Functional Status Predicts Major Complications and Death After Endovascular Repair of Abdominal Aortic Aneurysms

Abstract

Objective

Endovascular aortic repair (EVAR) is considered a lower risk option for treating abdominal aortic aneurysms (AAA), and is of particular utility in patients with poor functional status who may be poor candidates for open repair. However, the specific contribution of preoperative functional status EVAR outcomes remains poorly defined. We hypothesized that impaired functional status, based simply on the ability of patients to perform activities of daily living, is associated with worse outcomes after EVAR.

Methods

Patients undergoing non-emergent EVAR for AAA between 2010 and 2014 were identified in the National Surgical Quality Improvement Program (NSQIP) database. The primary outcomes were 30-day mortality and major operative and systemic complications. Secondary outcomes were inpatient length of stay, need for reoperation, and discharge disposition. Using NSQIP defined preoperative functional status, patients were stratified as Independent or Dependent (either partial or totally dependent), and compared by univariate and multivariable analyses.

Results

Of 13,432 patients undergoing EVAR between 2010 – 2014, 13,043 were independent (97%) and 389 were dependent (3%) prior to surgery. Dependent patients were older and more frequently minorities; had higher rates of chronic pulmonary, heart, and kidney disease; and were more likely to have an American Society of Anesthesiologists score of 4 or 5. On multivariable analysis, preoperative dependent status was an independently risk factor for operative complications (OR 3.1, 95% CI 2.5 – 3.9), systemic complications (2.8, 2.0 – 3.9), and 30-day mortality (3.4, 2.1 – 5.6). Secondary outcomes were worse among dependent patients.

Conclusions

Although EVAR is a minimally invasive procedure with substantially less physiologic stress than open aortic repair, preoperative functional status is a critical determinant of adverse outcomes after EVAR, in spite of the minimally invasive nature of the procedure. Functional status, as measured by performance of activities of daily living, can be used as a valuable marker of increased perioperative risk, and may identify patients who may benefit from preoperative conditioning and specialized perioperative care.

Introduction

An estimated 40,000 intact abdominal aortic aneurysm (AAA) repairs are performed annually in the United States. ¹ Endovascular aortic repair (EVAR) has become the primary method of AAA repair, with approximately 80% of patients in major registries undergoing endovascular repair. ^1–3 Compared to traditional open AAA repair, EVAR is less invasive and offers several advantages, including fewer perioperative complications and substantially lower short- and medium-term mortality rates. ^3–6 Indeed, advances in arterial access and closure enable EVAR to be safely performed as an entirely percutaneous procedure, often on an outpatient basis. ^7–9 As such, EVAR has particular utility in patients with suitable anatomy who are at higher risk for adverse perioperative outcomes, such as the elderly, those with multiple comorbidities, or patients with impaired functional status. ^5,10

Impaired functional status, manifest by dependence on others for activities of daily living, is a defining component of frailty, which in turn is associated with worse postoperative outcomes in multiple previous studies. ^11–14 Dependent functional status, alone, is independently associated with major complications and mortality after major general and peripheral vascular surgery, and frailty is associated with substantially increased morbidity and mortality after EVAR and open AAA repair. ^15–19 While EVAR may be preferable to open surgery in patients with marginal functional status requiring AAA repair, the actual effect of functional status on outcomes after EVAR remains poorly quantified and this has been identified as a research priority. ²⁰ We hypothesized that impaired functional status is an independent risk factor for adverse outcomes after EVAR, despite it being a minimally invasive procedure.

Methods

This was a retrospective analysis of patients undergoing non-emergent EVAR for infra-renal AAA in the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) database. The NSQIP database is a prospective sample of surgeries from participating hospitals that involves standardized data abstraction by trained reviewers, with internal mechanisms to optimize accuracy and 30-day follow-up. ²¹ Collected perioperative variables include detailed demographic, clinical, procedural, and outcomes through 30 days after the principal procedure. ²²

In particular, since 2010, preoperative functional status has been categorized as independent, or partially or totally dependent based on requirements for assistance with none, some, or all of an individual’s activities of daily living, respectively (Table I). ²³ Functional status, assessed in similar fashion, has been externally and prospectively validated as an independent predictor of surgical outcomes. ²⁴ The reported inter-rater agreement for assessment of preoperative functional status prior to the study period was 96.6%. ²¹ NSQIP participant use data files were obtained from the ACS. Because the database is de-identified and contains no identifiable individual data, neither Institutional Review Board approval nor patient consent was required. The data files were used under the terms of the NSQIP data use agreement. ²⁵

Table I.

Functional Statusa	Definition
Independent	No assistance from another person required for any ADLs, including patients who function independently with prosthetics, equipment, or other supportive devices.
Partially Dependent	Some assistance from another person required for ADLs.
Totally Dependent	Total assistance required for all ADLs.

Patients undergoing EVAR for infra-renal AAA during 2010 – 2014 were identified by the principal operative procedure CPT codes 34800, 34802, 34803, 34804, 34805, 34825, or 34826; the latter two codes, which refer to placement of aortic extension devices, were included after exploratory analysis of the detailed NSQIP AAA specific dataset indicated these were frequently being used as primary EVAR codes. Those patients with unspecified preoperative functional status or who underwent emergency EVAR were excluded. Patients were categorized according to preoperative functional status as Independent or Dependent (either partially or totally dependent; Table I). The primary outcomes were composite major operative complications, composite major systemic complications, and 30-day mortality (Table II). Secondary outcomes were hospital length of stay, any need for any major operation within 30 days of the index EVAR, and discharge disposition.

Table II.

Outcome	NSQIP Field
Operative Complications
Deep SSI	WNDINFD – occurrences of deep incisional SSI
Organ space SSI	ORGSPCSSI – occurrences organ space SSI
Wound dehiscence	DEHIS – occurrences wound disruption
Need for transfusion	OTHBLEED – occurrences bleeding transfusions
Device complication	OTHGRAFL – occurrences graft/prosthesis
Systemic Complications
Pneumonia	OUPNUEMO - pneumonia
Reintubation	REINTUB – occurrences unplanned reintubation
Pulmonary embolism	PULEMBOL – occurrences pulmonary embolism
Respiratory failure	FAILWEAN – occurrences ventilator > 48 hours
Renal failure	OPRENAFL – occurrences acute renal failure
Stroke	CNSCVA – CVA/stroke with neurologic deficit
Myocardial infarction	CDARREST – occurrences cardiac arrest
Cardiac arrest	CDMI – occurrences myocardial infarction
30-day Mortality	DOpertoD – days from operation to death

The Independent and Dependent groups were compared by 2-sided Pearson χ² or Fisher’s exact tests for categorical variables. Normally distributed continuous data was analyzed using independent Student’s t test, while non-normal data was compared using the Mann-Whitney U test. Demographic and preoperative clinical variables with a P value < .20 on univariate analysis were included in step-wise multivariable logistic regression, with dichotomization of age as < or ≥ 80 years and creatinine as ≤ or > 1.8 mg/dL based on prior studies and exploratory analyses. Variables were retained in the initial multivariable model for at a P < .05. Possible confounding was explored by adding variables into the model, with those resulting in changes in the coefficients of >20% considered as confounders. A final multivariable model was then run, controlling for baseline demographic characteristics. Variables that were independently associated with mortality, or operative or systemic complications with a P < .0001 were included in clusters to evaluate the effect of combined dependency and comorbidities on subsequent outcomes.

P < .05 was considered statistically significant, with Bonferroni correction of multiple comparisons of the primary (P < .017) and secondary outcomes (P < .0125). All eligible patients during the study period were included to constitute the final sample of 13,432 patients (13,043 independent and 389 dependent). Based on major contemporary AAA studies, a 1% mortality rate was expected for intact EVAR. ^1,2,5 An absolute increase in mortality of 4% among dependent patients (yielding 5% mortality, equivalent to mortality seen for standard open AAA repair), ^1,2,5 was determined a priori to be a clinically significant finding, for which the given samples had 99.6% power to detect with an alpha of .05. At an alpha of .05, the samples had 90% power to detect an absolute increase in mortality of 2.2%.

Results

Between 2010 and 2014, 13,485 patients underwent non-emergent EVAR for infra-renal AAA, of whom 13,432 (99.6%) had a specified preoperative functional status. Prior to surgery, 13,043 (97%) were reported as functionally independent, while 389 (3%) were dependent (340 partially and 49 totally dependent). Dependent patients were older, more often female, more frequently non-white, and were more likely to be transferred from another acute care facility or nursing home preoperatively (Table III). Rates of chronic cardiac, pulmonary, and renal disease were higher among the Dependent group. Last, operating times were moderately longer for dependent patients (median 147 [IQR 106 – 216] vs. 130 [99 – 173] minutes, P < .0001).

Table III.

Characteristic	Independent	Dependent	P
Number	13,043	389
Demographics
Age, years	73 ± 8	77 ± 8	< .0001
Age > 80 years, %	3,212 (25%)	170 (44%)	< .0001
Female, n (%)	2,466 (19%)	127 (33%)	< .0001
BMI, kg/m2	28 ± 6	27 ± 8	< .0001
Non-white race, %	1,913 (15%)	89 (23%)	< .0001
Admission source, n (%)
Home	12,536 (96%)	283 (73%)	< .0001
Another hospital or ED	454 (4%)	45 (12%)	< .0001
Nursing home	43 (0%)	61 (16%)	< .0001
Comorbidities, n (%)
Hypertension	81%	85%	.03
Diabetes mellitus	2,115 (16%)	77 (20%)	.06
Smoker	4,051 (31%)	109 (28%)	.20
Dyspnea at rest	164 (1%)	36 (9%)	< .0001
COPD	2,409 (19%)	127 (33%)	< .0001
CHF	185 (1%)	24 (6%)	< .0001
Prior myocardial infarction	20 (0.2%)	1 (0.3%)	.46
Prior coronary artery stent	605 (5%)	14 (4%)	.49
ESRD	156 (1%)	14 (4%)	< .0001
ASA 4 or 5	2,907 (22%)	177 (46%)	< .0001
Creatinine, mg/dL ± SD	1.25 ± 0.95	1.14 ± 0.58	.98

Impaired preoperative functional status was associated with stepwise increases in major complications and death (Figure I). The dependent patient group had a higher rate of operative complications (34% vs. 11%, P < .0001; Table IV), primarily due to higher transfusion rates. Dependent patients had higher rates of pulmonary complications, cardiac events, stroke, and renal failure, resulting in a significantly higher rate of composite systemic complications (13% vs. 4%, P < .0001). Overall mortality was 1.1%, but mortality was substantially higher in the dependent group (6% vs. 1%, P < .0001).

Figure 1.

Table IV.

Outcome	Independent	Dependent	P
Number	13,043	389
Primary, n (%)a
Operative complications	1,411 (11%)	134 (34%)	< .0001
Deep SSI	51 (0.4%)	4 (1%)
Organ space SSI	20 (0.2%)	1 (0.3%)
Wound dehiscence	26 (0.2%)	1 (0.3%)
Need for transfusion	1,340 (10%)	130 (33%)
Device complication	0	0
Systemic complications	452 (4%)	51 (13%)	< .0001
Pneumonia	117 (0.9%)	19 (5%)
Reintubation	157 (1%)	15 (4%)
Pulmonary embolism	20 (0.2%)	1 (0.3%)
Respiratory failure	91 (0.7%)	12 (3%)
Renal failure	85 (0.7%)	7 (2%)
Stroke	41 (0.3%)	4 (1%)
Myocardial infarction	117 (0.9%)	6 (2%)
Cardiac arrest	47 (0.4%)	9 (2%)
30-day mortality	128 (1%)	22 (6%)	< .0001
Secondaryb
Length of stay, median days (IQR)	2 (1 – 3)	4 (2 – 9)	< .0001
Return to OR, n (%)	515 (4%)	29 (7%)	.001
Discharge to home, n (%)	11,930 (92%)	199 (51%)	< .0001
Discharge to facility, n (%)	736 (6%)	134 (34%)	< .0001
At home preoperatively, n (%)	629 (85%)	82 (61%)

After multivariable analysis adjusting for demographics, comorbidities, and operative risk, dependent functional status was independently associated with operative and systemic complications, as well as 30-day mortality. Dependent functional status was the leading risk factor for both operative (OR 3.1, 95% CI 2.5 – 3.9) and systemic complications (OR 2.8, 95% CI 2.0 – 3.9; Table V). Similarly, following comorbid end-stage renal disease (ESRD; OR 4.6, 95% CI 2.5 – 9.6), dependent functional status was the strongest independent risk factor for 30-day mortality (OR 3.4, 95% CI 2.0 – 5.3).

Table V.

Outcome	Variable	OR (95% CI)	Pa
Operative complications	Dependent	3.1 (2.5 – 3.9)	< .0001
	Age ≥ 80	1.8 (1.6 – 2.0)	< .0001
	Female	1.9 (1.7 – 2.2)	< .0001
	Creatinine > 1.8 mg/dL	2.2 (1.9 – 2.6)	< .0001
	ASA 4 or 5	1.9 (1.7 – 2.2)	< .0001
Systemic complications	Dependent	2.8 (2.0 – 3.9)	< .0001
	Age ≥ 80	1.6 (1.3 – 2.0)	< .0001
	Female	1.4 (1.1 – 1.7)	.002
	COPD	2.0 (1.6 – 2.4)	< .0001
	Hypertension	1.5 (1.1 – 1.9)	.004
	Creatinine > 1.8 mg/dL	2.1 (1.7 – 2.7)	< .0001
	ASA 4 or 5	1.5 (1.2 – 1.8)	.0002
30-day mortality	Dependent	3.4 (2.0 – 5.3)	< .0001
	Age ≥ 80	2.4 (1.7 – 3.5)	< .0001
	COPD	2.3 (1.5 – 3.0)	< .0001
	ESRD	4.6 (2.5 – 9.6)	< .0001
	ASA 4 or 5	2.2 (1.5 – 3.1)	< .0001
	Female	1.7 (1.2 – 2.4)	< .0001

Multivariable analysis of risk factor combinations demonstrated increasing cumulative odds ratios for major complications and mortality in the setting of dependent functional status and other common comorbidities (Table VI). In particular, dependent functional status plus age ≥ 80 years and/or female sex were associated with significantly increased risk for operative complications, with chronic obstructive pulmonary disease (COPD) entailing further risk for systemic complications. Compared to the risk associated with dependent functional status alone, the odds for mortality more than doubled in the setting of COPD or age ≥ 80 years (OR 7.7 and 8.2, respectively), and was more than four times higher for dependent patients with ESRD (OR 15.6). Progressively higher risk was seen in the setting of combined dependent status, advanced age, and comorbidities.

Table VI.

Risk Factor Combination	OR (95% CI)	Pa
Operative complications
Dependent	3.1 (2.5 – 3.9)	< .0001
Dependent + Age ≥ 80	5.4 (4.2 – 7.0)	< .0001
Dependent + Female	5.8 (4.6 – 7.4)	< .0001
Dependent + Age ≥ 80 + Female	10.2 (7.6 – 13.3)	< .0001
Systemic complications
Dependent	2.8 (2.0 – 3.9)	< .0001
Dependent + Female	3.9 (2.7 – 5.7)	< .0001
Dependent + Age ≥ 80	5.0 (3.1 – 6.5)	< .0001
Dependent + COPD	5.5 (3.8 – 8.0)	< .0001
Dependent + Age ≥ 80 + Female	6.3 (4.2 – 9.4)	< .0001
Dependent + Age ≥ 80 + COPD	8.9 (5.9 – 13.5)	< .0001
30-day mortality
Dependent	3.2 (2.0 – 5.3)	< .0001
Dependent + COPD	6.9 (3.8 – 12.5)	< .0001
Dependent + Age ≥ 80	7.8 (4.4 – 14.0)	< .0001
Dependent + ESRD	15.8 (6.9 – 36.0)	< .0001
Dependent + Age ≥ 80 + COPD	16.7 (8.6 – 32.6)	< .0001
Dependent + ESRD + COPD	33.8 (14.0 – 81.8)	< .0001
Dependent + Age ≥ 80 + ESRD	38.2 (15.8 – 92.6)	< .0001
Dependent + Age ≥ 80 + COPD + ESRD	81.9 (31.9 – 210.0)	< .0001

Dependent patients had significantly worse secondary outcomes after EVAR (Table IV). The median hospital length of stay of Dependent patients was twice as long as for independent patients (4 vs. 2 days, P < .0001). Dependent patients required more major operations within 30 days of their index EVAR (7% vs. 4%, P = .001), although data on reoperations was not available for the entire study period. While nearly all Independent patients were discharged to home, one-third of Dependent patients required discharge to other healthcare facilities (34% vs. 6%, P < .0001), including many who were at home preoperatively (21% vs. 5%, P < .0001).

Discussion

Advances in technique and technology have enabled EVAR to be performed with such ease that it can potentially be performed on an outpatient basis without general anesthesia or a skin incision. ^7–9 However, despite being minimally invasive and associated with lower morbidity and mortality than open AAA repair, ^3,5,6 this study demonstrates that EVAR remains a high-risk procedure for patients with impaired functional status. Preoperative functional status is a critical predictor of adverse outcomes after EVAR, and identifies patients who may benefit from pre- and perioperative interventions targeted to improve outcomes among functionally compromised patients.

The overall 30-day mortality rate of 1.1% for non-emergent EVAR in this study was similar to outcomes from the National Inpatient Sample, Medicare beneficiaries, and the preceding five years in NSQIP. ^1,5,26 However, mortality among Dependent patients was 6%, comparable to the typical mortality associated with open AAA repair. ^1,5 While these patients may have faced even higher mortality had they undergone open repair, this study indicates that EVAR remains a high-risk procedure in this population. Further, this study clearly demonstrates that impaired functional status in the setting of advanced age and/or specific common comorbidities is associated with progressively increased risk for mortality. In particular, EVAR in select subgroups, particularly dependent patients who are 80 years or older with chronic renal and/or pulmonary disease, entails substantially increased risk that should be considered during preoperative evaluation, patient selection and counseling, and perioperative care.

The higher rate of composite operative complications among dependent patients was primarily due to more frequent blood transfusion in this group. It is unclear if this was due to greater intraoperative blood loss, a higher prevalence of chronic anemia, or a lower transfusion threshold in this group. Any scenario is plausible, as dependent patients had both longer operating times and a greater burden of comorbidities, but this distinction is not clear from this dataset. In contrast, except for pulmonary embolism, systemic complications were consistently more common among dependent patients. While the relationship between EVAR and subsequent systemic complications is unclear in this dataset, apart from the potential for contrast induced nephropathy, EVAR would not be expected to directly cause or account for these events. Instead, and consistent with other literature on functional status and frailty, the higher rates of systemic complications likely reflect greater general susceptibility to perioperative complications that is independent of the specific procedure, and is ultimately manifest in higher mortality. Indeed, the 1% and 6% mortality rates for independent and dependent patients in this study were nearly identical to the 1.5% and 6.1% mortality rates for similarly classified patients undergoing lower extremity bypass in NSQIP. ¹⁶

These findings are consistent with recent analyses of physical fitness and frailty among patients undergoing aortic repair. Boult et al reported that poor self-reported preoperative fitness predicted mortality among moderate risk patients undergoing EVAR. ²⁷ In a recent analysis by Arya et al that applied a modified frailty index to patients undergoing elective AAA repair in the NSQIP database, frailty was independently associated with increased mortality and complication rates, with similarly increased risk for both EVAR and open AAA repair. ¹⁹ Indeed, functional status may be more critical in determining postoperative outcomes than the type of aortic repair. In a study of nursing home patients undergoing aortic repair, Beffa et al found that preoperative functional scores were strongly associated with subsequent recovery, whereas repair modality was not. They concluded that EVAR “may be more physically demanding on patients than previously believed…the less invasive endovascular approach was not associated with improved functional preservation compared with open aortic repair.” ²⁸

This study adds to a growing body of literature demonstrating dependent functional status is a leading risk factor for adverse outcomes after major surgery. Analyses of institutional and NSQIP data indicate functional impairment or frailty is an independent risk factor for complications and death after a range surgical procedures. ^{11,12,15,17,18,24,29} A study by Hua et al of open and endovascular AAA repair using 2000 – 2003 NSQIP data previously reported dependent patients have higher morbidity and mortality after EVAR. Our study confirms and expands upon that finding, demonstrating that impaired functional status remains an important risk factor despite advances that have further reduced the invasiveness and physiologic stress of EVAR. Taken together with data that patients undergoing ambulatory surgery have significant rates of post-operative functional impairment, ³⁰ these findings confirm that even less invasive endovascular procedures can induce significant physiologic stress, in particularly the elderly and those with major comorbidities.

Functional status is a particularly useful marker of perioperative risk because it is easily assessed and is potentially modifiable and manageable. Interventions that improve functional status prior to surgery or maintain it during the perioperative period may provide an opportunity to improve outcomes for these high risk patients. For example, structured exercise improves functional status and physiologic reserve, ^31–33 which is associated with better postoperative outcomes. ³³ During hospitalization, care in a unit designed to promote independence reduces functional deterioration and need for discharge to a long-term care facility. ³⁴ As such, preoperative functional status screening could be used to improve risk assessment and identify patients who may benefit from preoperative conditioning and dedicated perioperative care. Further study is required to determine if such interventions are beneficial among patients undergoing EVAR.

Conversely, functional independence is strongly associated with complication-free survival after EVAR. Particularly for those without major comorbidities, independent patients can undergo EVAR with minimal risk for complications or death. As such, functional independence is a useful criterion for identifying patients who may be appropriate for outpatient EVAR or other enhanced recovery pathways. ⁹

Several mechanisms could account for the higher morbidity and mortality among dependent patients. Most directly, impaired functional status could lead to complications via impaired pulmonary hygiene, immobility, and poor nutrition. Major comorbidities, which are more prevalent among dependent patients, could increase the risk for events such as wound complications, respiratory or renal failure, or potentially fatal cardiopulmonary episodes. While our data indicates functional status and select comorbidities are independently associated with major adverse outcomes, our composite analyses and published frailty models suggest an additive effects from the accumulation of functional and medical deficits. ^12,35 Although not assessed in this study, additional components of frailty, such as impaired nutritional status or cognitive function, could potentially mediate complications among dependent patients.

This study has several important limitations. While functional status is accurately abstracted in NSQIP, its use in the database remains a subjectively assessed data element. Prospective assessment of functional status or objective measures, such as radiographically assessed core muscle size, may more accurately reflect functional condition and predict outcomes. ³⁶ Further, other studies have examined the relationship between frailty and surgical outcomes. While functional status is an important component of frailty, the latter is a more expansive concept, ¹⁴ and in practice can be more difficult and time-consuming to accurately assess. ^11,13,31 However, the potential ease of assessing functional status, and its consistent relationship with outcomes in other studies and datasets, suggests it remains a valuable measure of patient condition and operative risk.

The NSQIP database only includes perioperative details for patients undergoing intervention, and as such does not account for potential bias in selecting dependent patients for EVAR. While the results are consistent with other studies of functional status and surgical outcomes, it is unclear if these findings can be extrapolated to all dependent patients with intact infra-renal AAA who may be considered for EVAR. Using this dataset, we were unable to account for potential differences in hospital volume; it is possible that experienced centers with greater EVAR volumes in higher risk patients may have better outcomes than reported here. The standard NSQIP dataset used for this study does not report aneurysm size, which limited assessment of the potential benefit of EVAR for AAA repair in this study. However, it is likely that the operative risk for EVAR in certain dependent patients with moderate sized AAA exceeds their expected risk for rupture. In the setting of advanced age and major comorbidities, our data suggests select dependent patients may not benefit from prophylactic repair for AAA < 6.0 cm. ³⁷ Further, as recently reported by Gonçalves et al, after AAA repair patients have substantial non-aortic related late mortality from other cardiovascular disease and malignancy that limits the long-term benefit of AAA repair. This is most applicable to elderly patients, who are both more likely to have decreased functional capacity and be at risk for late mortality. ³⁸ As such, the utility of AAA repair in poorly functional patients and consideration of a higher threshold for intervention warrants further investigation. However, given the importance of functional status demonstrated in this analysis, it should be considered a critical component of caring for patients with AAA and should be assessed on an individual basis when assessing the patients for EVAR.

Conclusion

Endovascular aortic repair is a high-risk procedure in patients with impaired functional status. Because dependent functional status is easily assessed and independently identifies patients who are at increased risk for major complications and mortality, its assessment should be a standard part of the preoperative evaluation for patients undergoing EVAR and it should be considered in the risk/benefit assessment when selecting individual patients for intervention. Certain dependent subgroups, such as those with advanced age and/or major comorbidities, face substantially increased risk that may be prohibitive under certain circumstances, which may warrant a higher threshold for intervention.