAA at a size ≤5.5 cm,

362

particularly in those over 80 years, would improve long-term survival following EVAR, however

363

rupture risk for aneurysms smaller than 5.5 cm is less than 1-2%, which is lower than the peri-

364

operative mortality for elective EVAR in octogenarians.(13, 23, 32)

365

Reported outcomes after EVAR in octogenarians vary considerably. The largest group was 697

366

octogenarians in the EUROSTAR registry from 1996-2004 whose 8-year survival rate was

367

64%.(7, 11) In contrast, 322 octogenarians with a mean AAA diameter of 6.2 cm who underwent

368

EVAR from 1997-2007 at Mount Sinai Hospital in New York had an overall five year mortality

369

of 72.6%.(10) The EUROSTAR long-term survival rate is comparable to the KPSGR four-year

370

unadjusted survival rates, while the mortality in the Mount Sinai study exceeds that of the highest

371

risk subset in this analysis. Differences in patient selection may account for similar short-term

372

but variable mid and late-term outcomes reported in the different studies, with higher risk

373

patients not being offered EVAR in some settings. It is not surprising that age over 80 alone

374

results in a shortened life expectancy after EVAR but it is important to note that our analysis

375

indicates this effect is independent of medical co-morbidities. The group in our study with the

376

worst outlook after EVAR were those over 80 with AAA >5.5 cm where the hazard ratio for

377

mortality was nearly 4 times that of the population under 80 with AAA ≤5.5 cm. Nevertheless,

378

the two-year survival in all four comparison groups was greater than 80% (Fig.1). When

379

considering an annual rupture rate as high as 10% with AAA size in the range of 5.5 – 6.9

380

cm,(33) many vascular surgeons contend AAA repair is indicated, irrespective of age or AAA

381

size, as long as there are no prohibitive comorbidities to significantly reduce life expectancy.

382

Strengths of our study include a four and a half year follow-up period with robust internal

383

validation of study data. The KPSGR includes prospectively collected information about patients,

AC C

EP

TE D

M AN U

SC

RI PT

361

17

ACCEPTED MANUSCRIPT

operations, surgeons and hospitals associated with mortality and re-intervention. Clinical

385

measures such as history of MI, smoking status, comorbidities, and presence of coronary artery

386

bypass graft, etc. were chart reviewed and validated, and were adjusted for in the analysis.

387

Moreover, our study is generalizable as large and representative samples of patients, surgeons

388

and hospitals were studied within multiple geographic regions. The population of this integrated

389

healthcare system has been shown to be socio-economically and demographically representative

390

of the geographical areas it covers,(34) making our findings most likely representative of other

391

patients in our regions.

392

This study represents a large, contemporary “real world” experience but has several limitations.

393

First, although the study duration is four and a half years, the mean follow-up is less than two

394

years. In addition, this registry does not include data on cause of death or long-term aneurysm

395

related mortality. Morphologic data regarding anatomic factors such as tortuosity and

396

calcification, more common in the elderly, are not available and could not be adjusted for in the

397

analysis. We also could not identify how many of the primary procedures were performed

398

outside the device instructions for use (IFU) or control for that confounder. A prior publication

399

from our Northern California group found that 41.9% of procedures were done outside of IFUs

400

although they noted no significant impact on survival, re-interventions, or endoleak status

401

compared to the IFU-adherent cohort.(35) This report also excludes the outcomes of OSR

402

performed during this period. Consideration of patient comorbidities is important during clinical

403

decision making. Frailty is associated with age, BMI, smoking status, renal insufficiency, peripheral

404

vascular disease, and arteriosclerotic cardiovascular disease. We included all these variables in the

405

model but did not use a specific frailty profile or index as a separate measure for statistical analysis.

406

Finally, despite innovative methods to encourage surgeons to enter data into the KPSGR, some

AC C

EP

TE D

M AN U

SC

RI PT

384

18

ACCEPTED MANUSCRIPT

patient data were missing and had to be estimated using statistical methods previously described.

408

We are limited in comparing risk groups among patients enrolled in the registry, namely patients

409

who undergo EVAR in our system based on the recommendations of surgeons and medical

410

advice.

411

Conclusion

412

In this analysis from the KPSGR, age >80 years and AAA diameter >5.5 cm were independent

413

predictors of reduced survival after EVAR, but the effect was not amplified when both were

414

present. Mortality risk due to aneurysm size effect alone was constant at any given age.

415

Furthermore, age >80 years or AAA size >5.5 cm did not increase risk of re-intervention. No

416

specific AAA size, patient age or combination thereof were identified that would contraindicate

417

AAA repair.

M AN U

SC

RI PT

407

421 422 423 424 425 426

EP

420

AC C

419

TE D

418

427 428

19

ACCEPTED MANUSCRIPT

Acknowledgement

430

This study is dedicated to the memory of Robert J. Hye, MD, who was an integral champion and

431

mentor for the development of Kaiser Permanente Endovascular Stent Graft Registry. We

432

acknowledge all the Kaiser Permanente surgeons who contribute National Implant Registries and

433

the Surgical Outcomes and Analysis Department, which coordinates Registry operations. A

434

special thank you for Mary-Lou Kiley, the Stent registry project manager, for overseeing the

435

completion of the study; Kristen J. Mays, the department research associate, for validating the

436

data from the registry for this study; Jon Javines, department programmer, for extracting the

437

electronic medical records’ data; and Donna Means Leck, department admininstrative research

438

analyst, for providing editorial assistance.

AC C

EP

TE D

M AN U

SC

RI PT

429

20

ACCEPTED MANUSCRIPT

References

439 1.

Raval MV, Eskandari MK. Outcomes of elective abdominal aortic aneurysm repair among the

441

elderly: endovascular versus open repair. Surgery. 2012;151(2):245-60.

442

2.

443

with abdominal aortic aneurysm (EVAR trial 1), 30-day operative mortality results: randomised

444

controlled trial. The Lancet. 2004;364(9437):843-8.

445

3.

446

repair of abdominal aortic aneurysm: A randomized trial. JAMA. 2009;302(14):1535-42.

447

4.

448

small abdominal aortic aneurysms. The New England journal of medicine. 2002;346(19):1437-44.

449

5.

450

abdominal aortic aneurysms in matched cohorts using propensity score modeling. J Vasc Surg.

451

2015;62(2):304-11.e2.

452

6.

453

in elderly patients. International journal of surgery (London, England). 2015;15:117-23.

454

7.

455

octogenarians: A systematic review. Journal of Vascular Surgery. 2008;47(3):676-81.

456

8.

457

abdominal aortic aneurysms in octogenarians. J Vasc Surg. 2008;47(5):924-7.

458

9.

459

endovascular abdominal aortic aneurysm repair in octogenarians based on the Endurant Stent Graft

460

Natural Selection Global Postmarket Registry (ENGAGE). J Vasc Surg. 2012;56(1):27-35.

461

10.

462

aneurysm repair in octogenarians and nonagenarians. J Vasc Surg. 2010;51(6):1354-9.

RI PT

440

Greenhalgh RM. Comparison of endovascular aneurysm repair with open repair in patients

SC

Lederle FA, Freischlag JA, Kyriakides TC, et al. Outcomes following endovascular vs open

M AN U

Lederle FA, Wilson SE, Johnson GR, et al. Immediate repair compared with surveillance of

Huang Y, Gloviczki P, Oderich GS, et al. Outcome after open and endovascular repairs of

TE D

de Leur K, Flu HC, Ho GH, et al. Outcome of elective treatment of abdominal aortic aneurysm

EP

Henebiens M, Vahl A, Koelemay MJ. Elective surgery of abdominal aortic aneurysms in

AC C

Paolini D, Chahwan S, Wojnarowski D, et al. Elective endovascular and open repair of

Pol RA, Zeebregts CJ, van Sterkenburg SM, et al. Thirty-day outcome and quality of life after

Prenner SB, Turnbull IC, Malik R, et al. Outcome of elective endovascular abdominal aortic

21

ACCEPTED MANUSCRIPT

463

11.

Lange C, Leurs LJ, Buth J, et al. Endovascular repair of abdominal aortic aneurysm in

464

octogenarians: an analysis based on EUROSTAR data. J Vasc Surg. 2005;42(4):624-30; discussion

465

30.

466

12.

467

aneurysm: the Society for Vascular Surgery practice guidelines. J Vasc Surg. 2009;50(4 Suppl):S2-

468

49.

469

13.

470

in the UK Small Aneurysm Trial. The British journal of surgery. 2007;94(6):702-8.

471

14.

472

small aneurysm repair (CAESAR): results from a randomised trial. European journal of vascular and

473

endovascular surgery : the official journal of the European Society for Vascular Surgery.

474

2011;41(1):13-25.

475

15.

476

patients with small abdominal aortic aneurysms. J Vasc Surg. 2010;51(5):1081-7.

477

16.

478

in Octogenarians: Meta-analysis and Systematic Review. European journal of vascular and

479

endovascular surgery : the official journal of the European Society for Vascular Surgery.

480

2017;54(4):454-63.

481

17.

482

small versus large abdominal aortic aneurysm. J Vasc Surg. 2003;37(6):1206-12.

483

18.

484

outcome of endovascular aneurysm repair: does size matter? A report from EUROSTAR. J Vasc Surg.

485

2004;39(2):288-97.

486

19.

487

aneurysm diameter predict outcome? J Vasc Surg. 2006;44(5):920-29; discussion 9-31.

RI PT

Chaikof EL, Brewster DC, Dalman RL, et al. The care of patients with an abdominal aortic

SC

Powell JT, Brown LC, Forbes JF, et al. Final 12-year follow-up of surgery versus surveillance

M AN U

Cao P, De Rango P, Verzini F, et al. Comparison of surveillance versus aortic endografting for

TE D

Ouriel K, Clair DG, Kent KC, et al. Endovascular repair compared with surveillance for

EP

Han Y, Zhang S, Zhang J, et al. Outcomes of Endovascular Abdominal Aortic Aneurysm Repair

AC C

Ouriel K, Srivastava SD, Sarac TP, et al. Disparate outcome after endovascular treatment of

Peppelenbosch N, Buth J, Harris PL, et al. Diameter of abdominal aortic aneurysm and

Zarins CK, Crabtree T, Bloch DA, et al. Endovascular aneurysm repair at 5 years: Does

22

ACCEPTED MANUSCRIPT

488

20.

Keith CJ, Jr., Passman MA, Gaffud MJ, et al. Comparison of outcomes following endovascular

489

repair of abdominal aortic aneurysms based on size threshold. J Vasc Surg. 2013;58(6):1458-66.

490

21.

491

diameter influence clinical outcome of endovascular abdominal aortic repair: a 4-year EUROSTAR

492

experience. Ann Vasc Surg. 2005;19(6):755-61.

493

22.

494

long-term survival after endovascular aortic aneurysm repair. Vascular and endovascular surgery.

495

2012;46(7):530-5.

496

23.

497

Aneurysm Repairs: Should We Approach Octogenarians and Nonagenarians Differently? Journal of

498

the American College of Surgeons. 2012;215(5):690-701.

499

24.

500

medical record to optimize data capture for longitudinal outcomes in endovascular abdominal

501

aortic aneurysm repair. J Vasc Surg. 2015;61(5):1160-6.

502

25.

503

data. Medical care. 1998;36(1):8-27.

504

26.

505

Society: Series B (Statistical Methodology). 2014;76(1):243-63.

506

27.

507

specification. Statistical methods in medical research. 2007;16(3):219-42.

508

28.

509

1987 [

510

29.

511

with surveillance of small abdominal aortic aneurysms. The New England journal of medicine.

512

2002;346(19):1445-52.

RI PT

Waasdorp EJ, de Vries JP, Hobo R, et al. Aneurysm diameter and proximal aortic neck

SC

Tsilimparis N, Mitakidou D, Hanack U, et al. Effect of preoperative aneurysm diameter on

M AN U

Tsilimparis N, Perez S, Dayama A, et al. Age-Stratified Results from 20,095 Aortoiliac

TE D

Hye RJ, Inui TS, Anthony FF, et al. A multiregional registry experience using an electronic

Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative

EP

Imai K, Ratkovic M. Covariate balancing propensity score. Journal of the Royal Statistical

AC C

van Buuren S. Multiple imputation of discrete and continuous data by fully conditional

Rubin DB. Multiple Imputation for Nonresponse in Surveys New York: John Wiley & Sons;

Powell JT, Brady AR, Brown LC, et al. Long-term outcomes of immediate repair compared

23

ACCEPTED MANUSCRIPT

513

30.

514

abdominal aortic aneurysms. Ann Vasc Surg. 2011;25(3):306-14.

515

31.

516

patients physically ineligible for open repair. The New England journal of medicine.

517

2010;362(20):1872-80.

518

32.

519

surveillance. UK Small Aneurysm Trial Participants. Ann Surg. 1999;230(3):289-96; discussion 96-

520

7.

521

33.

522

in patients refusing or unfit for elective repair. Jama. 2002;287(22):2968-72.

523

34.

524

members of a large, integrated health care system: comparison with US Census Bureau data. Perm J.

525

2012;16(3):37-41.

526

35.

527

device instructions for use guidelines has no impact on outcomes. J Vasc Surg. 2015;61(5):1151-9.

RI PT

SC

Brown LC, Powell JT. Risk factors for aneurysm rupture in patients kept under ultrasound

M AN U

Lederle FA, Johnson GR, Wilson SE, et al. Rupture rate of large abdominal aortic aneurysms

Koebnick C, Langer-Gould AM, Gould MK, et al. Sociodemographic characteristics of

TE D

Walker J, Tucker LY, Goodney P, et al. Adherence to endovascular aortic aneurysm repair

EP

529

Greenhalgh RM, Brown LC, Powell JT, et al. Endovascular repair of aortic aneurysm in

AC C

528

Jim J, Rubin BG, Geraghty PJ, et al. Outcome of endovascular repair of small and large

24

ACCEPTED MANUSCRIPT

Table 1: Characteristics of patients underwent endovascular aneurysm repair, January 2010 June 2014 Age (years)

Gender: Male Race: White Asian Black Others BMI

≤5.5 N=193 82 83 86 (84.2± 3.0)

>5.5 N=265 82 85 87 (84.9± 3.1)

N=1967 69 74 80 (74.4± 7.9)

84% ( 651)

91% ( 664)

67% ( 130)

81% ( 215)

84% (1660)

72% ( 557) 8% ( 62) 8% ( 59) 12% ( 93) 25.0 27.7 30.6 (27.9± 4.7)

77% ( 556) 8% ( 56) 6% ( 43) 9% ( 65) 24.7 27.9 31.4 (28.3± 5.2)

69% ( 132) 15% ( 28) 5% ( 10) 11% ( 21) 23.3 26.2 28.8 (26.2± 4.2)

79% ( 205) 11% ( 28) 3% ( 8) 7% ( 19) 23.1 25.2 27.8 (25.5± 3.6)

75% (1450) 9% ( 174) 6% ( 120) 10% ( 198) 24.3 27.2 30.3 (27.6± 4.8)

31% ( 223) 7% ( 50) 62% ( 448)

29% ( 200) 7% ( 45) 64% ( 445)

9% ( 16) 30% ( 53) 60% ( 105)

11% ( 26) 14% ( 34) 75% ( 180)

25% ( 465) 10% ( 182) 65% (1178)

27% ( 210) 20% ( 146) 75% ( 559)

30% ( 222) 28% ( 190) 72% ( 490)

23% ( 44) 32% ( 60) 79% ( 147)

20% ( 54) 40% ( 99) 75% ( 185)

27% ( 530) 27% ( 495) 74% (1381)

20% ( 89) 27% ( 67) 11% ( 48) 7% ( 29) 4% ( 18)

23% ( 96) 37% ( 89) 16% ( 68) 9% ( 39) 3% ( 11)

21% ( 26) 31% ( 22) 15% ( 19) 13% ( 16) 3% ( 4)

29% ( 43) 42% ( 38) 18% ( 26) 10% ( 14) 3% ( 4)

22% (254) 33% (216) 14% (161) 9% ( 98) 3% ( 37)

18% 21% 39% 22%

19% 18% 39% 24%

21% 10% 47% 23%

21% 15% 35% 28%

19% 18% 39% 24%

AC C

EP

TE D

Smoking Status: Active Never smoked Quit Co-morbidities: Diabetes1 Renal insufficiency2 PVD2 Risk Factor: PVD/PAD/ASCVD History of MI Coronary artery bypass graft Congestive heart failure, EF<50% Cardiac chest pain, angina Implant: Cook - Zenith EndoLogix (AFX or IntuiTrak) Gore - Excluder Medtronic - Endurant

>5.5 N=731 68 73 77 (72.0± 5.7)

SC

Age

Max AAA diameter (AP or TR) 532 533 534 535

Total

≤5.5 N=778 67 71 75 (70.6± 6.3)

M AN U

AAA size (cm)

>80

≤80

RI PT

530 531

(143) (163) (302) (170)

4.9 5.2 5.4

5.7

(137) (133) (282) (179)

6.0 6.7

5.0

( 40) ( 19) ( 90) ( 44)

5.3 5.4

5.8

( 56) ( 40) ( 94) ( 75)

6.2 6.9

5.2

(376) (355) (768) (468)

5.6 6.1

(5.0±0.5) (6.3±0.8) (5.1±0.5) (6.4±0.8) (5.7±1.0) a b c represent the lower quartile a, the median b, and the upper quartile c for continuous variables. x±s represents X ̄ ±1 SD. N is the number of non–missing values. Numbers after percents are frequencies. 1 co-morbidities at the time of surgery 2 co-morbidities within one preceding year of operation 25

ACCEPTED MANUSCRIPT

EP

TE D

M AN U

SC

RI PT

Missing pattern: Race(N=25,1.3%), BMI (N=14,0.7%), smoking status (N=142,7.2%), renal insufficiency and PVD (N=103,5.2%), PVD/PAD/ASCVD (N=825,41.9%), history of MI (N=1321,67.2%), coronary artery bypass graft (N=826,42.0%), congestive heart failure, EF<50%(N=825,41.9%), cardiac chest pain, angina(N=902,45.9%) AAA = abdominal aortic aneurysm, BMI = body mass index, PVD= peripheral vascular disease, PAD = peripheral artery disease, ASCVD = arteriosclerotic cardiovascular, MI= Myocardial Infarction, EF = Ejection Fraction, AP = Anteroposterior, TR = transverse

AC C

536 537 538 539 540 541 542 543 544 545

26

ACCEPTED MANUSCRIPT

Table 2: Post-operation death hazard ratio (95% confidence interval) by age, aneurysm size, and average effects (N=1967)

Survival model1,2

Interaction3:

>80 vs ≤80 >80 vs ≤80 >80 vs ≤80

AAA size effect (≤5.5cm as ref. group): Average of all patients Among age≤80 patients Among age>80 patients

----

<0.001 <0.001 <0.001

0.67 (0.31, 1.45)

0.309

----

2.53 (1.73, 3.70) 2.98 (1.82, 4.90) 2.00 (1.10, 3.61)

<0.001 <0.001 0.022

>5.5 vs ≤5.5 >5.5 vs ≤5.5 >5.5 vs ≤5.5

1.75 (1.26, 2.45) 1.94 (1.32, 2.86) 1.30 (0.66, 2.54)

0.001 <0.001 0.445

TE D

mixed effects Cox model with propensity score weights was fitted Standard deviation of random effect (intercept): surgeon = 0.37, hospital =0.18 3 Interaction is the differential age effect between AAA size groups(or vice versa) 2

HR = hazard ratio CI = confidence interval AAA = abdominal artery aneurysm

EP

556

Age effect (≤80 as ref. group): Average of all patients Among AAA≤5.5cm patients Among AAA>5.5cm patients

1 Between-within

(Reference) 1.94 (1.32, 2.86) 2.98 (1.82, 4.90) 3.87 (2.47, 6.08)

--

AC C

548 549 550 551 552 553 554 555

--

p-value

SC

Test types

HR (95%CI)

RI PT

Risk groups Age AAA size (years) (cm) ≤80 ≤5.5 ≤80 >5.5 >80 ≤5.5 >80 >5.5

M AN U

546 547

27

ACCEPTED MANUSCRIPT

Table 3: Re-intervention hazard ratio (95% confidence interval) by age, aneurysm size, and average effects (N=1967)

Survival model1,2

Interaction3: Age effect (≤80 as ref. group): Average of all patients Among AAA≤5.5cm patients Among AAA>5.5cm patients

>80 vs ≤80 >80 vs ≤80 >80 vs ≤80

AAA size effect (≤5.5cm as ref. group): Average of all patients Among age≤80 patients Among age>80 patients

----

TE D

1 Between-within

(Reference) 1.38 (0.95, 1.99) 1.23 (0.68, 2.21) 1.15 (0.66, 2.03)

0.087 0.490 0.618

0.68 (0.28, 1.66)

0.398

----

1.04 (0.67, 1.62) 1.23 (0.68, 2.21) 0.84 (0.43, 1.64)

0.859 0.490 0.603

>5.5 vs ≤5.5 >5.5 vs ≤5.5 >5.5 vs ≤5.5

1.29 (0.92, 1.81) 1.38 (0.95, 1.99) 0.94 (0.42, 2.12)

0.135 0.087 0.878

--

mixed effects Cox model with propensity score weights was fitted Standard deviation of random effect (intercept): surgeon = 0.43, hospital =0.26 3 Interaction is the differential age effect between AAA size groups(or vice versa) 2

EP

HR = hazard ratio CI = confidence interval AAA = abdominal artery aneurysm

AC C

559 560 561 562 563 564 565 566 567 568

--

p-value

SC

Test types

HR (95%CI)

RI PT

Risk groups Age AAA size (years) (cm) ≤80 ≤5.5 ≤80 >5.5 >80 ≤5.5 >80 >5.5

M AN U

557 558

28

ACCEPTED MANUSCRIPT

Table 4. Post-operation death and re-intervention cumulative incidence rate and number of events, by age, aneurysm size, and average effects (N=1967)

Age (years)

>5.5 N=731

≤5.5 N=778

Outcomes Mortality Re-intervention

3.3% (51) 4.6% (67)

6.2% (86) 6.0% (78)

≤5.5 N=193

>5.5 N=265

RI PT

AAA size (cm)

9.9% (37) 5.4% (19)

14.6% (64) 5.9% (25)

* Rate (number of events) were shown, rate was calculated as number of events per 100 person years to account for different follow-up time. ** likelihood ratio test

EP

TE D

M AN U

AAA = abdominal artery aneurysm

AC C

574 575 576 577 578 579

>80

≤80

SC

569 570 571 572 573

29

P** <0.001 0.600

ACCEPTED MANUSCRIPT

580

Figure 1: Post-operative survival probability by aneurysm size and age

581

Footnotes: survival curves were adjusted based on the model for comparison

582

Figure 2: Post-operative re-intervention free probability by aneurysm size and age

584

Footnotes: survival curves were adjusted based on the model for comparison

RI PT

583

585

AC C

EP

TE D

M AN U

SC

586

30

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT