Percutaneous Thrombectomy for Acute Limb Ischemia is Associated With Equivalent Limb and Mortality Outcomes Compared to Open Thrombectomy

Abstract

Background:

Acute limb ischemia (ALI) carries a 15–20% risk of combined death or amputation at 30 days and 50–60% at 1 year. Percutaneous mechanical thrombectomy (PT) is an emerging minimally invasive alternative to open thrombectomy (OT). However, ALI thrombectomy cases are omitted from most quality databases, limiting comparisons of limb and survival outcomes between PT and OT. Therefore, our aim was to compare in-hospital outcomes between PT and OT using the National Inpatient Sample (NIS).

Methods:

We analyzed survey-weighted NIS data (2015–2020) to include emergent admissions of aged adults (50+ years) with a primary diagnosis of lower extremity ALI undergoing index procedures within two days of hospitalization. We excluded hospitalizations with concurrent trauma or dissection diagnoses and index procedures using catheter-directed thrombolysis. Our primary outcome was composite in-hospital major amputation or death. Secondary outcomes included in-hospital major amputation, death, in-hospital reintervention (including angioplasty/stent, thrombolysis, PT, OT, or bypass), and extended length of stay (eLOS; defined as LOS >75^th percentile). Adjusted odds ratios (aORs) with 95% confidence intervals (95%CI) were generated by multivariable logistic regression, adjusting for demographics, frailty (Risk Analysis Index), secondary diagnoses including atrial fibrillation and peripheral artery disease (PAD), hospital characteristics, and index procedure data including the anatomic thrombectomy level and fasciotomy. A priori subgroup analyses were performed using interaction terms.

Results:

We included 23,795 survey-weighted ALI hospitalizations (mean age 72.2 years, 50.4% female, 79.2% White, 22.3% frail) with 7,335 (30.8%) undergoing PT. Hospitalization characteristics for PT vs OT differed by atrial fibrillation (28.7% vs 36.5%, p<.0001), frequency of intervention at the femoropopliteal level (86.2% vs 88.8%, p=.009), and fasciotomy (4.8% vs. 6.9%, p=.006). In total, 2,530 (10.6%) underwent major amputation or died. Unadjusted (10.1% vs 10.9%, p=.43) and adjusted (aOR=0.96 [95%CI, 0.77–1.20], p=.74) risk did not differ between groups. PT was associated with increased odds of reintervention (aOR=2.10 [95%CI 1.72–2.56], p<.0001) when compared to OT, but this was not seen in the tibial subgroup (aOR=1.31 [95% CI, 0.86–2.01], p=.21, p_interaction<.0001). Further, 79.1% of PT hospitalizations undergoing reintervention were salvaged with endovascular therapy. Lastly, PT was associated with significantly decreased odds of eLOS (aOR=0.80 [95%CI 0.69–0.94], p=.005).

Conclusion:

PT was associated with comparable in-hospital limb salvage and mortality rates compared to OT. Despite an increased risk of reintervention, most PT reinterventions avoided open surgery and PT was associated with a decreased risk of eLOS. Thus, PT may be an appropriate alternative to OT in appropriately selected patients.

Table of Contents Summary

Limb salvage and mortality outcomes were equivalent in percutaneous vs. open thrombectomy in a retrospective study of 23,795 hospitalizations for acute limb ischemia undergoing revascularization within the National Inpatient Sample. The authors suggest percutaneous thrombectomy is a suitable alternative to open thrombectomy in properly selected patients.

Introduction

Acute limb ischemia (ALI) results from a sudden decrease in limb perfusion that threatens limb viability. Among adults in the United States (U.S.), ALI occurs in 1.5 per 10,000 person-years and despite significant advances in medical and surgical therapy, over half will ultimately require major amputation within one year^1–3. Traditional treatment options include open thrombectomy (OT), surgical bypass, and catheter directed thrombolysis (CDT), with treatment selection based on both etiology and degree of ischemia as determined by Rutherford classification of ALI^2,4,5. Typically, open approaches are preferred in patients with ALI secondary to embolic events or with compromised motor function, requiring rapid revascularization. Conversely, CDT is frequently favored among patients with acute thrombosis of pre-existing peripheral arterial disease (PAD) and/or prior interventions, as they often present with milder symptoms^2,4,5.

Minimally invasive approaches, when appropriately selected, can be associated with lower physiologic stress when compared to open surgical interventions and may improve outcomes, especially in an aged and frail patient population^6–8. Endovascular approaches can also be attractive in re-operative fields due to increased wound infection risk with open surgery⁹. However, the existing endovascular treatment options have limitations. Specifically, CDT can require 24–48 hours of thrombolytic infusion to clear thrombus and restore perfusion. Thus, CDT is not feasible in those with a high bleeding risk or Rutherford IIB ischemia, who have a dense motor deficit requiring prompt revascularization.

More recently, the addition of percutaneous thrombectomy (PT) to the armamentarium has broadened treatment options for ALI. Several devices are available with different mechanisms for thrombus clearance, including mechanical and suction thrombectomy, both alone and in combination with thrombolytics. PT offers several potential benefits. It allows for minimally invasive, single-session thrombus clearance, making it a viable option in Rutherford IIB ischemia. It also potentially obviates the need for thrombolytics, which carry an increased risk of bleeding, particularly amongst elderly patients^4,10,11.

The study of ALI is excluded from commonly used quality improvement databases (OT is excluded from the Vascular Quality Initiative [VQI] and PT is excluded from the National Surgical Quality Improvement Program [NSQIP]). Thus, current literature on PT is limited to single-center retrospective studies comparing PT to CDT and studies comparing open revascularization to CDT only^12–17. We therefore examined in-hospital outcomes between PT and OT using the National Inpatient Sample (NIS), hypothesizing that PT would be associated with improved combined amputation and death rates.

Methods

Our retrospective study using the NIS database was determined to be exempt from human subjects’ review by the University of Pittsburgh’s Human Research Protection Office (STUDY20110441). Data were presented in accordance with the NIS, Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality Use Agreements and Strengthening the Reporting of Observational Studies in Epidemiology guidelines^18,19.

Data Source and Patient Cohort

We examined all hospitalizations for ALI in which patients underwent PT or OT in the U.S. using NIS (2015–2020) data. The NIS samples 20% of non-federal acute care hospital discharges in the United States from all payers, allowing for population-level, survey-weighted estimates of hospitalizations and outcomes^20,21. NIS data includes demographics, up to 40 diagnosis codes using International Classification of Disease, 10^th Edition, Clinical Modifications (ICD-10-CM), up to 25 procedure codes using International Classification of Disease, 10^th Edition, Procedure Coding System (ICD-10-PCS), procedure day, in-hospital mortality and discharge disposition²².

In our primary cohort, we included non-elective hospitalizations with a primary ICD-10-CM diagnosis code of lower extremity ALI and underwent PT or OT within 2 days of admission as defined by ICD-10-PCS (Figure 1, Supplementary Table 1), mapped from previously described ICD-9-PCS^3,23. Using procedure days from admission, we defined the index thrombectomy procedure as the one that occurred on the earliest hospital day. Those who underwent both PT and OT on the first procedural day were classified as PT and analyzed in an intention-to-treat fashion, assuming that both procedures were done within the same operation and the likelihood of PT failure converting to OT was higher than OT failure converting to PT. To minimize misclassification, we excluded hospitalizations with secondary diagnoses of trauma²⁴ or dissection²⁵ (Figure 1, Supplementary Table 1). We further excluded patients <50 years old to minimize the inclusion of outliers, such as patients with traumatic ischemia, within our cohort. We also eliminated hospitalizations with concomitant CDT, because its use is only a viable treatment among patients with limited ischemia and low bleeding risk.

Figure 1.

Inclusion and Exclusion Criteria

Patient demographic and admitting hospital information was available within the NIS at time of hospitalization. Hospital facility information includes geographic location, teaching status, and bed size. Comorbid conditions were defined using secondary ICD-10-CM codes in accordance with NIS guidance, using Elixhauser Comorbidity Index²⁶ and Risk Analysis Index (RAI)²⁷. Other covariates were taken from previously published codes^3,28–30 (Supplementary Table 1). The RAI is a semi-continuous frailty score ranging from 0–81 that includes a variety of demographics and comorbidities (cancer, weight loss, renal failure, congestive heart failure, dyspnea, cognitive impairment, and functional status) that has been validated to predict postoperative complications and mortality in patients with vascular disease²⁷. The RAI was used as a continuous variable in our initial multivariable model but was dichotomized into frail (RAI≥ 26) and not frail (RAI<26) for subgroup analyses²⁷. Fasciotomy was used as a surrogate for ALI severity, as it’s independently associated with severity³¹. In-hospital amputation and reintervention procedures were also defined using ICD-10-PCS, recapitulated from the VQI³² (Supplementary Table 1).

Outcomes

The primary outcome was defined as the composite of in-hospital ipsilateral major (above-ankle) amputation and death. Secondary outcomes included in-hospital ipsilateral major amputation, death, in-hospital reintervention (including angioplasty/stent, thrombolysis, PT, OT, or bypass, regardless of laterality), and extended length of stay (eLOS; defined as LOS greater than 75^th percentile within the entire thrombectomy cohort²⁷).

Statistical Analysis

Survey-weighted data were reported in accordance with NIS guidelines. Continuous variables were presented as means with standard errors (SE) and categorical variables were presented as proportions and 95% confidence intervals (95%CI). Baseline characteristics were compared using Students t- and chi-squared testing.

Multivariable logistic regression generated adjusted odds ratios (aOR) for outcomes. A priori covariates included patient demographics (i.e. age, sex, race, insurance type, hospital type, and median income of the patients’ residential zip code), pertinent comorbid conditions (i.e. pre-existing PAD, atrial fibrillation, hypercoagulable states, and smoking status), ALI severity (using fasciotomy³¹), frailty (i.e. RAI), and treated anatomic level (i.e. aortoiliac, femoropopliteal, and tibial). Multilevel thrombectomy was defined as more than one anatomic level treated, as granular data were only present for procedures, not anatomic level of disease. A priori subgroup analyses (age over 70, sex, race, frailty, PAD, atrial fibrillation, diabetes, hypercoagulable state, smoker, multilevel treatment, distal anatomic level, and fasciotomy) were performed assessing the moderation of the treatment effect using interaction terms within multivariable modeling. Post-prediction rates were calculated from our multivariable model and used to determine adjusted outcome equivalence. Outcome equivalence was defined by a margin of 3% difference between treatment groups based off objective performance goals of catheter-based treatments in chronic limb threatening ischemia, with 90% confidence intervals (90%CI) as this is a two-sided test (CI=[1–2α]*100%)^33,34.

Five sensitivity analyses were performed using alternative cohort and exposure definitions. First, we included an additional ICD-10-CM in our definition of ALI (I99.8, Other disorder of the circulatory system), given this non-specific code includes “ischemia of lower extremity” within its definition³⁵. Second, we included patients who underwent concomitant thrombolysis during the index thrombectomy procedure. Third, hospitalizations with both PT and OT on the first procedural day were classified as OT instead of PT, as they were within the primary analysis. Fourth, we only included patients who underwent index thrombectomy within 1 day of hospitalization. Finally, we completed a sensitivity analysis including other clinically relevant patient (chronic antiplatelet/anticoagulation use, hypertension, prior angioplasty/stent, prior bypass), intraoperative (bilateral intervention), and postoperative factors (compartment syndrome), which were omitted from the primary a priori model in an effort to minimize both overfitting of the model and use of non-validated ICD-10 codes.

Data analysis and figures were completed using Stata 17.0 (Stata Corp, College Station, TX) and PRISM 10 (GraphPad Software, La Jolla, CA). All graphs were created in Prism (GraphPad 9.0).

Results

Baseline characteristics

Among 183,649,844 estimated hospitalizations in the NIS (2015–2020), 67,095 (3.7%) had a primary diagnosis of ALI. Of those, 23,795 (35.5%) hospitalizations met procedure specific inclusion and exclusion criteria (mean age 72.2 [SE 0.2], 50.4% female, 79.2% White and 22.3% frail) and 7,335 (30.8%) underwent PT (Figure 1). Across index thrombectomy procedures, 35.8% were multilevel and 6.2% included fasciotomy. PT hospitalizations were more frequently located at urban non-teaching hospitals and in the Southern U.S., and less frequently located in the Northeast when compared to OT hospitalizations. PT hospitalizations were less likely to have diagnoses of atrial fibrillation, congestive heart failure, and chronic lung disease. PT procedures were less likely to include the femoropopliteal segment and fasciotomy (Table 1).

Table 1.

Baseline Characteristics by Index Thrombectomy Procedure

Preoperative variables	Index Percutaneous Thrombectomy (n=7,335)	Index Open Thrombectomy (n=16,460)	p-value
Demographics
Age, years	72.2(0.2)	72.6(0.2)	.002
Female sex	49.1(46.5–51.6)	50.9(49.3–52.6)	.23
Race/ethnicity*			.97
White	79.5(77.2–81.6)	79.1(77.6–80.6)
Black	10.5(9–12.3)	10.8(9.8–12)
Hispanic	6.8(5.5–8.4)	6.6(5.7–7.6)
Asian or Pacific Islander	1(0.6–1.7)	1.2(0.9–1.7)
American Indian or Alaska Native	0.1(0–0.6)	0.2(0.1–0.4)
other	2(1.4–2.9)	2(1.6–2.6)
Hospital Region			.0003
Northeast	14.8(12.9–17)	20.2(18.4–22.2)
Midwest	24.6(22.1–27.3)	24.1(22.2–26.2)
South	42.4(39.4–45.4)	38(35.8–40.3)
West	18.2(16–20.7)	17.6(16–19.4)
Bed Size°			.40
Small	14.4(12.5–16.5)	13.3(11.9–14.9)
Medium	27.5(24.9–30.2)	26.4(24.5–28.4)
Large	58.1(55.2–61.1)	60.3(58–62.5)
Hospital Type			.001
Rural	3.1(2.3–41)	3.3(2.6–4)
Urban nonteaching	19.8(17.5–22.2)	15.2(13.8–16.8)
Urban teaching	77.2(74.6–79.5)	81.5(79.8–83.1)
Insurance type			.08
Medicare	67.8(65.3–70.1)	71(69.4–72.6)
Medicaid	8.4(7.1–9.9)	8.1(7.2–9.1)
Private	18.5(16.6–20.6)	15.6(14.4–16.9)
Self-pay	2.4(1.7–3.3)	2.8(2.3–3.5)
Other**	2.9(2.2–4)	2.4(1.9–3)
Income quartile⧺			.66
quartile 1	31.1(28.6–33.6)	29.8(28.1–31.6)
quartile 2	28.5(26.2–30.9)	28(26.5–29.7)
quartile 3	22.2(20.1–24.5)	23.7(22.3–25.2)
quartile 4	18.2(16.2–20.4)	18.4(17–19.9)
Comorbid conditions
Peripheral arterial disease	70.8(68.4–73)	68.6(66.9–70.2)	.13
Prior peripheral vascular intervention	6.4(5.3–7.9)	4.8(4.2–5.6)	.021
Lower extremity aneurysm	4.0(3.1–5.2)	2.7(2.3–3.4)	.024
Atrial fibrillation	28.7(26.4–31.1)	36.5(34.9–38.2)	<0.0001
Congestive heart failure	25.8(23.6–28.1)	29.6(28–31.1)	.007
Hypertension	79.1(76.9–81.1)	80.3(78.8–81.6)	.35
Smoker	51.5(48.8–54.1)	53.2(51.5–54.9)	.27
Chronic lung disease	23.7(21.6–26)	28(26.5–29.6)	.002
Renal failure	21.4(19.3–23.6)	21.1(19.8–22.6)	.84
Diabetes	30.8(28.5–33.3)	29.5(28–31.1)	.38
Hypercoagulable	3.4(2.6–4.5)	4.1(3.5–4.8)	.25
Preoperative Anticoagulation	22.6(20.5–24.7	23.9(22.4–25.4)	.32
Cancer	6.7(5.6–8.1)	6.3(5.6–7.2)	.60
Risk Analysis Index	29.5(0.2)	30.1(0.2)	.04
Operative details
Multilevel	34.8(32.3–37.3)	36.3(34.6–38)	.32
Bilateral Treated Levels	13.0(11.3–14.8)	14.1(13.0–15.5)	.27
Aortoiliac	27.1(24.8–29.4)	25.5(24–27)	.25
Femoropopliteal	86.2(84.3–87.8)	88.8(87.7–89.8)	.009
Tibial	23.2(21.1–25.5)	25.8(24.3–27.3)	.07
Fasciotomy	4.8(3.8–6)	6.9(6–7.8)	.006
Compartment Syndrome	3.9(3.01–5.0)	5.5(4.8–6.4)	.02
Weekend admission	23.9(21.7–26.1)	24.4(23–26)	.67

Amputation and Death

2,530 (10.6%) estimated ALI hospitalizations resulted in ipsilateral major amputation or death, with no difference between PT and OT on univariable (10.1% vs 10.9%, p=.43) or multivariable analysis (aOR=0.96 [95%CI, 0.77–1.20], p=.74). Post-prediction estimates of the composite primary outcome were equivalent for PT (10.2% [90%CI, 10.0–10.5%]) compared to OT (10.9% [90%CI, 10.7–11.1%]) based on a priori equivalence margins. Significant predictors for combined death and amputation include increasing frailty, pre-existing PAD, multilevel thrombectomy, and concurrent fasciotomy (Figure 2).

Figure 2.

Multivariable Logistic Regression of Composite Amputation and Death

Major amputations occurred in 1,120 (4.7%) ALI hospitalizations with no difference between PT and OT on univariable (4.1% vs 5.0%, p=.28) or multivariable (aOR=0.82 [95%CI, 0.60–1.11]), p=.21) analysis. Older age was associated with a reduced risk of amputation. Procedural predictors for amputation included both femoropopliteal or tibial thrombectomy compared to aortoiliac thrombectomy as well as concurrent fasciotomy (Figure 3a)

Figure 3.

Multivariable Logistic Regression of Secondary Outcomes

Death occurred in 1,495 (6.3%) ALI hospitalizations with no difference between PT and OT for on univariable (6.2 vs 6.3%; p=.88) or multivariable (aOR=1.05 [95%CI, 0.79–1.38], p=.75) analysis. Predictors of death included multilevel thrombectomy and fasciotomy. Femoropopliteal or tibial thrombectomy (compared to aortoiliac thrombectomy) and smoking were protective (Figure 3b).

In prespecified subgroup analyses, no interaction terms evaluating moderation reached significance for composite amputation and death, amputation alone, or death alone.

In-hospital Reintervention

In-hospital reintervention occurred in 2,480 (10.4%) ALI hospitalizations, with significantly more observed reinterventions in the PT group on univariable (15.7% vs 8.1%, p<.0001) and multivariable analysis (aOR=2.10 [95%CI, 1.72–2.56], p<.0001). Multilevel treatment was associated with decreased risk, while PAD was associated with increased risk (Figure 3c). In subgroup analysis, the association between PT and increased in-hospital reintervention was not present for those who underwent tibial (aOR=1.31 [95% CI, 0.86–2.01], p=.21) compared to femoropopliteal interventions (aOR=2.82 [95% CI, 2.21–3.60], p<.0001; p_interaction<.0001). Additionally, the association was lesser in magnitude for those who underwent multilevel (aOR=1.50 [95%CI 1.05–2.14] compared to single-level thrombectomy p_interaction=0.03; Supplementary Figure 1a).

In the PT group, 79.1% of hospitalizations requiring reintervention underwent endovascular salvage, 13.9% underwent open reintervention, and 7.0% underwent hybrid reintervention. Angioplasty/stent (53.5%) reinterventions were the most common, while 49.1% underwent repeat PT and 30.9% underwent CDT. In the OT group, only 17.7% of hospitalizations requiring reintervention underwent endovascular salvage, with the majority (60.5%) undergoing repeat open revascularization, and 21.8% undergoing both. The most common reintervention was repeated OT (72.9%) while 20.3% required a bypass (Table 2).

Table 2.

Reintervention Type by Index Thrombectomy Procedure

Reintervention	Index Percutaenous Thrombectomy, (n=1150)	Index Open Thrombectomy (n=1330)
Bypass	7.8(4.9–12.3)	20.3(15.5–26.5)
Open Thrombectomy	16.1(11.7–22.1)	72.9(62.8–84.6)
Percutaneous Thrombectomy	49.1(40.7–59.1)	10.5(7.3–15.2)
Lysis	30.9(24.4–38.9)	12.0(8.4–17.1)
Angioplasty/stent	53.5(45.0–63.4)	31.2(24.9–39.1)

Extended Length of Stay

eLOS occurred in 6,170 (25.9%) hospitalizations, with significantly less eLOS in PT compared to OT hospitalizations on univariable (22.6% vs 27.4%, p=.0005) and multivariable analysis (aOR=0.80 [95%CI, 0.69–0.94], p=.005). Black and other non-White races were associated with an increased risk of eLOS, as well as those with Medicaid or self-pay insurance. Greater frailty, pre-existing PAD, and hypercoagulable state were also associated with higher risk of eLOS (Figure 3d). The association between reduced eLOS and PT was moderated by race, with only white individuals having decreased eLOS with PT compared to black individuals (p_interaction=0.046; Supplementary Figure 1b).

Sensitivity Analysis

We performed five sensitivity analyses of our primary outcome, by altering the definition of ALI to include an additional ICD-CM-10 code, including concomitant CDT, reclassifying those who underwent both PT and OT as OT, limiting patients to those who underwent intervention within 1 day of hospitalization, and including additional covariates. The odds of amputation and death in the PT group remained similar and non-significant across sensitivity analyses (Table 3).

Table 3.

Sensitivity Analysis of Composite Amputation and Death

Cohort	total n	PT n (%)	Adjusted Odds Ratio*
Primary cohort	23,795	7,335 (30.8)	0.98 (0.79–1.12), p=.89
Sensitivity Analysis 1: including I998 ICD-CM-10** code	25,235	7,960 (31.5)	0.95 (0.77–1.17), p=64
Sensitivity Analysis 2: including concomittant lysis	28,310	10,860 (38.4)	0.91 (0.75–1.11), p=35
Sensitivity Analysis 3: concomittant PT and OT reclassified as OT	23,795	6,515 (27.4)	0.93 (0.74–1.18), p=0.57
Sensitivity Analysis 4: OR within 1 day of hospitalization	21,915	6,570 (30.0)	0.95 (0.76–1.20), p=0.68
Sensitivity Analysis 5: Additional covariates⧺ included	23,795	7,335 (30.8)	0.96 (0.78–1.20), p=0.75

Discussion

This present, large national study of 23,795 U.S. hospitalizations with a primary diagnosis of lower extremity ALI suggests that PT and OT are associated with equivalent in-hospital major amputation and mortality outcomes as measured by equivalence margins. However, our multivariable analysis of in-hospital outcomes suggests that these treatments are associated with differing secondary outcomes. PT was associated with an 110% increased relative risk of in-hospital major reintervention, although the majority of these reinterventions were endovascular. Additionally, this association was not present tibial interventions. We also demonstrated a 20% decreased relative risk of eLOS amongst PT hospitalizations. Our robust results were consistently observed across subgroups and in numerous sensitivity analyses.

Half of patients with ALI die or lose the affected limb within one year, reducing patient quality of life and functional independence¹. Yet, the evaluation of ALI treatments is extremely limited with only 48 registered trials evaluating ALI³⁶ on ClinicalTrials.gov in comparison to 334 evaluating chronic limb-threatening ischemia³⁷ and 10,352 evaluating coronary artery disease³⁸. The study of ALI is further impeded by the exclusion of thrombectomy procedures from important quality improvement datasets: OT is excluded from the VQI and PT is excluded from the NSQIP.

The prior data evaluating the association between ALI outcomes and open versus endovascular treatments are limited to three randomized controlled trials (STILE, TOPAS, and the Rochester trial), which compared open revascularization to CDT as opposed to PT^39–41. Interestingly, two of three trials (TOPAS and Rochester) included patients with motor deficits. Two of the three trials (STILE and Rochester) reported decreased 30-day composite amputation and death rate with CDT, attributed to higher numbers of cardiopulmonary complications in the open surgical group^39,40.

The evaluation of PT is limited to retrospective, single center studies or within larger studies comparing endovascular vs. open approaches. A retrospective analysis of endovascular (AngioJet^™ [Boston Scientific^®, MA] or CDT) vs. open surgical approaches showed similar major amputation risk but a higher 30-day mortality risk in the open revascularization group, attributed to the increased physiologic stress of open surgical interventions⁴². Another single-center, retrospective study comparing outcomes of Angiojet PT vs. CDT in patients with Rutherford I and II ALI showed PT had similar limb salvage outcomes with a mortality benefit, demonstrating a nearly 50% reduction in the adjusted risk of mortality¹².

Based on these studies, we hypothesized that limb salvage rates would be similar between PT and OT with decreased in-hospital mortality in the PT group. However, we uncovered similar rates of amputation and mortality for PT and OT across all a priori subgroups and sensitivity analyses. Our findings may differ for several reasons. The aforementioned studies included bypass with OT in their analyses, and including higher surgical stress of bypass may have elevated the mortality rate above that of the endovascular group. Furthermore, in certain cases, a surgical embolectomy can be expeditious and minimally stressful, while a lengthy endovascular intervention can prolong exposure to anesthesia, stressing the importance of appropriate patient selection. Additionally, while our cohort was over 200 times larger than existing studies, the available NIS data cannot differentiate between Rutherford classes of ALI, necessitating use of fasciotomies as a surrogate for severity of ischemia. Lastly, NIS data is limited to in-hospital outcomes, rather than 30-day, mid or long-term outcomes. Our overall in-hospital amputation (4.7%) and mortality (6.3%) rates are comparable to previously reported rates of in-hospital events (6–9% amputation and 9–12% mortality), with outcome data from 1998 to 2009¹. However, morbidity remains high after hospitalization, with reported 1-year amputation rates ranging from 11–18% and 1-year mortality ranging from 26–42%^1,43. Further study will be necessary to determine if the use of PT can decrease longer term mortality.

Several secondary outcomes and predictors of poor outcomes were consistent with existing literature. We demonstrated that PT had increased re-intervention but decreased eLOS, consistent with randomized controlled trials comparing open and endovascular treatments of chronic limb threatening ischemia^44,45. We found that pre-existing PAD had increased amputation, mortality, reintervention, and eLOS, consistent with a retrospective review that showed ALI patients with thrombosed bypasses had 4-times the risk of 5-year amputation than those with ALI secondary to embolic disease⁴³.

Counterintuitively, we found that younger patients were significantly more likely to undergo in-hospital amputation, consistent with a retrospective review comparing outcomes of patients over or above 50 years of age undergoing revascularization for ALI⁴⁶. We believe this is likely due to the lower likelihood of underlying vascular disease and less resultant collateralization in younger patients, who thus may present with more dense ischemia and higher likelihood of underlying hypercoagulability, leading to worse outcomes⁴⁷. One contradictory finding was that in our data, smoking was associated with a decreased risk of in-hospital death, as prior studies that show no association between ALI mortality and smoking^1,43. We hypothesize this may be due to healthy user effect (i.e. those who have diagnosis codes for smoking are in cessation programs) or that smokers may represent a different population who is more likely to present with ALI at a younger, healthier state compared to those with other comorbidities such as diabetes or renal failure.

Our analysis suggests that PT is a viable alternative to OT, meeting the requirements for outcome equivalence as quantified by the non-inferiority margin³³. Despite increased in-hospital reintervention rates for PT hospitalizations, almost 80% were able to avoid open salvage operations. Additionally, mortality and amputation rates were not affected, and eLOS frequency was decreased. This supports that this strategy may be attractive, particularly in poor surgical candidates who are frail, have contraindications to general anesthesia, or complex re-operative fields. While tibial disease appeared to be a strong predictor for amputation, we demonstrated no risk of reintervention with PT within the tibial subgroup. This may suggest that these patients had more complete revascularization, lessening the necessity of returning to the OR, tibial interventions with PT are more durable compared to OT, or these patients may have had a no reflow phenomenon such that reintervention was felt to be futile, leading to amputation. We also found that multilevel treatment was associated with increased mortality, likely due to both the greater ischemic burden and extent of the physiologically stressful operative intervention required. Therefore, the reduced risk of reintervention for this population may represent that many of these patients didn’t survive to undergo revascularization. ALI is extremely heterogeneous. Our work here, using administrative data to investigate outcomes in a national sample, must be complemented with a more granular dataset to further elucidate which patients may benefit most from this emerging technology. As our understanding of ALI expands to meet that of CTLI, we hope to develop evidence-based criteria to guide therapeutic interventions^48,49.

Our study has several limitations. NIS data provide a large sample representative of the U.S., encompassing many sociodemographic communities and practice patterns, and remains the one of the only large databases that captures all ALI cases. However, it relies on ICD-10 coding for diagnoses and procedures, which cannot capture all clinical characteristics (e.g. Rutherford classification, reason for reintervention/fasciotomy, laterality, and device data), introducing potential selection bias. To combat these limitations, we used validated ICD coding strategies^{1,3,23,26,28,29,32} and excluded those who underwent CDT, with an age <50, and had concurrent trauma/dissection, capturing a patient cohort that was appropriate for either PT or OT based on severity and etiology. Furthermore, because Rutherford classification is not available within these data (Supplementary Table 1), we used intraoperative fasciotomy, occurring in 5–7% of hospitalizations, as a surrogate for ischemic severity. Although the reported rate of fasciotomy use ranges broadly (5–50% ALI revascularization) in the literature, representing variable practice patterns, our data mirror those reported in the literature^46,50. Interestingly, Rutherford classification is also omitted from both the NSQIP and VQI, highlighting the omission of ALI specific data available in the literature. Additionally, outcomes may be misclassified as these data only include in-hospital events and do not account for planned reoperations or the competing risk of survival for those undergoing secondary procedures. Lastly, PT is a relatively recent treatment modality for ALI and the 2021 to 2023 NIS data were not yet released, omitting the most current uses and outcomes of PT from our analysis.

Conclusions

PT is an appropriate alternative to OT for the management of ALI with comparable in-hospital limb salvage and mortality rates. Despite the increased risk of in-hospital major reintervention associated with PT, most patients underwent endovascular reinterventions and the risk of eLOS remained lower than with OT. Further, our a priori subgroup analyses suggest that patients undergoing tibial intervention do not have an increased risk of reintervention with PT. However, more study working to identify which patients will benefit most from PT is necessary for appropriate, evidence-based application, particularly due to the higher rate of reintervention. Furthermore, the long-term outcomes of PT such as recurrent ALI and limb salvage could not be assessed and are important considerations that will need to be examined in future studies.

Article Highlights.

Type of Research:

Retrospective analysis of the prospectively collected National Inpatient Sample (NIS)

Key Findings:

In a cohort of 23,795 hospitalizations for acute limb ischemia undergoing thrombectomy, we found that 30.8% underwent percutaneous mechanical thrombectomy. Importantly, we found equivalent in-hospital limb-salvage and mortality between percutaneous and open thrombectomy (univariable 10.1% vs 10.9%, p=.43; multivariable aOR=0.96 [95%CI, 0.77–1.20], p=.74), which was consistent across subgroup analyses. However, despite having a higher likelihood of in-hospital reintervention (univariable 15.7% vs 8.1%, p<.0001; multivariable aOR=2.10 [95%CI, 1.72–2.56], p<.0001), 79.1% of the percutaneous group was able to avoid physiologically stressful open salvage operations. Furthermore, PT was not associated with increased reintervention for tibial interventions on subgroup analysis (aOR=1.31 [95% CI, 0.86–2.01], p=.21, p_interaction<.0001).

Take Home Message:

Percutaneous thrombectomy is an appropriate alternative to open thrombectomy given equivalent short-term limb salvage and mortality outcomes, but may potentially be most useful in patients with tibial disease or contraindications to open surgery.