Contemporary Incidence, Outcomes and Survival Associated with EVAR Conversion to Open Repair Among Medicare Beneficiaries R1WC: 348/3535

Abstract

Objective:

The widespread application of endovascular abdominal aortic aneurysm repair (EVAR) has ushered in an era of requisite post-operative surveillance and the potential need for re-intervention. The national prevalence and results of EVAR conversion to open repair, however, remain poorly defined. The purpose of this analysis was to define the incidence of open conversion and its associated outcomes.

Methods:

The SVS VQI EVAR registry linked to Medicare claims via Vascular Implants Surveillance and Interventional Outcomes Network (VISION) was queried for open conversions after initial EVAR procedures from 2003–2016. Cumulative conversion incidence within up to five years following EVAR and outcomes following open interventions were determined. Multivariable logistic regressions were used to identify independent predictors of conversion and mortality.

Results:

Among 15,937 EVAR patients, 309 (1.9%) underwent an open conversion: 43% (n=132) early (<30-days) and 57% (n=177) late (>30-days). The longitudinally observed rate of conversion was constant over time, as well as by geographic region. Independent predictors of conversion included female sex (HR 1.49, p<.001), aneurysm diameter >6.0 cm at the time of index EVAR (HR 1.74, p<.001), non-elective repair (compared to elective presentation: HR 1.72, p<.001), and aorto-uni-iliac repairs (HR 2.19, p<.001). In contrast, adjunctive operative procedures such as endo-anchors or cuff extensions (HR 0.62, p=.06) were protective against long-term conversion. Both early (HR 1.6, p<.001) and late (HR 1.26, p=.07) open conversions were associated with significant 30-day (total cohort-15%) and 1-year mortality (total cohort-25%). Patients undergoing open conversion experienced high rates of 30-day readmission (42%) and cardiac (45%), renal (32%), and pulmonary (30%) complications.

Conclusions:

This large registry based analysis is among the first to document the incidence and outcomes for open conversion after EVAR in a national cohort with long-term follow-up. Importantly, women, patients with large aneurysms, and complex anatomy, as well as urgent/emergent EVARs are at increased risks for open conversion. It appears that more conversions are performed in the early post-operative period, despite perceptions that conversion is a delayed phenomenon. In all instances, conversion is associated with significant morbidity and mortality and highlights the importance of appropriate patient selection at the time of index EVAR.

Table of Contents Summary

This retrospective VQI-VISION Medicare claims linkage analysis of EVARs determined the national incidence of conversion to open repair. The study suggests that 2% of EVAR patients will undergo subsequent conversion and key predictors are identified that portend the highest subsequent probability of conversion to open repair.

Introduction

The therapeutic landscape of abdominal aortic aneurysm (AAA) repair has continued to evolve since the advent of endovascular (EVAR) techniques.¹ In the early 2000s, EVAR utilization surpassed open repair as the predominant AAA treatment modality and not surprisingly, has remained the more prevalent operative approach to date. Previously, we demonstrated that ~80% AAAs are now repaired with an endograft among Medicare beneficiaries.² Moreover, EVAR utilization has similarly surpassed >50% of all intact AAA repairs in various other countries, including Australia, Sweden, Italy, and the United Kingdom, in part reflecting current societal consensus guidelines which recommend EVAR as the preferred treatment modality in patients with both suitable anatomy and reasonable life expectancy.^{3, 4}

Despite these secular trends, EVAR has also been shown to be associated with an increased risk of reintervention over time, due to a myriad of underlying etiologies including various modes of graft failure and/or aneurysm expansion among others. Notably, contemporary rates of reintervention have been documented to be 15% - 33% over three to ten years post-implantation.⁵ Among the gamut of reinterventions, EVAR conversion to open repair remains a significant undertaking likely reflecting an underlying failure of the index EVAR intervention. It has been estimated that approximately 2% of EVAR patients may require conversion to open repair due to graft failure, infection, graft thrombosis, persistent endoleak leading to aneurysm expansion, and/or late rupture, among other indications.^{6, 7} In fact, one institution recently reported that EVAR conversion has become an increasingly common indication for any form of open AAA surgery and increased from 9% prior to 2009 to 27% in 2010–2019.⁸ Unfortunately, small single institution case series of EVAR conversions remain limited in scope and are likely underpowered to yield any definitive clinical insights about national trends or temporal variation in prevalence.

Therefore, the purpose of this analysis was to examine the national incidence of EVAR conversion to open AAA repair using a large validated data source. Furthermore, we sought to identify patient, procedure and anatomic variables associated with conversion to open repair and document procedure associated outcomes.

Methods

Human Subjects Protection

We obtained approval from the Institutional Review Board at Dartmouth Hitchcock Medical Center and Weill Cornell Medical Center for this project. All datasets were maintained on a secure server in accordance with CMS data management regulations. The need for consent was waived since no direct patient contact or harm resulted from the study.

Registry Data Description

The Vascular Implant Surveillance and Interventional Outcomes Network (VISION) Coordinated Registry Network (CRN) is a collaboration between the Society for Vascular Surgery Vascular Quality Initiative (SVS VQI), the US Food and Drug Administration (FDA), and various stakeholders including industry.⁹ It merges VQI patient-level data for vascular procedures with Medicare claims data. The strategic coordination of registries with other data sources, including claims, has been promoted by the FDA and the Health and Human Services (HHS) Office of the Assistant Secretary for Planning and Evaluation (ASPE) using the PCOR Infrastructure and Technology Innovation through Coordinated Registry Networks (CRN) Community of Practice Initiative (https://aspe.hhs.gov/bridging-pcor-infrastructure-and-technology-innovation-through-coordinated-registry-networks-crn-community-practice).

This combined effort enables long-term surveillance of vascular implants and permits enhanced longitudinal follow-up accordingly. Patients were matched between the SVS VQI EVAR registry and claims using direct identifiers through the CMS contractor. When direct identifiers were not available, patients were matched using key data components including the patient’s date of birth, sex, zip code, procedure date, and identifier or state of the facility where the procedure was performed. This previously validated, indirect matching algorithm has achieved an over 90% success rate and over 99% accuracy, and provides reliable outcomes at five-year follow-up.^10–12

Cohort

Accordingly, we queried the EVAR dataset of the VQI-VISION collaboration. The linked data contained patients who underwent AAA repair with an aortic endograft implant and were matched to their Medicare claims. All patients who underwent EVAR (including elective, urgent, and emergent presentations) from 2003–2016 were eligible for inclusion. Patients with mycotic aneurysms, aorto-enteric fistula/erosion were not included in the analysis. Patients who were not enrolled in Medicare part A and part B at the time of the index EVAR procedure were excluded as they lacked complete outcomes data.

Variables and Outcomes

The index operation was defined as the initial EVAR implant and was marked as time zero in the analysis. From the VISION dataset, we collected patient demographics (age, sex, race, geographic region), comorbidities within 1 year prior to the index EVAR, and procedural information including aortic sac diameter at time of index EVAR, endograft type, urgency level, and adjunct device utilization (defined as endo-anchors and/or cuff extensions with the initial procedure). We analyzed annual surgeon EVAR volume, defined as the number of EVAR procedures performed by each surgeon during the calendar year of each patient’s index procedure. Surgeon experience was defined as how long a provider has been in practice contributing data to the VQI EVAR registry.

The primary outcome of the study was EVAR conversion to open repair, defined as any open AAA operation that occurred during the initial hospitalization or after discharge from the index EVAR procedure. EVAR conversion to open was identified from Medicare inpatient and physician-billed claims with ICD9 and ICD10 codes and from VQI recorded conversions. The secondary outcome was overall mortality following EVAR, identified from Medicare Master Beneficiary Summary File. When analyzing EVAR conversion events, we censored patients at death, at the end of the Medicare part A and part B entitlement, at the end of the study period (Dec 31, 2016), or at the end of 5 years following EVAR (VISION Medicare linked claims), whichever was the earliest. When analyzing mortality, we censored patients at the end of the Medicare part A and part B entitlement and the end of the study period, whichever occurred first. Primary and secondary outcomes were analyzed for all patients.

Among the subgroup of patients who underwent EVAR conversion to open repair, we additionally analyzed the length of stay, discharge disposition status, and 30-day readmission after the hospitalization for the conversion to open AAA repair. We also examined the one-year risk of major pulmonary, cardiac, and renal failure complications after the conversion to open AAA repair. Complications and renal failure were identified using ICD9 and ICD10 diagnostic codes and are listed in the Appendix Table I.

Statistical analysis

We reported median with inter-quartile range for age and frequencies with percentages for categorical variables. Categorical variables were compared between patients experiencing open conversion after EVAR vs. those without conversion using the Chi-square tests; however, age was compared between the two groups using the Wilcoxon Rank-sum test. We evaluated the risk of conversion stratified by the year of the index EVAR, urgency level of the index EVAR, and geographic region using the Kaplan-Meier survival method and the Log-rank test. We used a cox regression model to assess independent risk predictors of undergoing an open conversion. Candidate variables for the Cox model were all baseline characteristics. When assessing independent risk predictors of all-cause mortality, we included conversion to open repair as a time-dependent variable in the Cox model and all baseline characteristics. In the secondary analysis focusing on patients with EVAR conversion to open repair, categorical outcomes were summarized with frequencies and counts. Length of stay was summarized with median and inter-quartile range. All analyses were carried out using SAS version 9.4(SAS Institute Inc., Cary, NC.)

Results

Incidence of Conversion to Open AAA Repair

After matching SVS VQI EVAR patients with their Medicare claims, 15,937 patients met inclusion criteria for analysis. Among these, 1.9%(n=309) underwent conversion to open AAA repair at some point following their index EVAR within five years. Within this subset of patients, 43%(n=132) experienced early conversion(e.g. < 30-days) after their index EVAR, while 57%(n=177) underwent conversion 30 days or later after the initial endograft implant(Appendix Figure).

Baseline Characteristics

The overall study cohort was predominantly male(79.2%, n=12,627), white(92.1%, n=14,685) and the median age was 76 years. Of the EVAR procedures, 6.2%(n=982) were performed between 2003–2009 and 93.8%(n=14,955) occurred from 2010–2016, reflecting the initial advent, growth, and subsequent patient accrual in the VQI registry.

EVAR conversion patients were more commonly female(27.8%, n=86/309) compared to women who did not undergo conversion(20.6%, n=3,224/15,628, p<.001). Otherwise, patient demographics and comorbidities were similar between groups in terms of age, race, hypertension, coronary artery disease, COPD, diabetes mellitus, and congestive heart failure. However, only chronic kidney disease was less prevalent among patients who underwent conversion(16.8%, n=52) compared to those who did not(21.5%, n=3,366, p=.05)(Table I).

Table I.

Patient Characteristics

Variable, No. (%)	No Conversion N=15,628	Conversion to Open N=309	p-value
Age, median years (IQR)	76 (70–82)	76 (71–82)	.58
Female sex	3224 (20.6%)	86 (27.8%)	<.001
Black Race	711 (4.5%)	17 (5.5%)	.58
Hypertension	13408 (85.8%)	259 (83.8%)	.33
Coronary Artery Disease	9052 (57.9%)	169 (54.7%)	.25
COPD	6108 (39.1%)	126 (40.8%)	.55
Diabetes Mellitus	4414 (28.2%)	78 (25.2%)	.25
Chronic Kidney Disease	3366 (21.5%)	52 (16.8%)	.05
Congestive Heart Failure	2561 (16.4%)	55 (17.8%)	.51

IQR, Interquartile range; COPD, chronic obstructive pulmonary disease

There were significant and notable index procedural differences among patients who subsequently experienced EVAR conversion and those who did not(Table II). Subjects who underwent conversion more commonly had aneurysms measuring 6.0cm or larger at the time of their index EVAR(>36% versus 24.5%, p<.001). Notably, conversion patients were more likely to have received their EVAR in a non-elective(e.g. symptomatic/urgent or emergent/rupture) setting(22.3% versus 12.4%, p<.001). Additionally, open conversion patients were twice as likely to have an aorto-uni-iliac endograft during the index EVAR as compared to those without conversion(13.3% versus 6.9%, p<.001). Furthermore, surgeon annual EVAR volume was not significantly associated with the risk of conversion to open repair. Interestingly however, surgeon experience did appear to paradoxically correlate with receiving EVAR conversion. Specifically, 8.1% of converted EVARs versus 4.8% of non-converted EVARs were originally performed by surgeons with ≥ 7-years of practice experience(p=.009).

Table II.

Procedural Characteristics for Index EVAR

Variable, No. (%)	No Conversion N=15,628	Conversion to Open N=309	p-value
AAA size at Time of Index EVAR			<.001
≤ 6.0cm	11545 (73.9%)	186 (60.3%)
> 6.0cm	3824 (24.5%)	112 (36.2%)
Missing	259 (1.6%)	11 (3.5%)
Non-elective (urgent/emergent) EVAR	1945 (12.4%)	69 (22.3%)	<.001
Aorto Uni-Iliac Endograft	1080 (6.9%)	41 (13.3%)	<.001
Surgeon Annual EVAR Volume			0.8
• 1–25 EVARs/yr.	13746 (88.0%)	274 (88.7%)
• 26–50 EVARs/yr.	1790 (11.5%)	NR
• 51–75 EVARs/yr.	92 (0.6%)	NR
Surgeon Experience/VQI Participation			.009
• 0–6 years	14872 (95.2%)	284 (91.9%)
• ≥ 7 years	756 (4.8%)	25 (8.1%)

NR, not reported due to low numbers within this cohort

Cumulative Incidence of Conversion by EVAR Procedure Years, Geographic Region, and Urgency Level

Figure 1a depicts the cumulative incidence of EVAR conversion stratified by time-interval. We initially hypothesized that EVARs performed during an earlier time period might be more vulnerable to failure for a myriad of reasons including early device iterations and fewer available adjuncts among other underlying causes. However, while we determined that the absolute overall incidence of conversion over time was greater for EVAR procedures performed from 2003–2009(5%) when compared to the later time period(2010–2016, 2%, p<.05), but the relative change for each time interval was similar, with an incidence of ~2% over 5 years.

Figure 1. Cumulative Incidence of Conversion to Open Repair after EVAR.

a. Stratified by Time Interval when Index EVAR was performed

b. Stratified by Geographic Region

c. Stratified by Elective vs. Urgent/Emergent Index EVAR

Correspondingly, we determined that the cumulative incidence of open conversion was similar when stratified by geographic region. Specifically, the curves in Figure 1b represent the incidence of conversion among patients residing in the Midwest, Northeast, South, and West regions of the United States and demonstrate that no region was an outlier for the likelihood of EVAR conversion during follow-up.

Importantly, the cumulative rate of conversion was significantly different among patients who received their EVARs electively versus those who underwent EVAR non-electively, including symptomatic/urgent or emergent/ruptured presentations(elective, 2% vs. non-elective 5% at 5-years; Figure 1c). In fact, the cumulative incidence of conversion to open repair increased significantly following non-elective EVARs and continued to diverge from the elective cohort over time(p<.001).

Predictors of EVAR Conversion

Cox regression analysis revealed several variables associated with the risk of undergoing EVAR conversion(Table III). Patients who underwent aorto-uni-iliac endograft placement were more than two-fold as likely to undergo conversion compared to those who received a standard bifurcated endograft(hazard ratio; HR 2.19, p<.001). In addition, patients treated by more experienced surgeons(≥ 7 years of practice) were twice as likely to undergo open conversion (HR 1.96, p<.001) when compared to subjects receiving implants from less experienced surgeons. Moreover, EVARs performed in the setting of larger aneurysms(>6.0cm) were 74% more likely to experience conversion than smaller aneurysms measuring less than 6.0cm in maximal diameter(HR 1.74, p<.001). Non-elective EVARs were significantly more likely to undergo subsequent conversion compared to elective implants(HR 1.72, p<.001). Interestingly, women were at a higher risk of conversion(HR 1.49, p<.001) when compared to men. In contrast, the use of adjunctive devices, such as endo-anchors and/or extension cuffs during the index EVAR, appeared to reduce the likelihood of conversion over time(HR 0.62, p=.06).

Table III.

Cox Proportional Hazards Model for Predictors of Conversion to Open Repair

Predictors of Open Conversion	Hazard Ratio	95% CI	p-value
Aorto Uni-Iliac Endograft	2.19	1.56–3.07	<.001
Surgeon Experience ≥ 7 years in practice	1.96	1.24–3.12	<.001
AAA >6.0cm at Time of Index EVAR	1.74	1.35–2.23	<.001
Non-Elective (urgent/emergent) EVAR	1.72	1.28–2.3	<.001
Female sex	1.49	1.15–1.94	<.001
Use of Adjunct (Endoanchors, cuff)	0.62	0.38–1.02	0.06
Procedure in 2010–2016 vs. 2003–2009	0.49	0.33–0.71	<.001

Non-significant variables were: all remaining patient demographic and comorbidity characteristics, geographic region of index EVAR procedure, & annual surgeon EVAR volume.

Outcomes following Conversion to Open Repair

We next performed a secondary analysis of patients who underwent conversion to open AAA repair(Table IV). In-hospital mortality in this cohort after the open AAA repair was 10.7%(n=33). More specifically, 14.6%(n=45) died within 30-days, 18.8%(n=58) died within 90-days, and 24.9%(n=77) died within one year of their conversion operation. The median length of stay after the open conversion was 6.5 days. Following the open AAA repair conversion procedure, 33.7%(n=104) of patients were discharged to home with the remainder of the cohort discharged to a facility with some level of nursing assistance, while the 30-day readmission rate was 40.8%(n=126). Overall, among the aggregate of elective and non-elective conversions, complications were prevalent. Specifically, 44.7%(n=138) suffered cardiac complications, 32%(n=99) had temporary or permanent renal failure, and 30.4%(n=94) experienced a pulmonary complication following conversion to open repair after EVAR. The subgroup analysis for outcomes after elective and non-elective EVAR conversion are featured in the Appendix Table II. It is important to highlight that the 30-day mortality rate for non-elective conversion was not surprisingly 3-fold higher than elective repair (28% vs. 8%, p<.001).

Table IV.

Outcomes after Conversion to Open Repair

Outcome, No. (%)	All Conversions N=309	Early Conversion < 30 Days (N=170, 55%)	Late Conversion ≥ 30 Days (N=139, 45%)	p-value
Mortality
In-hospital	33 (10.7%)	19 (11.2%)	14 (10.1%)	0.75
30-day	45 (14.6%)	26 (15.3%)	19 (13.7%)	0.69
90-day	58 (18.8%)	34 (20.0%)	24 (17.3%)	0.54
1-year	77 (24.9%)	46 (27.1%)	31 (22.3%)	0.34
LOS, days Median (IQR)	6.5 (3,10)	4 (2, 9)	8 (6, 13)	<0.001
30-day Readmission	126 (40.8%)	65 (38.2%)	61 (43.9%)	0.31
Discharge to Home	104 (33.7%)	66 (38.8%)	38 (27.3%)	0.03
Complications
Cardiac	138 (44.7%)	75 (44.1%)	63 (45.3%)	0.83
Renal Failure	99 (32%)	48 (28.2%)	51 (36.7%)	0.11
Pulmonary	94 (30.4%)	48 (28.2%)	46 (33.1%)	0.36

Cumulative Risk and Predictors of Mortality after EVAR

Our secondary outcome was all-cause mortality following EVAR. This endpoint occurred in 25.1%(n=3,999) of the patients. The single most important predictor for all-cause mortality among all EVAR patients was conversion to open repair in the adjusted Cox regression model. Specifically, EVAR conversion patients experienced a more than 2.5-fold increased risk of death compared to patients who did not undergo conversion(HR 2.63, p<.001). The variable most protective against mortality appeared to be an elective EVAR(HR 0.55, p<.001). All other significant variables associated with mortality following EVAR are depicted in Table V. Notably, AAA diameter ≥6.0cm at time of the index EVAR was associated with a 40% increased likelihood of death(HR 1.41, p<.001) during follow-up. Several additional comorbidities were similarly associated with mortality following EVAR, including: congestive heart failure(HR 1.62, p<.001), COPD(HR 1.60, p<.001), and chronic kidney disease(HR 1.59, p<.001). Lastly, peripheral arterial disease was associated with a decreased mortality among EVAR patients(HR 0.73, p<.001).

Table V.

Cox Proportional Hazards Model for Mortality after EVAR

Predictors of Mortality after EVAR	Hazard Ratio	95% CI	p-value
Conversion to Open	2.63	2.33–2.97	<.001
Non-Elective (urgent/emergent) EVAR	1.8	1.6–2.0	<.001
Congestive Heart Failure	1.62	1.50–1.74	<.001
COPD	1.60	1.50–1.71	<.001
Chronic Kidney Disease	1.59	1.47–1.71	<.001
AAA >6.0cm at Time of Index EVAR	1.41	1.32–1.51	<.001
Aorto Uni-Iliac Endograft	1.23	1.10–1.37	<.001
Procedure in 2010–2016 vs. 2003–2009	1.15	1.03–1.30	.02
Black Race	1.14	0.98–1.32	.08
Female sex	1.12	1.04–1.21	<.001
Age (per year)	1.05	1.05–1.06	<.001
Peripheral Arterial Disease	0.73	0.63–0.85	<.001

Discussion

This SVS VQI EVAR registry-Medicare claims linkage analysis using the Vascular Implant Surveillance and Interventional Outcomes Network(VISION) was comprised of nearly 16,000 patients who underwent an endograft repair of an AAA with extended longitudinal follow-up. Most importantly, it demonstrates that the cumulative incidence of EVAR conversion has been linear since 2003 and remains 1.9%. To our surprise, nearly half of all open conversions occurred within 30 days of the index EVAR, while the remainder occurred at a regular interval thereafter. Women, patients with larger aneurysms(6.0cm diameter or larger), complex aneurysm anatomy as reflected by use of AUI devices, and non-elective(urgent/emergent) EVAR procedures were associated with significantly higher rates of open conversion over time. Importantly, open conversion was associated with significant morbidity and mortality, with one in four conversion patients experiencing 1-year mortality.

The national VQI EVAR conversion incidence of 1.9% in our cohort mirrors rates published in other smaller institutional series. For example, one single-center report documented late conversion in 1.5%(n=16/1,060) of EVARs performed between 1999–2015.⁶ Several other reports presented conversion rates ranging from 0.8% to 5.9% respectively.^13–17 Our study findings corroborated these reports and further characterized the time to conversion. As with prior single center reports, the results documented in this analysis clearly illustrate the impact of surgeon/center practice variation on subsequent risk of EVAR conversion. Surprisingly, this study demonstrated that nearly half of all open conversions occurred within 30 days of the index EVAR. One potential explanation for this observation may be that some endografts were implanted urgently/emergently to provide a less invasive rescue therapy in a high-risk patient who might otherwise be a less suitable open repair candidate at the time of the non-elective presentation. Therefore, it is conceivable that some of these implants were performed as a temporizing intervention with the anticipated need for a more definitive open repair when the patient was clinically more stable. While this explanation does not entirely account for these findings, our observation that urgent/emergent EVARs were associated with significantly higher rates of open conversion does support this hypothesis as depicted in Figure 1c.

The risk factors associated with an increased likelihood of EVAR conversion in this analysis align with those listed in previous reports and has important implications for surveillance paradigms. For example, in an analysis by Columbo and associates, gender based disparity in post-EVAR reintervention rates has been previously documented.¹⁸ The same analysis also demonstrated that large aneurysms and urgent/emergent repairs were associated with increased conversion rates. Our analysis confirmed these findings in a larger national cohort. Accordingly, it would appear justifiable that female patients, subjects with large aneurysms, as well as those who undergo non-elective repairs warrant not only close but potentially different surveillance intensity following EVAR.

Interestingly, it appeared that adjunctive surgical interventions, such as the placement of endo-anchors and/or cuffs, conferred a longitudinal protective effect against open conversion over time. These interventions are frequently used in sub-optimal anatomic situations to facilitate the suitability and improve the durability of the index EVAR. Accordingly, one might infer that these patients would be at a higher risk for open conversion over time since these technical adjuncts were employed in the setting of potentially hostile EVAR anatomy, which may be outside the conventional instructions for use(IFU). ¹⁹ Our findings herein appear to refute this notion. Conceivably, these patients might have received more intensive surveillance and/or repeated intervention without conversion to open repair or were deemed too high-risk for endograft removal, but we are unable to definitively determine this from the claims linkage. Accordingly, it is important to emphasize that procedural adjuncts such as cuffs or endoanchors may not confer longitudinal protection against potential open conversion.

Overall, our analysis demonstrated that there is a persistent group of EVAR patients who will be at risk for reintervention over time and potential rupture. Accordingly, there appears to be a small but consistent subset of patients who will undergo EVAR conversion despite its associated morbidity and mortality.^{20, 21} Somewhat unexpectedly, it appears that reintervention rates following EVAR, even when performed under the most favorable conditions, have been reported to be ~15% - 30% depending on aneurysm size.²² It is worth re-emphasizing that our study demonstrated that larger aneurysm diameter at the time of EVAR was a significant predictor of future conversion to open repair. This observation thus warrants either enhanced pre-EVAR patient discussion disclosing this risk and close follow-up or consideration of open repair with the index presentation and avoidance of placing a stent-graft, especially in younger healthy subjects with AAAs ≥ 6.0cm

Our study has several intrinsic limitations. By nature of the Medicare claims linkage, certain anatomic and procedure-specific details are not available when examining events after the index EVAR procedure and thus cannot be included in the analysis. Medicare claims are predominantly for patients age 65 or older so the findings of this analysis may not be generalizable to younger subjects. Although the VQI collects granular anatomic details with the initial implant, due to data missingness, some anatomic variables were not accounted for in the risk adjustment. For example, proximal landing zone neck angulation, conical neck configuration, aortic mural thrombus or calcification, which are known to be associated with endoleak and subsequent need for reintervention, remained unquantified in our cohort.²³ Moreover, we are unable to determine device manufacturer or whether the index EVAR implant was placed within its intended and tested IFU.

Importantly, due to the limitations of the claims-based data, we were unable to ascertain the presence and type of endoleak after the first year of VQI EVAR surveillance and what the specific indication was for the conversion to open repair. Similarly, we were unable to specify whether the open conversion operation was for underlying infection, sac expansion, or another etiology. Likewise, we are unable to characterize the antecedent longitudinal history of re-interventions following the index EVAR which may or may not predispose patients to the risk of conversion. However, our finding that nearly 50% of EVAR conversion occurred within 1-month of the initial procedure would suggest that there are multiple risk factors which may place patients at risk for conversion irrespective of reintervention. In addition, it should be noted that we are unable to comment on the timing of the early conversion events. Thus, we are unable to determine if these were intraoperative or conversions at other time points within 30 days.

Further, technical details surrounding the conversion procedure are not available due to the lack of claims coding specificity for these variables. It is also worth noting that unexpectedly patients with documented PAD in the study cohort experienced decreased mortality. Unfortunately, we are unable to comment further on this counterintuitive finding. Nevertheless, this analysis constitutes the largest real world longitudinal study of EVAR procedures that characterizes open conversion events in the literature and offers the most robust follow-up to date among this complex group of patients.

Conclusions

In conclusion, this study is among the first to document the incidence and outcomes of EVAR conversion using a national cohort with long-term follow-up. Conversion to open repair poses significant risks to patients and is associated with demonstrably high morbidity and mortality rates. The higher than anticipated early conversion rate justifies vigilant early surveillance after EVAR, while the linear cumulative incidence of conversion over subsequent years highlights the importance of dedicated indefinite long-term EVAR surveillance. Our analysis underscores the importance of appropriate patient selection for EVAR in order to mitigate the potential for conversion to open repair. This study also illustrates the value of strategic linking of national registry data to claims to efficiently study long-term outcomes in the interventional setting.

ARTICLE HIGHLIGHTS.

Type of Research:

A retrospective analysis of prospectively collected national quality registry data from the Vascular Quality Initiative (VQI) with Medicare claims linkage after endovascular aortic aneurysm repair (EVAR).

Key Findings:

Among nearly 16,000 EVAR patients, 2% underwent subsequent open conversion. Female sex, large aneurysms (>6cm) and non-elective repairs were all independently associated with subsequent conversion. Open conversion occurred in both the early and late post-operative periods and was associated with significant 30-day and 1-year mortality.

Take Home Message:

Open aortic conversion is associated with significant morbidity and mortality after EVAR. Roughly 2% of EVAR patients will require open conversion and this may occur in both the early and late post-operative periods. Given the associated morbidity and mortality with conversion, prudent patient selection for EVAR repair remains critically important and should inform operative decision making, as well as postoperative surveillance paradigms.

Appendix Figure.

Study Cohort

Appendix Table I.

Diagnosis Codes Used to Define Complications

Complication	ICD9	ICD10
Pulmonary Complications	5121, 5184, 5185, 9973, 41511	J9589, J95821, J95822, J9600, J9601, J9602, J9690, J9691,J9692
Cardiac Complications	99702, 410, 4275, 4289, 42841, 42843, 42831, 42833, 42821, 42823, 998, 99801	I219,I509, I499
Temporary or Permanent Renal Failure	5849, V451, V560, 3995, V561, V568, 5498	N990, N178, N179

Appendix Table II.

Outcomes After Elective and Non-elective EVAR Conversion

Outcome, No. (%)	Elective conversion(N=177)	Non-elective conversion(N=132)
Mortality
In-hospital	13(7.34%)	17(12.88%)
30-day	15(8.47%)	37(28.03%)
90-day	22(12.43%)	43(32.58%)
1-year	32(18.08%)	45(34.09%)
LOS(IQR)	2(6,9)	5(0,11)
Readmission-30d	51(28.81%)	52(39.39%)
Discharge destination
Home	70(39.55%)	14(10.61%)
Short-term Hospital	1(0.56%)	1(0.76%)
Intermediate/Long-term	35(19.77%)	38(28.79%)
Home Healthcare	42(23.73%)	14(10.61%)
Dead	13(7.34%)	18(13.64%)
Missing	16(9.04%)	47(35.61%)
Complications
Pulmonary	44(24.86%)	51(38.64%)
Cardiac	77(43.50%)	68(51.52%)
Renal Failure	53(29.94%)	48(36.36%)