Association of Frailty With Treatment Selection and Long-Term Outcomes Among Patients With Chronic Limb-Threatening Ischemia.

Background The optimal treatment strategy for patients with chronic limb-threatening ischemia (CLTI) is often unclear. Frailty has emerged as an important factor that can identify patients at greater risk of poor outcomes and guide treatment selection, but few studies have explored its utility among the CLTI population. We examine the association of a health record-based frailty measure with treatment choice and long-term outcomes among patients hospitalized with CLTI. Methods and Results We included patients aged >65 years hospitalized with CLTI in the Medicare Provider Analysis and Review data set between October 1, 2009 and September 30, 2015. The primary exposure was frailty, defined by the Claims-based Frailty Indicator. Baseline frailty status and revascularization choice were examined using logistic regression. Cox proportional hazards regression was used to determine the association between frailty and death or amputation, stratifying by treatment strategy. Of 85 060 patients, 35 484 (42%) were classified as frail. Frail patients had lower likelihood of revascularization (adjusted odds ratio [OR], 0.78; 95% CI, 0.75‒0.82). Among those revascularized, frailty was associated with lower likelihood of surgical versus endovascular treatment (adjusted OR, 0.76; CI, 0.72‒0.81). Frail patients experienced increased risk of amputation or death, regardless of revascularization status (revascularized: adjusted hazard ratio [HR], 1.34; CI, 1.30‒1.38; non-revascularized: adjusted HR, 1.22; CI, 1.17‒1.27). Among those revascularized, frailty was independently associated with amputation or death irrespective of revascularization strategy (surgical: adjusted HR, 1.36; CI, 1.31‒1.42; endovascular: aHR, 1.29; CI, 1.243‒1.35). Conclusions Among patients hospitalized with CLTI, frailty is an important independent predictor of revascularization strategy and longitudinal adverse outcomes.

amputation‐free survival

Claims‐based Frailty Indicator

chronic limb‐threatening ischemia

Clinical Perspective

What Is New?

Among patients hospitalized with chronic limb‐threatening ischemia, frailty is an important independent predictor of revascularization strategy and longitudinal adverse outcomes.

Frail patients with chronic limb‐threatening ischemia were less likely to receive aggressive treatments and more likely to experience death or amputation, regardless of treatment strategy.

What Are the Clinical Implications?

Clinicians should assess frailty in patients with chronic limb‐threatening ischemia and use such information to guide shared decision‐making with patients about their prognosis and the various treatment options available, ranging from invasive surgical revascularization to palliation.

Chronic limb‐threatening ischemia (CLTI) represents the most severe stage of peripheral artery disease and has devastating consequences if left untreated, with a 22% cardiovascular mortality rate and a 22% major amputation rate at 1 year. Determining the optimal treatment strategy for each individual patient in the setting of these aggregate poor outcomes remains unclear. Although prompt revascularization is recommended in multiple major societal guidelines, , it is often uncertain whether an endovascular or surgical approach is superior for a particular patient, as each strategy carries its unique advantages and drawbacks. Furthermore, there may be competing risks of morbidity and mortality that negate the benefit of revascularization altogether and may favor palliation. As such, it is critical that factors associated with prognosis are identified and used to guide clinical decision‐making to improve CLTI outcomes.

Prior attempts at creating clinical risk stratification tools have focused on traditional comorbidities and their cumulative burden. , , , , , However, these characteristics may not fully represent a patient’s risk or candidacy for a specific invasive strategy. Recently, frailty has emerged as an important prognostic factor that can guide treatment selection and help identify patients with cardiovascular disease at greater risk of poor outcomes. , , Multiple methods have been adopted to measure frailty, including in‐person and health record‐based assessments. Despite the demonstrated utility of identifying patients with cardiovascular disease who are frail, the extension of these methods to larger CLTI populations is limited. ,

Therefore, this study involving patients hospitalized with CLTI aimed to examine the association of a health record‐based frailty measure with: (1) treatment choice and (2) outcomes. Such results can not only help inform shared decision‐making with patients on CLTI treatment options but can also help guide future studies examining the optimal treatment strategy for patients with CLTI.

METHODS

Study Population

We included all unique adults aged ≥66 years with an inpatient hospitalization for CLTI in the Centers for Medicare and Medicaid Services Medicare Provider Analysis and Review (MedPAR) database between October 1, 2009 and September 30, 2015. The cut‐off of the study corresponded with the transition to the International Classification of Diseases, Tenth Revision (ICD‐10) coding, as the frailty measure used in this study relies on the International Classification of Diseases, Ninth Revision (ICD‐9) claims codes and has yet to be validated in ICD‐10. CLTI hospitalizations were identified based on whether patients had a primary discharge diagnosis code for atherosclerosis of arteries of the extremities with rest pain (440.22), ulceration (440.23), or gangrene (440.24), as has been done previously (Table S1). CLTI hospitalizations were excluded if patients did not have at least 1 continuous year of data within the MedPAR data set before the CLTI episode, since these data were used to ascertain comorbidities, or if patients had a previous diagnosis of CLTI in this 1‐year lookback period. We also excluded CLTI hospitalizations if patients died or received a major amputation during the index hospitalization, as we aimed to examine a population that was theoretically considered for a revascularization strategy as a definitive CLTI treatment. After exclusion criteria were applied, the first CLTI hospitalization for each patient was included in our study. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Exposure

The primary exposure in the study was frailty, as measured by the Claims‐based Frailty Indicator (CFI) developed by Segal et al, based on administrative claims in the 12 months before CLTI admission. The CFI is a 21‐variable indicator derived from a lookback of primarily ICD‐9 administrative claims that was validated against the Fried Frailty Phenotype, a clinical tool for measuring frailty, in the Cardiovascular Health Study (Tables S2 and S3). We examined an ICD‐9 code in any position in any of the admissions in the 12‐month lookback period to identify frailty. The CFI has subsequently been externally validated, and has been used extensively in studies of populations with cardiovascular disease. , Notably, this frailty measure does not directly include many of the traditional cardiovascular risk factors (eg, hypertension, hyperlipidemia, diabetes, obesity, smoking status), though it does consider congestive heart failure and stroke. For this study, we ascertained frailty status based on ICD‐9 codes from inpatient admissions in the year before the CLTI hospitalization. We used the established CFI threshold of 0.25 to define frailty. We also performed multiple sensitivity analyses using different CFI score thresholds to supplement these results, as outlined below.

Characteristics

Patient characteristics included demographics (age, sex, race), Elixhauser comorbidities, and current or prior tobacco use given known prior associations between these variables and outcomes. All comorbidities were ascertained during the 1‐year lookback period. Revascularization was defined as either endovascular or surgical using claims‐based ICD‐9 procedure codes from the index hospitalization (Table S1). , If a revascularization code was not found during the index admission, a patient was considered not to have had revascularization. Patients who underwent revascularizations with both surgical and endovascular types of procedures during the same index hospitalization contributed data to the surgical revascularization group.

Outcomes

The primary outcome was amputation‐free survival (AFS) measured from the discharge date. Amputation was defined as any major amputation and was identified through validated ICD‐9 and ICD‐10 coding algorithms (Table S1). , Mortality was determined using the vital status information from the Medicare Beneficiary Summary File. Secondary outcomes included the individual components of the composite outcome. Follow‐up data were available through December 2017.

Statistical Analysis

All metrics and normally distributed variables were reported as mean±SD. Non‐normally distributed variables were presented as median (interquartile range). Categorical variables were presented as frequency and percentage. Baseline characteristics were compared between those who were frail and not frail using standardized differences. A standardized difference ≥10% was considered significant.

The association between frailty status and revascularization among patients with CLTI was then examined. Crude rates of revascularization versus no revascularization were compared by frailty status using chi‐squared tests. Multivariable logistic regression was used to determine the adjusted association between revascularization and frailty status. The first model evaluated the association between revascularization (compared with no revascularization) and frailty status among all patients with CLTI, adjusting for demographics, Elixhauser comorbidities, and tobacco use. The second model evaluated the association between surgical revascularization (compared with endovascular revascularization) and frailty status among all patients who underwent a revascularization procedure, again adjusting for demographics, Elixhauser comorbidities, and tobacco use.

Finally, the association between frailty status and outcomes was analyzed, stratified by treatment strategy. Kaplan‒Meier methods were used to estimate AFS by frailty status among patients who did not receive revascularization and, separately, among patients who received revascularization. In addition, among patients who received revascularization, patients were stratified by frailty status among those receiving endovascular treatment and, separately, surgical treatment. Cox proportional hazards regression was then used to evaluate the association between frailty status and outcomes, adjusted for patient characteristics. The proportional hazards assumption was evaluated by plotting the hazard ratios (HRs) between frail and non‐frail individuals with survival over time (Figure S1). Because of the possibility of unmeasured confounding and treatment selection bias, we elected to create separate models for patients who did not receive revascularization and for patients who received revascularization. Patients receiving revascularization were further stratified into those receiving endovascular or surgical treatment. To account for the competing risk of death, Fine‐Gray methods were used to analyze the outcome of major amputation.

As sensitivity analyses, the relationship between the degree of frailty and the risk of death or major amputation at 1 year was evaluated, considering frailty as a continuous variable using a cubic spline model with 4 knots. We additionally examined baseline characteristics and receipt of revascularization across quartiles of the frailty indicator, and repeated Kaplan‒Meier analyses of AFS by quartiles among patients who did not receive revascularization and, separately, among patients who received revascularization. Similarly, we repeated Kaplan‒Meier analyses of AFS by quartiles among patients who received surgical revascularization and, separately, among patients who received endovascular revascularization.

A 2‐sided P<0.05 was considered statistically significant without adjustment for multiple comparisons. All analyses were performed using SAS 9.4 (Cary, NC, USA). The study was approved by the institutional review board of Beth Israel Deaconess Medical Center, with a waiver of informed consent for retrospective data analysis.

RESULTS

During the study period, 85 060 patients met criteria and were included in the analysis, of which 35 484 (42%) were classified as frail (Figures S2 and S3). Median follow‐up in the whole cohort was 2.34 years (interquartile range, 3.32 years). Patients who were frail were more likely to be older, women, of Black or other race, and non‐smokers (Table 1). Frail patients also had a greater burden of both cardiovascular and non‐cardiovascular conditions, including congestive heart failure, valvular heart disease, pulmonary circulation disease, renal failure, and anemia. Mean number admissions before the index admission among non‐frail patients was 1.2 versus 2.1 for frail patients.

Table 1

Baseline Characteristics of Patients With CLTI by Frailty Status

Subject characteristic	Not frail (n=49 576)	Frail (n=35 484)	Standardized difference*
Demographics
Age (y), mean (SD)	75.77 (5.36)	86.01 (6.01)	−1.798
Men	28 833 (58.16)	13 655 (38.48)	0.402
Race
White	41 773 (84.26)	26 081 (73.50)	0.266
Black	6043 (12.19)	6777 (19.10)	−0.191
Other †	1760 (3.55)	2626 (7.40)	−0.170
Elixhauser comorbidity variables
Summary comorbidity index, mean (SD)	3.79 (2.05)	4.58 (2.40)	0.35
Acquired immune deficiency syndrome	45 (0.09)	13 (0.04)	0.020
Alcohol abuse	147 (0.30)	40 (0.11)	0.042
Chronic blood loss anemia	893 (1.80)	847 (2.39)	−0.041
Chronic pulmonary disease	15 984 (32.24)	9881 (27.85)	0.096
Coagulopathy	2436 (4.91)	2422 (6.83)	−0.082
Congestive heart failure	6798 (13.71)	12 670 (35.71)	−0.527
Deficiency anemias	10 665 (21.51)	11 509 (32.43)	−0.248
Depression	2797 (5.64)	4374 (12.33)	−0.236
Diabetes w/ chronic complications	10 659 (21.50)	7570 (21.33)	0.004
Diabetes w/o chronic complications	19 750 (39.84)	12 933 (36.45)	0.070
Drug abuse	34 (0.07)	13 (0.04)	0.013
Fluid and electrolyte disorders	12 826 (25.87)	14 613 (41.18)	−0.329
Hypertension	39 953 (80.59)	28 786 (81.12)	−0.013
Hypothyroidism	5640 (11.38)	6529 (18.40)	−0.198
Liver disease	605 (1.22)	352 (0.99)	0.022
Lymphoma	452 (0.91)	358 (1.01)	−0.010
Metastatic cancer	603 (1.22)	336 (0.95)	0.026
Obesity	3788 (7.64)	1837 (5.18)	0.101
Other neurological disorders	3189 (6.43)	5895 (16.61)	−0.323
Paralysis	2031 (4.10)	2677 (7.54)	−0.147
Peptic ulcer disease without bleeding	21 (0.04)	24 (0.07)	−0.013
Psychoses	973 (1.96)	1531 (4.31)	−0.135
Pulmonary circulation disease	1032 (2.08)	1931 (5.44)	−0.177
Renal failure	12 660 (25.54)	12 404 (34.96)	−0.206
Rheum. arthritis/collagen vascular diseases	1930 (3.89)	2131 (6.01)	−0.098
Solid tumor w/out metastasis	1452 (2.93)	1129 (3.18)	−0.015
Valvular disease	1607 (3.24)	2968 (8.36)	−0.220
Weight loss	3482 (7.02)	5160 (14.54)	−0.244
Additional variables
Smoking	25 386 (51.21)	11 626 (32.76)	0.381
No. admissions preceding index (mean, SD) ‡	1.20 (1.79)	2.14 (2.48)	0.435

Standardized difference calculated as: for continuous variables and for categorical variables. Comorbidities were ascertained during both the index admission and the 1‐year lookback period.

†

Other race or ethnicity includes those who identify as Asian, Hispanic, North American Native, or Other and those with race unknown.

‡Number of admissions preceding index admission was not included in adjusted models given collinearity with frailty measure.

CLTI indicates chronic limb‐threatening ischemia.

†

Frailty and Treatment Choice

In unadjusted analyses, frail patients were less likely to receive revascularization than non‐frail patients (60.2% versus 76.4%, respectively; P<0.001) (Table 2). This relationship persisted after adjustment (adjusted OR, 0.78; 95% CI, 0.75‒0.82; P<0.001). Of patients who received revascularization, those who were frail were more likely to receive endovascular treatment when compared with non‐frail patients (59.5% versus 40.0%, respectively; P<0.001). Conversely, non‐frail patients who underwent revascularization were more likely to receive surgical treatment (60.0% versus 40.5% of frail patients; P<0.001). After adjustment, being classified as frail remained associated with a lower likelihood of receiving surgical treatment (adjusted OR, 0.76; 95% CI, 0.72‒0.81; P<0.001).

Table 2

Association of Treatment Choice for CLTI by Frailty Status

Cohort	Not frail	Frail	Chi‐squared P value	Adjusted OR for treatment modality for frailty vs non‐frail individuals* (95% CI)	P value
	n (%)	n (%)
All patients	49 576	35 484	<0.0001	0.784 (0.746‒0.824) †	<0.0001
Revascularization	37 887 (76.4)	21 379 (60.2)
No revascularization	11 689 (23.6)	14 105 (39.8)
Patients treated with revascularization	37 887	21 379	<0.0001	0.764 (0.724‒0.808) ‡	<0.0001
Surgical	22 734 (60.0)	8654 (40.5)
Endovascular	15 153 (40.0)	12 725 (59.5)

CLTI indicates chronic limb‐threatening ischemia; and OR, odds ratio.

Multivariable logistic regression models controlling for all covariates listed in Table 1 unless otherwise specified.

OR of revascularization (relative to no revascularization).

OR of surgical treatment (relative to endovascular treatment).

Frailty and Outcomes

Overall, the median AFS was 2.5 years (interquartile range, 0.7 years‒4.1 years) for non‐frail patients and 0.8 years (interquartile range, 0.2 years‒2.6 years) for frail patients. The Kaplan‒Meier cumulative estimate of AFS at 8.25 years of follow‐up was greater among non‐frail patients compared with frail patients for both those who were revascularized (17.9% non‐frail versus 4.0% frail, P<0.001) and those who were not revascularized (10.8% non‐frail versus 1.8% frail, P<0.001) (Figure 1). The association between frailty and worse AFS also persisted when all patients were considered in aggregate (Figure S4). Of those who received revascularization, non‐frail patients also had a greater cumulative AFS than frail patients, irrespective of revascularization strategy (surgical: 19.9% non‐frail versus 5.0% frail, P<0.001; endovascular: 15.1% non‐frail versus 3.4% frail, P<0.001) (Figure 2).

Figure 1

Amputation‐free survival among patients with chronic limb‐threatening ischemia by frailty status, stratified by treatment choice.

A, Among patients undergoing revascularization. B, Among patients not undergoing revascularization. HR indicates hazard ratio.

Figure 2

Amputation‐free survival among patients with chronic limb‐threatening ischemia undergoing revascularization by frailty status, stratified by revascularization strategy.

A, Surgical revascularization. B, Endovascular revascularization. HR indicates hazard ratio.

In adjusted analysis, frailty remained associated with an increased risk of death or major amputation, regardless of whether revascularization was performed (revascularized cohort: hazard ratio [HR], 1.34; 95% CI, 1.30‒1.38; P<0.001; non‐revascularized cohort: HR, 1.22; 95% CI, 1.17‒1.27; P<0.001) (Figure 1). The difference in death or major amputation was primarily driven by an increased risk of death among those that were frail (Figures S5 and S6). Notably, frailty was a stronger independent predictor of death or major amputation compared with many traditional risk factors (including tobacco use, diabetes, and obesity), among both the revascularized and non‐revascularized cohorts (Figure 3, Table S4).

Figure 3

Top predictors of amputation or death for patients with chronic limb‐threatening ischemia using Cox proportional hazards.

A, Among patients undergoing revascularization. B, Among patients not undergoing revascularization. C, Among patients undergoing surgical revascularization. D, Among patients undergoing endovascular revascularization. Figures include predictors with hazard ratio >1.2. See Tables S4 and S5 for full model results.

Of patients who were revascularized, frailty was also a strong independent predictor of AFS, irrespective of revascularization strategy (surgical cohort: adjusted HR, 1.36; 95% CI, 1.31‒1.42; P<0.001; endovascular cohort: adjusted HR, 1.29; 95% CI, 1.24‒1.35; P<0.001) (Figure 2). The difference in death or major amputation was again primarily driven by an increased risk of death among those that were frail (Figures S7 and S8). Frailty was a stronger predictor of death or major amputation compared with many traditional risk factors, including age, race, sex, tobacco use, diabetes, and obesity (Figure 3, Table S5).

Frailty as a Continuous Measure

When frailty was examined as a continuous variable, higher values of frailty were associated with a higher risk of death or major amputation at 1 year among the total cohort, as well as when stratified by revascularization status (Figure S9). When frailty was examined by quartiles, patients with greater degree of frailty showed greater burden of both cardiovascular and non‐cardiovascular conditions as well as a sequentially lower likelihood of receiving any revascularization, as well as surgical revascularization, if revascularized (Tables S6 and S7). Patients with a greater degree of frailty had sequentially worse outcomes, regardless of whether revascularization was pursued or the type of revascularization (Figures S10 and S11).

DISCUSSION

In this retrospective cohort study of Medicare beneficiaries with long‐term follow‐up, we evaluated the impact of a claims‐based measurement of frailty on treatment selection and outcomes among patients hospitalized with CLTI. We found the following notable results. Patients hospitalized for CLTI were, on average, more frail than the community‐dwelling Medicare beneficiary population. Frailty was a useful tool to discriminate whether a patient underwent a revascularization strategy versus a non‐invasive treatment approach, irrespective of most traditional comorbidities. The presence of frailty also impacted who underwent an endovascular revascularization strategy versus surgical. Furthermore, regardless of treatment strategy, frailty was independently associated with worse outcomes, even after adjusting for an array of demographics and comorbidities. This association was primarily driven by an increased risk of death and not necessarily by differences in amputation. These results suggest that frailty can be used to prognosticate future outcomes and guide shared decision‐making with patients about the various treatment options available.

We found evidence that treatment selection for patients hospitalized with CLTI is influenced by frailty, a multidimensional syndrome characterized by decreased reserve and diminished resistance to stressors. Previous studies have shown that patients with CLTI who receive conservative or endovascular therapy are likely to have more comorbidities than those who receive surgical therapy. , Guidelines currently recommend endovascular over surgical treatment for CLTI in patients with substantial comorbidities, which may place them at higher risk of postoperative complications from surgical revascularization. Our finding that frail patients are more likely to receive endovascular intervention, despite controlling for comorbidities, reflects how clinicians may already consider frailty in the selection of appropriate treatment options for patients with CLTI. However, a frailty assessment may be one method to improve communication about what characteristics of the patient may make them a better or worse candidate for a particular revascularization strategy, thereby improving patient counseling and comprehension.

This study extends existing investigation about factors affecting CLTI outcomes. Several risk scores have been developed to predict 30‐day or 1‐year amputation‐free survival among patients with CLTI undergoing surgical revascularization, , or mortality after endovascular or surgical revascularization at 2 or 5 years. , However these scores only capture select comorbidities and are not frequently used in clinical practice. We include many of these risk score components in our multivariate analysis and show that frailty remains a strong, independent factor that improves patient risk stratification for future clinically important events, regardless of treatment strategy. As such, frailty may be a final common pathway that integrates information from other established factors and risk scores. Several studies using varied frailty metrics have shown an association between frailty and adverse functional outcomes after peripheral artery disease procedures more broadly. Additionally, small, single‐center studies have found that frailty was associated with a lower AFS or overall survival at 2 years among patients with CLTI who received revascularization. , We found similar results in a large national sample of >85 000 Medicare beneficiaries with CLTI with follow‐up available up to 8 years. Additionally, we found that frailty was associated with worse outcomes among hospitalized patients who do not receive revascularization, which comprises a large proportion of Medicare patients with CLTI. Thus, our study further establishes the role of phenotypic frailty as a strong, independent secular predictor of CLTI outcomes, independent of physician management.

These results have important implications for the treatment of patients hospitalized with CLTI and for future studies evaluating CLTI treatment options. These findings suggest that physicians should routinely assess frailty in clinical management of patients hospitalized with CLTI. While multiple different conceptualizations of frailty exist, the one used for the current study, which is anchored to a physical phenotype, can be easily quantified using a comprehensive geriatric assessment, and is thus readily translatable into practice. Therefore, clinicians can use such information to guide shared decision‐making with patients about their prognosis and the various treatment options available, ranging from invasive surgical revascularization to palliation. For instance, if a frail patient hospitalized with CLTI is informed of an expected overall median AFS of 0.8 years (versus 2.5 years in a non‐frail patient) in considering treatment options and recovery times, less invasive treatment may be chosen in shared decision‐making. These results highlight the need for team‐based care of patients with CLTI as well as the need for additional patient‐centered outcomes to enhance the shared‐decision making process. Additionally, the additive prognostic value of frailty suggests that observational studies comparing treatment options among patients hospitalized with CLTI may be confounded if frailty is not considered. Finally, our evidence that frailty affects CLTI outcomes suggests that future randomized trials of CLTI treatment should consider examining outcomes among frail patients separately, to individualize treatment for this high‐risk subgroup.

Our analysis must be interpreted in light of its limitations. First, for the primary analysis, we considered frailty as a dichotomous exposure for pragmatism. However, in supplemental analysis, similar relationships between quartiles of frailty and AFS were observed, and a continuous measure of frailty was associated with increased 1‐year mortality in a dose‐dependent manner, further emphasizing the importance of frailty as a predictor of outcomes. Second, given evidence that frailty likely impacts treatment selection, we chose not to compare outcomes among frail patients across different treatment strategies to avoid confounding by indication. Third, the frailty scale used is based on longitudinal ICD‐9 claims data. As such, the application of this specific scale for future investigation or clinical utility is limited, both by the availability of these data and by the transition to the ICD‐10 coding system, though work to crosswalk these claims to ICD‐10 codes is ongoing. Furthermore, we studied a more historic cohort of patients; although this enabled us to capture long‐term follow‐up of these patients, treatment options may have since evolved. Nonetheless, the utility of this analysis is supported by the correlation between the claims‐based frailty scale used in this study and the widely adopted in‐person Fried frailty assessment. Fourth, we do not account for patients who may become frail in the follow‐up time; however, such analysis pertains primarily to the treatment decision faced at time of CLI and time‐updating is of modest benefit with a non‐frail median survival of 2.4 years. Fifth, we only included patients hospitalized for CLTI, which may represent a more severe form of the disease and did not include the minority of patients who were discharged with plans for outpatient intervention or treated primarily in the outpatient setting. Lastly, the MedPAR data set lacks granular anatomical and procedural characteristics or medications that may also be prognostic of future adverse events.

CONCLUSIONS

In this large nationwide analysis of hospitalized Medicare beneficiaries with CLTI, we find that frailty is an important independent predictor of revascularization strategy and longitudinal adverse outcomes in patients who did and did not undergo revascularization. Future work investigating the optimal treatment strategy for this high‐risk subgroup is warranted.

Sources of Funding

None.

Disclosures

Dr Butala is funded by the John S. LaDue Memorial Fellowship at Harvard Medical School, Boston, MA and reports consulting fees and ownership interest in HiLabs, outside the submitted work. Dr Secemsky receives grants from AstraZeneca, BD Bard, Boston Scientific, Cook Medical, CSI, Medtronic, Philips, and UCSF. He consults for CSI, Medtronic, and Philips and is on the speaking bureau of BD Bard, Cook Medical and Medtronic, outside the submitted work. Dr Schermerhorn has received support from Abbott, Cook Medical, Endologix, Medtronic, and Philips, outside the submitted work. Dr Beckman has served as a consultant for AstraZeneca, Boehringer Ingelheim, Bristol‐Myers Squibb, Merck, Novo Nordisk, and Sanofi; and has served on the Data Safety and Monitoring Board of Bayer and Novartis, outside the submitted work. Dr Shishehbor has served on the Scientific Advisory Boards of Medtronic, Abbott Vascular, Phillips, Terumo, and Boston Scientific, outside the submitted work. Dr Yeh reports additional grant support from Abiomed, Astra Zeneca, and Boston Scientific and consulting fees from Abbott, Boston Scientific, Medtronic, and Teleflex, outside the submitted work. The remaining authors have no disclosures to report.

Tables S1–S7

Figures S1–S11

Click here for additional data file.