Follow-Up Compliance After Endovascular Abdominal Aortic Aneurysm Repair in Medicare Beneficiaries

Abstract

Objective

Life-long imaging follow-up is essential to the safe and appropriate management of patients who undergo endovascular abdominal aortic aneurysm repair (EVAR). We sought to evaluate the rate of compliance with imaging follow-up post EVAR and to identify factors associated with being lost to imaging follow-up.

Methods

We identified a 20% sample of continuously enrolled Medicare beneficiaries who underwent EVAR between 2001 and 2008. Using data through 2010 from Medicare Inpatient, Outpatient, and Carrier files, we identified all abdominal imaging studies that may have been performed for EVAR follow-up. Patients were considered lost to annual imaging follow-up if they did not undergo any abdominal imaging study within their last two years of follow-up. Multivariable models were constructed to identify independent factors associated with being lost to annual imaging follow-up.

Results

Among 19,962 patients who underwent EVAR, the incidence of loss to annual imaging follow-up at 5 years post EVAR was 50%. Primary factors associated with being lost to annual imaging follow-up were advanced age (age 65-69 years, reference, age 75-79, HR, 1.23 [95% CI, 1.15-1.32]), (age 80-85, HR, 1.45, [95% CI 1.35-1.55]), (age >85, HR 2.03, [95% CI 1.88-2.20]) and presentation with an urgent/emergent intact aneurysm (HR, 1.27 [95% CI, 1.20-1.35]) or ruptured aneurysm (HR, 1.84 [95% CI, 1.63-2.08]). Additional independent factors included several previously diagnosed chronic diseases and South and West regions of the US.

Conclusions

Annual imaging follow-up compliance post EVAR in the United States is significantly below recommended levels. Quality improvement efforts to encourage improved compliance with imaging follow-up, especially in older patients with multiple comorbidities and in those who underwent EVAR urgently or for rupture, are necessary.

Introduction

The management of abdominal aortic aneurysms (AAA) has traditionally depended upon open surgical repair.^{1, 2} Recent advances in catheter-based, endovascular techniques have led to a substantial increase in the proportion of AAAs managed with endovascular aortic aneurysm repair (EVAR). In 2006, 21,725 EVAR procedures were performed in the United States, exceeding for the first time the number of open surgical AAA repairs.³ Currently, greater than 80% of AAA repairs in the United States are performed utilizing EVAR.⁴

Multiple guidelines recommend annual life-long imaging follow-up after EVAR to identify and correct complications such as endoleaks and/or residual aortic sac enlargement and, thereby, prevent death due to aneurysm rupture post EVAR.^{5, 6} Recent evidence from several longitudinal investigations indicates that rates of late aneurysm sac enlargement are unexpectedly high, increasing the risk of death due to rupture post EVAR.^7-9 In a study of 27 patients presenting with AAA rupture post EVAR over the period 2002-2009, approximately three quarters had been lost to follow-up prior to rupture.¹⁰ Accordingly, annual life-long imaging follow-up with either computed tomography (CT), magnetic resonance imaging (MRI), or duplex ultrasound is essential to the safe and appropriate management of patients undergoing EVAR.

Despite the critical importance of annual life-long imaging follow-up after EVAR, little is known about compliance with this requirement.^{11, 12} We hypothesized that among Medicare beneficiaries treated with EVAR, a significant proportion of patients are lost to annual imaging follow-up, and that there are several patient characteristics associated with loss to annual imaging follow-up. To test this hypothesis, we examined Medicare claims data in order to identify a group of patients who underwent EVAR between 2001 and 2008 and estimated the proportion lost to imaging follow-up over an extended duration of follow-up. A secondary objective of this observational study was to examine patient-related factors associated with loss to annual imaging follow-up in whom increased efforts directed at follow-up compliance are warranted.

Methods

Data Sources

We conducted a retrospective cohort study using 2001-2010 enrollment and claims data for a 20% national sample of Medicare beneficiaries selected by the Centers for Medicare and Medicaid Services (CMS) in constructing their standard Part B (physician services) claims files (needed to identify imaging follow-up in our study). Data on EVAR procedures were obtained from the United States Medicare Provider Analysis and Review (MedPAR) files, which contain hospital discharge abstracts for the acute care hospitalizations of all Medicare beneficiaries with Part A coverage. The data include admission and discharge dates, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis and procedure codes and dates, admission type (categorized as elective, urgent, and emergent), and discharge disposition. Imaging data were obtained from Hospital Inpatient, Outpatient, and Carrier files. These files were linked to Medicare Denominator files, which include information on beneficiaries' date of birth, sex, race/ethnicity (categorized as black, white, or other), enrollment status, region of residence (Midwest, Northeast, South, or West), and vital status (including date of death). Follow-up imaging was evaluated for all patients through December 31, 2010.

Cohort Identification

The study cohort was defined by identifying Medicare fee-for-service beneficiaries 65 years or older who underwent an EVAR between January 1, 2001 and December 31, 2008. Eligibility was limited to those continuously enrolled in Medicare Parts A and B (as identified through Denominator files) from the time of their EVAR until the end of the study period or death. Patients were excluded if they resided outside the United States or were enrolled in a Medicare health maintenance organization during the study window.

Patients who underwent EVAR were identified based on the presence of an ICD-9 diagnosis code for AAA with (441.3) or without rupture (441.4) as well as an ICD-9 procedure code for EVAR (39.71). To limit the cohort to patients who underwent EVAR for infrarenal AAA, and to minimize misclassification errors, patients were excluded for diagnosis and procedure codes related to the thoracic aorta or visceral segment aorta (eTable 1, online only).

Identification of Imaging Studies Performed During EVAR Follow-up

To identify any abdominal imaging study that might have been performed for EVAR follow-up, the Inpatient, Outpatient, and Carrier files were queried for billing codes consistent with an abdominal or pelvic CT, MRI, or duplex ultrasound (eTable 2, online only). The billing codes included were intentionally broad to mitigate the risk of missing potentially relevant imaging studies.

Patient Covariates

Patient sociodemographic characteristics examined included age, sex, race, and residential location by United States region. Median household income by beneficiary's zip-code of residence was obtained from United States Census files.¹³ The indication for EVAR was classified as either elective intact aneurysm, urgent or emergent intact aneurysm, or ruptured aneurysm. A partial list of relevant comorbidities used in the Klabunde adaptation of the Charlson comorbidity index were identified using a one-year look-back from the index EVAR admission date.^{14, 15} Comorbidities included ischemic heart disease, congestive heart failure, cerebrovascular disease, chronic obstructive pulmonary disease, chronic renal dysfunction, diabetes, cancer, hypertension, and hyperlipidemia.

Outcome Measures

The primary study endpoint was loss to annual imaging follow-up. Patients were considered lost to annual imaging follow-up in the following two situations:

1. If the patient was alive at the end of the follow up period (December 31, 2010) and did not have any claim submitted for an imaging study (CT, MRI, or duplex ultrasound) within the two years prior to the end of the follow up period.

2. If the patient died during the follow up period and did not have any claim submitted for an imaging study (CT, MRI, or duplex ultrasound) within the two years prior to the date of death.

A two year span was utilized to provide an adequate time window to ensure that loss to annual imaging follow-up was not overestimated.

The secondary endpoint evaluated was the presence of a prolonged imaging gap post EVAR. Patients were considered to have a prolonged imaging gap if they did not undergo at least one imaging study during each two year interval that they were alive following EVAR.

Statistical Methods

Annual incidence of EVAR procedures performed between 2001 and 2008 is reported as procedure rates per 100,000 Medicare beneficiaries. After a patient was determined to be lost to annual imaging follow-up using the methodology presented above, the time-to-event occurrence of loss to annual imaging follow-up was analyzed using the Kaplan-Meier method with the time duration assigned being 2 years from the last obtained imaging study. Between group differences in stratified analyses were compared using the log-rank test. For all survival analyses, observations were censored at the time of death and Kaplan-Meier curves were generated without the use of a smoothing function. To identify independent predictors of loss to annual imaging follow-up, all potential sociodemographic risk factors in Table 1 were introduced into a multivariable Cox proportional hazards model using backwards stepwise selection.

Table 1.

Baseline characteristics of Medicare Beneficiaries who underwent endovascular aneurysm repair between 2001 and 2008 (n=19,962).

Variable	n (%)
Men	16368 (82%)
Age
65-69	3393 (17%)
70-74	4791 (24%)
75-79	5390 (27%)
80-84	4192 (21%)
>85	2196 (11%)
Race
White	18764 (94%)
Black	599 (3%)
Other	599 (3%)
Mean Family Income Quartile
Lowest Quartile (0-25%)	4969 (24.89)
Low Medium Quartile (26-50%)	4988 (24.99)
High Medium Quartile (51-75%)	4954 (24.82)
Highest Quartile (76-100%)	5051 (25.30)
Region of the United States
Northeast	3793 (19%)
Midwest	5390 (27%)
South	8384 (42%)
West	2395 (12%)
Comorbidities
Ischemic Heart Disease	9182 (46%)
Congestive Heart Failure	2595 (13%)
Cerebrovascular Disease	2195 (11%)
Chronic Obstructive Pulmonary Disease	6387 (32%)
Chronic Renal Dysfunction	1596 (8%)
Diabetes	3992 (20%)
Cancer	2994 (15%)
Hypertension	14173 (71%)
Hyperlipidemia	10779 (54%)
Presentation
Elective Intact Aneurysm	16856 (84%)
Urgent or Emergent Intact Aneurysm	2626 (13%)
Ruptured Aneurysm	480 (2%)

The proportion of patients with a prolonged imaging gap for each two year window following EVAR (0-1.9 years, 2-3.9 years, 4-5.9 years, and 6-8 years) was calculated at the patient level using the following equation:

$$100\%-\left(\frac{\#\text{patients}\text{with}\ge 1\text{imaging}\text{study}\text{during}2\text{year}\text{interval}}{\#\text{patients}\text{alive}\text{during}2\text{year}\text{interval}}\ast 100\right)$$

All analyses were performed using SAS 9.2 (Cary, NC); two-tailed p-values <0.05 were considered statistically significant. This study was approved by the Institutional Review Board at the University of Massachusetts Medical School.

Results

The study population consisted of 19,962 Medicare beneficiaries aged 65 years and older who underwent EVAR between 2001 and 2008. The patients were primarily men (82%), white (94%), with an average age of 76.3 years, a median household income of $49,746, and represented all United States regions (Table 1). Patients who underwent EVAR had a high burden of cardiovascular risk factors including ischemic heart disease (46%), diabetes (20%), hypertension (71%), and hyperlipidemia (54%). Eighty-four percent of EVARs were performed electively, 13% urgently or emergently for intact aneurysm, and 2% emergently for ruptured aneurysm.

Over the study period, the absolute number of hospitalizations for EVAR increased markedly (from 1,400 in 2001 to 3,529 in 2008), representing a more than 2.5 fold increase in the annual incidence of EVAR procedures (from 25 EVARs per 100,000 Medicare beneficiaries in 2001 to 65 EVARs per 100,000 Medicare beneficiaries in 2008, p-value for trend <0.001) (Figure 1).

Figure 1.

The number of hospitalizations for EVAR procedures per 100,000 Medicare beneficiaries, between 2001 and 2008.

Loss to Annual Imaging Follow-up

The mean duration of follow-up was 5 years ± 2.6 years. The proportion of patients lost to annual imaging follow-up on Kaplan-Meier analysis at 1, 3, and 5 years after EVAR was 22% (SE .0003), 38% (SE .004), and 50% (SE .004), respectively. Loss to annual imaging follow-up appeared to increase steadily throughout the study period with no evidence of a plateau or slowdown in the slope of the Kaplan-Meier curve (Figure 2).

Figure 2.

Kaplan-Meier analysis of all patients who underwent EVAR between 2001 and 2008 demonstrating the proportion of patients not lost to imaging follow up.

On univariable analysis, the proportion of patients lost to annual imaging follow-up was significantly associated with the relative urgency of the EVAR procedure (Figure 3). Patients treated with EVAR for ruptured AAA were significantly more likely to be lost to annual imaging follow-up than were patients treated with EVAR for intact AAA (p<0.001). This effect was also seen when patients treated for intact aneurysms were further stratified based on urgent/emergent status versus elective status. Patients undergoing intact AAA repair urgently or emergently had higher rates of being lost to annual imaging follow-up than did patients undergoing elective AAA repair (p<0.001). This observed increase in being lost to annual imaging follow-up occurred predominantly during the first 2 years of follow-up. The subsequent rate of being lost to imaging follow-up during years 3, 4, and 5 occurred at a similar rate across all groups (Figure 3).

Figure 3.

Kaplan-Meier analysis of all patients who underwent EVAR between 2001 and 2008 demonstrating the proportion of patients not lost to imaging follow up, stratified by presentation at the time of repair—elective, urgent/emergent, ruptured.

On multivariable analysis, the factors associated with loss to annual imaging follow-up that had the largest hazards ratios were advanced age and aneurysm rupture on presentation (Table 2). Additional independent predictors of loss to annual imaging follow-up included previous histories of congestive heart failure, chronic renal insufficiency, chronic obstructive pulmonary disease, cancer, cerebrovascular disease, diabetes, and South and West regions of the United States.

Table 2.

Independent factors associated with being lost to imaging follow-up following EVAR.

Variable	Hazards Ratio	95% Confidence Interval
Age
65-69	ref	ref
70-74	1.04	0.97-1.12
75-79	1.23	1.15-1.32
80-85	1.45	1.35-1.55
>85	2.03	1.88-2.20
US Region
East	ref	ref
Midwest	1.05	0.98-1.11
South	1.1	1.03-1.17
West	1.16	1.07-1.25
Comorbidities
Congestive Heart Failure	1.52	1.43-1.61
Cerebrovascular Disease	1.07	1.01-1.15
COPD	1.29	1.23-1.35
Chronic Renal Insufficiency	1.37	1.27-1.49
Diabetes	1.07	1.02-1.13
Cancer	1.12	1.06-1.19
Presentation
Elective Non-ruptured	ref	ref
Urgent/Emergent Non-Ruptured	1.27	1.20-1.35
Ruptured	1.84	1.63-2.08

Prolonged Imaging Gap

Figure 4 demonstrates the proportion of patients who received ≥1 imaging study (CT, MRI, or duplex ultrasound) during each 2 year interval following EVAR. Of the 18,914 patients alive 2 years after EVAR, 94.8% received ≥1 imaging study post EVAR. For the 13,307 patients alive 4 years after EVAR, 69.6% received ≥1 imaging study between years 2 and 4 post EVAR. For the 7,666 patients alive 6 years after EVAR, 49.4% received ≥1 imaging study between years 4 and 6 post EVAR. For the 3,239 patients alive 8 years after EVAR, 36.6% received ≥1 imaging study between years 6 and 8 post EVAR.

Figure 4.

Prolonged imaging gap. Proportion of patients who received at least one imaging study (CT, MRI, or duplex) during each two year interval that they were alive following EVAR.

Discussion

This study demonstrates that, in a large population of Medicare beneficiaries who underwent EVAR between 2001 and 2008, 50% of patients were lost to annual imaging follow-up by 5 years post EVAR. For the subset of patients with 8 years of follow-up data post EVAR, substantive declines in imaging follow-up continued with only 37% undergoing an imaging study between years 6 and 8. These rates are surprisingly low given the clear evidence supporting the importance of annual imaging follow-up for identifying correctable problems after EVAR that can lead to aneurysm sac enlargement, rupture, and death.^5-10

It is possible that a proportion of the patients categorized as, “lost to annual imaging follow-up” were not truly “lost” but may have stopped receiving follow-up either by choice or they may have been told by their physician that they no longer needed further follow-up. Either way, compliance with recommended lifelong imaging follow-up after EVAR was poor. The exact motivation leading to this poor compliance is an interesting and important question that, unfortunately, is beyond the scope of this study and cannot be addressed with this dataset.

Despite the critical importance of annual imaging follow-up post EVAR, few studies have examined compliance with this guideline-recommended management and United States data are limited to single-institution studies.^{5, 6} Jones and colleagues analyzed follow-up compliance in 302 patients who underwent EVAR at a single institution (1999-2005).¹¹ Over an average follow-up of 30 months, 33% of patients had incomplete follow-up (defined as missing ≥2 consecutive imaging studies). In another single center review with a more heterogeneous population (patients underwent EVAR, endovascular thoracic aneurysm repair, or medical management for Type B aortic dissection), imaging follow-up compliance was assessed in 204 patients.¹⁶ At an average follow-up of 28 months, 56% of patients were lost to imaging follow-up (defined as >1 year since last imaging).

To our knowledge, our study provides the first population based estimates of post EVAR annual imaging follow-up rates for the United States. In Europe, an attempt has been made to better understand factors associated with loss to imaging follow-up.¹⁷ The European Collaborating Group on Stent-Graft Techniques for Abdominal Aortic Aneurysm Repair recommends lifelong annual imaging follow-up post EVAR. Using data from 4,433 patients enrolled in this registry over the period 1996-2004, 65% failed to comply with the recommended follow-up regimen. On multivariable analysis, compliance was associated with the number of comorbidities and cardiovascular risk factors present. Patients who were hypertensive, actively smoked, and deemed medically too high risk for open repair, were more likely to comply with imaging follow-up.

In contrast to the European findings, it appears that Medicare beneficiaries are more likely to be lost to annual imaging follow-up if they have multiple comorbidities previously diagnosed. Presence of congestive heart failure, cerebrovascular disease, chronic obstructive pulmonary disease, chronic renal insufficiency, diabetes, and cancer were associated with a significantly increased hazard of loss to annual imaging follow-up. One possible explanation for this finding may be that in patients with a number of active long-term medical issues, the patient and treating physicians may lose focus on surveillance of an AAA post EVAR due to the demands of caring for their high burden of competing medical risks.

Older individuals were also more likely to be lost to annual imaging follow-up. Not surprisingly, each 5 year increment of advanced age was associated with a significant increase in being lost to imaging follow-up. Since the duration of remaining life expectancy decreases with advancing age, the perceived benefit of regular surveillance post EVAR may decrease as well. In a patient older than 85 years (hazards ratio for loss to imaging follow-up 2.03), the value of performing a reintervention, even if a problem is identified post EVAR, becomes less clear.

Surprisingly, patients who underwent EVAR in the urgent/emergent setting and those who presented with a ruptured AAA were significantly more likely to be lost to annual imaging follow-up. The reasons for this are likely multifactorial. First, patients undergoing elective EVAR have been counseled by the treating physician, often on multiple visits, and may have a better underlying understanding of their disease and the critical importance of life-long imaging follow-up. Second, patients being treated in the urgent/emergent setting are often treated outside of their established health care network and are more likely to be lost to follow-up. These findings point to an important opportunity for national quality improvement efforts to enhance care coordination across providers of care and to reward appropriate imaging follow-up among patients at high risk for loss to follow-up. Novel ideas to improve imaging follow-up compliance, such as physician-patient contracts, improved preoperative education materials, and financial incentives directed at both patients and physicians need to be developed and tested.

There are several strengths and limitations inherent to this study design. First, our cohort was limited to elderly Medicare beneficiaries and we cannot comment on post-EVAR follow-up imaging rates among individuals younger than 65 years. Second, we restricted analyses to Medicare fee-for-service enrollees, as Medicare managed care enrollees frequently have incomplete claims and, therefore, incomplete capture of follow-up imaging data. If follow-up imaging compliance is higher among Medicare Advantage enrollees, our imaging compliance estimates may be low. Offsetting this potential bias, we defined loss to annual imaging follow-up quite liberally (2 years without an imaging study) instead of using a more stringent definition (1 year without an imaging study). This was deliberate in order to avoid overestimating the rate of loss to annual imaging follow-up. As a result, the true rate of loss to annual imaging follow-up is likely to be significantly higher than reported here. Finally, it is possible that a proportion of the imaging studies identified during the follow-up period were indicated for reasons other than post EVAR surveillance. While the AAA may have been imaged, we were unable to know if a physician with expertise in evaluating an EVAR repair reviewed the imaging study.

An important question that arises based on these data, is how many patients who were lost to imaging follow-up progressed to AAA rupture post EVAR compared to those who were not lost to imaging follow-up. In many ways, this is the key clinical question that begins to quantify the importance of annual life-long imaging follow-up after EVAR. Unfortunately, this is a difficult question to answer with administrative claims data as the majority of patients who experience AAA rupture do not make it to the hospital. As a result, these events are not captured in administrative data such as the Medicare files used in the present study. Furthermore, we are unable to comment on the proportion of follow-up imaging studies that demonstrated abnormalities related to the EVAR procedure as the dataset contains no results from the imaging study. We are limited to only being able to report whether or not they had an imaging study based on the claims submitted.

In conclusion, these findings suggest that compliance with imaging follow-up recommendations after EVAR in the United States is well below the recommended rate. All patients, and especially those of advanced age and those with multiple comorbidities, should be counseled that their AAA, despite having undergone endovascular treatment, still carries a small but lifelong risk of rupture and, therefore, requires lifelong annual imaging surveillance. Also, patients undergoing EVAR urgently/emergently or for a ruptured AAA appear to be at especially high risk for loss to annual imaging follow-up and warrant increased attention to improve long-term imaging follow-up rates. These findings point to an important opportunity for quality improvement, particularly in an era in which Medicare seeks to expand its armament of quality improvement measures.

Supplementary Material

Online Appendix

eTable 1: ICD 9 diagnosis codes and ICD 9 procedure codes utilized to create the study cohort.

eTable 2: Billing codes utilized to identify claims for an abdominal or pelvic imaging study.