The Contemporary Natural History of Minor Amputation Among Diabetic Patients with Peripheral Arterial Disease

Abstract

Objectives:

The growing prevalence of diabetes and concomitant PAD(DM/PAD) has led to an increase in patients at-risk for adverse limb events in current practice. Despite a widespread perception that minor amputation may result in both limb salvage and preserved functionality, the natural history of minor amputations remains unknown. Thus, we sought to quantify the rates of subsequent major amputation and survival among DM/PAD patients with any prior minor amputation.

Methods:

We performed a retrospective cohort study using US Medicare claims from 2007–2019. We included patients with DM/PAD based on ICD-9 and ICD-10 diagnosis codes. We excluded patients lacking continuous fee-for-service coverage or with incomplete demographic data. The primary exposure was prior minor(below-ankle) amputation. The primary outcome was major(above-ankle) amputation. Statistical analyses were performed using the Kaplan-Meier method and Cox proportional hazards modeling.

Results:

We identified 12,257,174 patients(age 73±11 years,48% male,76% White) with DM/PAD. Of these patients, 2.2%(N=274,225) underwent prior minor amputation. Patients with prior minor amputation were more likely to be male(63% versus 47%,p<0.0001), Black non-Hispanic(17% versus 13%,p<0.0001), and rural(25% versus 21%,p<0.0001) than those without prior minor amputation. The five-year Kaplan-Meier cumulative incidence of major amputation was 27%(N=58,613) of patients with prior minor amputation, compared to 1.4%(N=129,872) of patients without prior minor amputation. After risk-adjustment, patients with prior minor amputations were 6.1-fold more likely to require a subsequent major amputation(HR 6.11[6.04–6.18]) compared to those without prior minor amputations.

Conclusions:

This contemporary claims-based analysis demonstrates that approximately 25% of Medicare beneficiaries with DM/PAD and prior minor amputation will necessitate a major amputation within five years. Prior minor amputation carries a risk of major amputation comparable to de novo tissue loss, and is a stronger predictor than any demographic or socioeconomic exposure. These results help to inform both clinical decision-making and anticipated real-world outcomes among those at greatest risk for limb loss.

Table of Contents Summary

Nearly one-quarter of 274,225 Medicare beneficiaries with DM/PAD and a minor amputation succumbed to major limb amputation within five years in this retrospective claims-based study. A healed prior minor amputation represents a risk factor similar to tissue loss.

Introduction:

Peripheral arterial disease (PAD) currently affects approximately 12 million Americans with an increasing prevalence over the past decade.¹ The most severe form of PAD, chronic limb-threatening ischemia (CLTI), is associated with a one-year risk of major amputation as high as 22% in the setting of ischemic tissue loss.¹ Diabetes mellitus (DM) is a frequent comorbid condition among patients with PAD, with co-incidence rates of 20–30%, and amplifies the odds of amputation as much as 50-fold when compared with diabetics who do not have PAD.^1,2

The therapeutic gold standard for patients with concomitant DM/PAD is revascularization to optimize perfusion to the foot, in the context of a multidisciplinary approach including wound care, offloading, and atherosclerotic risk factor modification to achieve limb salvage. However, when limb salvage is not possible, amputation remains clinically necessary. Surgical strategies that preserve ambulatory function and mobility guide operative paradigms, focusing on distal, minor amputations when technically feasible.³ Although the rates of amputations in patients with DM/PAD are increasing, this has been largely attributed to minor amputations, under the notion that therapeutic advancements have led to more minor amputations and therefore enhanced limb salvage.⁴

Recent work published by our group focused on identifying the predominant demographic, preventive, and socioeconomic factors that modulate the risk of limb amputation in this population.⁵ However, the natural history of patients with minor amputations remains poorly characterized, and whether these potentially at risk patients ultimately succumb to major amputation in the contemporary era remains unknown. Published estimates of major amputation risk after an index minor amputation are highly variable, are limited to short follow-up intervals, and study differing patient populations.^6–8 Moreover, while future major amputation after initial minor amputation has been studied among patients with diabetes and CLTI, outcomes in these cohorts tend to be driven by the CLTI subgroup. Ascertaining the outcomes of patients with DM/PAD and a foot ulcer is an active research priority in the 2024 Intersocietal IWGDF/ESVS/SVS guidelines.⁹ Lastly, although recent population-based studies have been undertaken to fill these gaps,^7,10 data on these outcomes in the United States is scarce to date.

Therefore, the objective of this study was to determine the natural history of minor amputation in patients with concomitant DM/PAD in the United States, including quantification of the increased risk of major amputation among this population compared to patients without prior minor amputation. A secondary aim was to identify risk factors for progression to major amputation. We hypothesized that patients with DM/PAD with a prior minor amputation would be at elevated risk for progression to major amputation, with increasing age, Black non-Hispanic, Hispanic or Native American race, dual eligibility for Medicare and Medicaid, and medical comorbidities heightening this risk.

Methods:

Cohort creation: Medicare patients with concomitant DM/PAD

This project is part of a larger Strategically Focused Research Network (SFRN) effort led by the American Heart Association to study patients with concomitant diabetes and peripheral arterial disease.¹¹ The methodology used to create this cohort has been described in prior publications by our group.⁵ In brief, we performed a retrospective cohort study using Medicare claims data from 2007–2019. In order to generate our cohort of patients diagnosed with diabetes and peripheral arterial disease (DM/PAD), we used corresponding ICD-9 or ICD-10 codes in the Part B carrier, MedPAR, and outpatient claim files. The index claim date was defined as the patients’ first instance of an ICD-9 or ICD-10 code for diabetes within the study period, and all patients eligible for inclusion were required to have a PAD diagnosis code during the same calendar year. Exclusion criteria were: lacking continuous fee-for-service coverage for the year of their DM and PAD diagnoses, living outside of the United States, or having incomplete or missing demographic data (Supplemental Figure 1). A comprehensive list of ICD-9 and ICD-10 codes used is provided in Supplemental Table 1.

Primary exposure: minor amputation

Our primary exposure of interest was minor amputation, defined as any amputation below the ankle level (e.g., toe amputation or transmetatarsal amputation). We created two groups: 1) Patients with DM/PAD who underwent minor amputation within five years after DM/PAD diagnosis, and 2) patients with DM/PAD without a minor amputation within five years after DM/PAD diagnosis. Secondary exposures of interest were chosen based on factors identified in management strategies put forth by organizations including the American Heart Association, the American Diabetes Association, the Society for Vascular Surgery, and the American Podiatric Medical Association.^12–14 These included patient age (<65, 65–74, 75–85, and >85 years old), sex, race, rurality,¹⁵ patient comorbidities (Charlson Comorbidity Index),¹⁶ diabetic foot ulcers,¹⁷ and preventive measures (foot examination, vascular imaging, or hemoglobin A1c testing).¹⁸

Primary outcome: Major amputation

Our primary outcome was major amputation, defined as any above the ankle amputation (e.g., below- or above-knee amputation). Major amputations were identified using the relevant CPT codes (Supplemental Figure 1).¹⁹ Only outcomes that occurred after the index DM/PAD claim date were counted. We also examined the composite outcome of major amputation and death as well as death alone as secondary outcomes.

Identification of comorbidities and model covariates

The process of identifying and defining demographic, preventive, and socioeconomic variables from Centers for Medicare and Medicaid Services (CMS) data was performed similarly to our prior work.⁵ Part B carrier files and MedPAR claim files were linked to the Master Beneficiary Summary File (MBSF) in order to obtain demographic information, vital status, and Medicare-Medicaid dual eligibility status. Information on patient race was derived from standard CMS groups until 2008, at which time Research Triangle Institute (RTI) race groups were included in the CMS data.²⁰ These groups are defined by CMS as White non-Hispanic, Black non-Hispanic, Hispanic, Asian/Asian Pacific Islander (API), and American Indian/Alaskan Native. Although we used the variables made available by CMS to characterize race/ethnicity, we recognize their limitations in capturing structural racism in vascular disease.^21,22

Patient comorbidities were characterized according to the enhanced Quan et al. Charlson Comorbidity Index algorithm.²³ Income was measured at the patient-level using eligibility for both Medicare and Medicaid as reported previously,²⁴ and at the aggregate-level using zip-code level median household income.²⁵ Presence of a foot ulcer, which we defined as a composite of gangrene, non-pressure ulcers, and pressure ulcers, after DM/PAD claim date and before major amputation, death, or loss to follow-up was identified using ICD-9 and ICD-10 codes.¹⁷ Finally, we captured three preventive measures (foot examination, vascular imaging, or hemoglobin A1c testing) within the first 6 months after index claim date of DM/PAD as reported in prior work.¹⁸

Lastly, we identified all open and endovascular revascularization procedures performed after DM/PAD claim date and prior to major amputation (if occurred), death, or loss to follow-up using ICD-9, ICD-10, and CPT codes (Supplemental Table 1). We then performed a stratified analysis among patients with CPT codes for claudication, rest pain, and tissue loss to assess variation by PAD severity (Supplemental Table 1). These factors were included given their potential to confound the association between prior minor amputation and subsequent major amputation.

Statistical analysis

First, we performed Kaplan-Meier estimation for progression to any major amputation, for DM/PAD patients with a prior minor amputation versus without a prior minor amputation. In our primary analysis, we defined the start time for the estimator in the group with prior minor amputation as the index claim date for DM/PAD. We then performed a sensitivity analysis in which we defined the start time as the date of minor amputation for patients with a prior minor amputation. In both our primary and sensitivity analyses, the start time for the estimator for patients without a prior minor amputation was defined as the index claim date for DM/PAD.

For the analyses of major amputation, patients were censored at death, loss of fee-for-service coverage, or the end of data availability, whichever was earliest. For the analyses of death or the composite of major amputation and death, patients were censored at loss of fee for service coverage, or the end of data availability, whichever was earliest. Next, to study the relationship of our exposures of interest to the risk of amputation, we created a Cox proportional hazards regression model to estimate the hazard ratio (HR) of the outcomes between the groups. We included a total of 33 aforementioned covariates (prior forefoot amputation status, foot ulceration status, three preventive measure categories, three categories for age at time of DM diagnosis, five race categories, sex, rural-urban area classification, revascularization status, 14 Charlson comorbidity categories, dual-eligibility status, and two zip-code level median household income categories) in the model.

Continuous variables are reported as means with standard deviations or medians with interquartile ranges as appropriate. Categorical variables are reported as percentages. Two-sample t-tests were used to evaluate univariate associations for continuous outcomes, and the chi-square test was used to evaluate univariate associations for dichotomous outcomes. All statistical analyses were performed using SAS version 9.4 (SAS Institute., Cary, NC).

Subgroup analysis: Dialysis dependence

Lastly, we performed a subgroup analysis for patients in our cohort on dialysis, since prior studies have shown wide variation in lower limb amputation rates among this group.^26,27 Patients on dialysis were identified using ICD-9 and ICD-10 codes (Supplemental Table 1). Kaplan-Meier analyses out to five years were performed for the outcomes of major amputation, combined major amputation and death, and death. A dichotomous covariate for dialysis dependence was also added to the Cox model.

Protection of human subjects

This study received approval from our Institutional Review Board for the protection of human subjects. All data were accessed under the Centers for Medicare and Medicaid Services (CMS) data use agreement RSCH-2018-52223.

Results:

Diabetes and peripheral arterial disease cohort

The final cohort included 12,257,174 unique patients with concomitant diabetes and peripheral arterial disease (Table 1). Of this cohort, 274,225 (2.2%) patients had undergone a minor amputation. Among patients who had undergone a minor amputation, their mean age at the time of minor amputation was 69.0±12.0 years old. Patients who had undergone a minor amputation were more likely to be male (63% versus 48%, p<0.0001) and more likely to be Black non-Hispanic (17% versus 13%, p<0.0001), Hispanic (7.4% versus 6.6%, p<0.0001) or Native American race (1.35% versus 0.54%, p<0.0001). Patients who had undergone a minor amputation were more likely to have a history of congestive heart failure (29% versus 26%, p<0.0001), chronic kidney disease (42% versus 29%, p<0.0001), to be chronically dependent on dialysis (26% versus 13%, p<0.0001), to live in rural areas (25% versus 21%, p<0.0001), and to be dual eligible for Medicare and Medicaid (33% versus 28%, p<0.0001). Patients with a prior minor amputation were also more likely to have a lower zip-code level median household income ($24,927 versus $27,142, p<0.0001).

Table 1.

Demographics of Medicare patients with concomitant diabetes and peripheral arterial disease, with and without prior minor amputation (n=12,257,174).

Characteristic1	Overall	Prior Minor Amputation		p-value2
	(n=12,257,174)	Yes (n=274,225)	No (n=11,982,949)
Age (years)	73.2 (10.8)	67.4 (11.9)	73.3 (10.8)	<0.0001
Sex				<0.0001
Male	5,891,034 (48%)	173,027 (63%)	5,718,007 (48%)
Female	6,366,140 (52%)	101,198 (37%)	6,264,942 (52%)
Race				<0.0001
White	9,345,595 (76%)	196,890 (72%)	9,148,705 (76%)
Black non-Hispanic	1,567,329 (13%)	46,896 (17%)	1,520,433 (13%)
Asian/API	299,064 (2.4%)	2,796 (1.0%)	296,268 (2.5%)
Hispanic	809,824 (6.6%)	20,325 (7.4%)	789,499 (6.6%)
Native American	68,457 (0.56%)	3,712 (1.4%)	64,745 (0.54%)
Unknown/other	166,905 (1.4%)	3,606 (1.3%)	163,299 (1.4%)
Comorbidities				<0.0001
Myocardial Infarction	1,247,186 (10%)	31,193 (11%)	1,215,993 (10%)
Congestive Heart Failure	3,153,057 (26%)	78,200 (29%)	3,074,857 (26%)
Chronic Obstructive Pulmonary Disease	3,139,012 (26%)	54,476 (20%)	3,084,536 (26%)
Chronic Kidney Disease	3,643,348 (30%)	116,432 (42%)	3,526,916 (29%)
Dialysis Dependence	1,602,445 (13%)	71,921 (26%)	1,530,524 (13%)
Revascularization †
Open	126,637 (1.0%)	21,192 (7.7%)	105,445 (0.88%)	<0.0001
Endovascular	368,671 (3.0%)	53,669 (19.6%)	315,002 (2.6%)	<0.0001
Charlson Comorbidity Score	2.15 (2.29)	2.27 (2.17)	2.15 (2.29)	<0.0001
Rural Classification (RUCA)	2,611,966 (21%)	67,802 (25%)	2,544,164 (21%)	<0.0001
Dual Eligibility (Medicare and Medicaid)	3,410,623 (28%)	90,602 (33%)	3,320,021 (28%)	<0.0001
Zipcode-Level Median Household Income ($)	60,644 (27,100)	57,118 (24,927)	60,725 (27,142)	<0.0001

¹Numerical variables are presented as mean (SD), and categorical variables are presented as number of patients (percentage).

²The two-sample t-test (at a 0.05 significance level, two-tailed) is used to compare groups for numerical variables. The chi-square test is used to compare groups for categorical variables.

^‡History of open or endovascular revascularization after DM/PAD claim date.

Open revascularization was performed in 1.03% of all patients after DM/PAD claim date: 7.73% of patients with minor amputation had an open revascularization performed and 0.88% of patients without a minor amputation had an open revascularization procedure performed, respectively. Endovascular revascularization was performed in 3.01% of all patients after DM/PAD claim date: 19.6% of patients with minor amputation had an endovascular revascularization performed and 2.63% of patients without a minor amputation had an endovascular revascularization procedure performed, respectively.

Lastly, when all patients with DM/PAD (irrespective of minor amputation status) were stratified by PAD severity, 7.8% of patients with a CPT code for claudication, 8.1% of patients with a CPT code for rest pain, and 13.3% of patients with a CPT code for gangrene/ulceration underwent a revascularization procedure after DM/PAD claim date and before major amputation, death, or loss to follow-up. Among DM/PAD patients with chronic limb-threatening ischemia (i.e., a CPT code for rest pain or gangrene/ulceration) and a minor amputation, 31% underwent a revascularization procedure prior to major amputation, death, or loss to follow-up.

Among patients without prior minor amputation, the Kaplan-Meier estimated cumulative incidence of major amputation was 0.55% (95%CI 0.55%−0.55%) at one year and 1.37% (95%CI 1.36%−1.38%) at five years (Table 2). The cumulative incidence of major amputation and death as a composite outcome was 10.8% (95%CI 10.8%−10.8%) at one year and 41.0% (95%CI 40.9%−41.0%) at five years (Figure 1).

Table 2.

Kaplan-Meier cumulative incidence of major amputation and cumulative incidence of major amputation and death among patients with concomitant diabetes and peripheral arterial disease (DM/PAD) with prior versus without prior minor amputation.

Kaplan-Meier Cumulative Incidences	No Prior Minor Amputation (n=11,982,949)‡		Prior Minor Amputation (n=274,225) (Primary Analysis)‡		Prior Minor Amputation (n=274,225) (Sensitivity Analysis)‡
	Cumulative Incidence of Major Amputation (years)	Cumulative Incidence of Major Amputation and Death (years)	Cumulative Incidence of Major Amputation (years)	Cumulative Incidence of Major Amputation and Death (years)	Cumulative Incidence of Major Amputation (years)	Cumulative Incidence of Major Amputation and Death (years)
1 Year	0.55% (0.55–0.55)	10.8% (10.8–10.8)	7.76% (7.66–7.86)	12.8% (12.6–12.9)	16.6% (16.5–16.8)	31.6% (31.4–31.7)
3 Years	0.97% (0.97–0.97)	25.8% (25.8–25.9)	17.2% (17.0–17.3)	33.9% (33.7–34.0)	22.5% (22.3–22.7)	51.5% (51.3–51.7)
5 Years	1.37% (1.36–1.38)	41.0% (40.9–41.0)	27.0% (26.8–27.3)	57.8% (57.6–58.1)	27.5% (27.3–27.8)	68.2% (68.0–68.4)

^‡For patients with a prior minor amputation, t=0 is defined as index claim date of DM/PAD in the primary analysis and as the date of minor amputation in the sensitivity analysis. t=0 is the index claim date of DM/PAD in the group without prior minor amputation.

Figure 1.

Kaplan-Meier cumulative incidence of major amputation (A) and major amputation/death (B) among patients with concomitant diabetes and peripheral arterial disease (DM/PAD) with prior minor amputation (blue) versus patients without prior minor amputation (red). t=0 for both groups is defined as the index claim date of DM/PAD.

Unadjusted rates of major amputation among patients with prior minor amputation

Among patients with prior minor amputation, the Kaplan-Meier estimated cumulative incidence of major amputation was 7.76% (95%CI 7.66%−7.86%) at one year and 27.0% (95%CI 26.8%−27.3%) at five years (Table 2). The cumulative incidence of major amputation and death as a composite outcome was 12.8% (95%CI 12.6%−12.9%) at one year and 57.8% (95%CI 57.6%−58.1%) at five years (Figure 1).

Sensitivity analysis for patients with prior minor amputation

For our sensitivity analysis, we defined the start time for the Kaplan-Meier estimator as the date of prior minor amputation (rather than the index claim date of DM/PAD) among patients with a prior minor amputation. Among patients with prior minor amputation, the estimated cumulative incidence of major amputation was 16.6% (95%CI 16.5%−16.8%) at one year and 27.5% (95%CI 27.3%−27.8%) at five years (Table 2). The cumulative incidence of major amputation and death as a composite outcome was 31.6% (95%CI 31.4%−31.7%) at one year and 68.2% (95%CI 68.0%−68.4%) at five years (Figure 2).

Figure 2.

Kaplan-Meier cumulative incidence of major amputation (A) and major amputation/death (B) among patients with concomitant diabetes and peripheral arterial disease (DM/PAD) with prior minor amputation (blue) versus patients without prior minor amputation (red). t=0 for the prior minor amputation group is defined as the date of minor amputation. t=0 for the no prior minor amputation group is defined as the index claim date of DM/PAD.

Mortality rates among patients with and without prior minor amputation

Among patients without prior minor amputation, the Kaplan-Meier estimated cumulative incidence of death was 10.4% (95%CI 10.4%−10.4%) at one year and 40.6% (95%CI 40.5%−40.6%) at five years (Supplemental Table 2, Supplemental Figure 2). Among patients with prior minor amputation in our primary analysis, the estimated cumulative incidence of death was 6.61% (95%CI 6.52%−6.71%) at one year and 48.0% (95%CI 47.7%−48.2%) at five years. Among patients with prior minor amputation in our sensitivity analysis, the estimated cumulative incidence of death was 20.7% (95%CI 20.5%−20.9%) at one year and 60.9% (95%CI 60.7%−61.1%) at five years.

Multivariable analysis: Factors associated with major amputation

In our multivariable Cox model, patients with a prior minor amputation were 6.1-fold as likely to subsequently undergo a major amputation than patients without a prior minor amputation (aHR=6.11, 95%CI 6.04–6.18; Figure 3). Presence of gangrene, non-pressure ulcers, or pressure ulcers was associated with a 6.8-times as likely to undergo major amputation compared to patients with no ulceration (aHR=6.76, 95%CI 6.66–6.85). Although revascularization (either open or endovascular) after DM/PAD claim date was independently associated with the outcome of major amputation (aHR=1.15, 95%CI 1.12–1.18), addition of revascularization to the multivariable model had a minimal effect on the hazard ratio associated with prior minor amputation.

Figure 3.

Hazard ratio for progression to major amputation among patients with concomitant diabetes and peripheral arterial disease with a prior minor amputation (n=274,225). Confidence intervals are displayed but are so small that they are encompassed within each dot. HR, hazard ratio; CCI: Charlson comorbidity score; API: Asian/Pacific Islander.

^§Adjusted effect sizes are the result of a fully adjusted model including all above covariates.

^†Exposures are defined as foot imaging, vascular testing, or hemoglobin A1c testing within the first 6 months after index claim date of DM/PAD.

^‡History of open or endovascular revascularization after DM/PAD claim date.

*The foot ulcer covariate is a combination of gangrene, non-pressure ulcers, and pressure ulcers.

Patients who were diagnosed with diabetes at age 75–84 or age >85 years old were 12% (aHR=0.88, 95%CI 0.87–0.89) and 15% less likely (aHR=0.85, 95%CI 0.83–0.87) to undergo major amputation than those diagnosed between age 65–74, respectively. People of Native American (aHR=1.8, 95%CI 1.7–1.9) and Black non-Hispanic race (aHR=1.9, 95%CI 1.9–2.0) had a higher likelihood of major amputation than people of White race. Patients with a Charlson comorbidity index score ≥3 also had a statistically significant increase in the likelihood of major amputation (aHR=1.1, 95%CI 1.1–1.1). Male sex was associated with an increased likelihood of major amputation than female sex (aHR=1.6, 95%CI 1.6–1.6).

The number of preventive measures (foot examination, vascular imaging, and/or hemoglobin A1c testing) within the first six months of DM/PAD diagnosis was correlated with an increasing likelihood of major amputation, even after controlling for other risk factors. Patients who utilized one, two, or three of these laboratory tests or imaging modalities had a 19%, 53%, and 95% increased likelihood of major amputation, respectively, compared to patients not undergoing any preventive testing.

Patients eligible for both Medicare and Medicaid had a 40% higher likelihood of undergoing major amputation (aHR=1.4, 95%CI 1.4–1.4). Patients living in areas with high-level (greater than $60,000 annually) income were 16% less likely to undergo major amputation than those living in low income areas (less than $40,000 annually) (aHR=0.84, 95%CI 0.82–0.86).

Subgroup analysis: Dialysis dependence

Patients with DM/PAD on dialysis had higher Kaplan-Meier cumulative incidences of major amputation (Supplemental Figure 3; Supplemental Table 3), combined major amputation and death (Supplemental Figure 4; Supplemental Table 3), and death (Supplemental Figure 5; Supplemental Table 4) than their counterparts not on dialysis. For instance, the Kaplan-Meier estimated cumulative incidence of major amputation was 3.89% (95%CI, 3.84–3.93) at five years among dialysis-dependent DM/PAD patients without a prior minor amputation, compared to a 1.37% five-year cumulative incidence of major amputation among similar patients not on dialysis (Supplemental Table 3). Even considering the higher event rates among dialysis-dependent patients, patients on dialysis with prior minor amputation had significantly higher cumulative incidences of every outcome compared to patients on dialysis without prior minor amputation—regardless of how the analysis start time was defined. In particular, 41.1% (95%CI, 40.6–41.6%) of DM/PAD patients on dialysis with a prior minor amputation underwent major amputation with five years of DM/PAD diagnosis, whereas only 3.89% (95%CI, 3.84–3.93%) of dialysis patients without a prior minor amputation underwent major amputation over the same time period. Lastly, addition of dialysis dependence to the Cox model did not significantly impact the hazard ratio associated with prior minor amputation (Figure 3).

Discussion:

This is among the first studies to demonstrate that nearly a quarter of patients who undergo minor amputation will succumb to major limb loss within five years. Moreover, the likelihood of patients with a prior minor amputation progressing to major amputation is comparable to the likelihood of major amputation among patients with active ulceration or tissue loss, even after controlling for comorbidities, prior revascularization procedures, and sociodemographic factors. This advocates for the interpretation of a prior minor amputation as a longitudinal risk factor for major amputation and refutes any widely held perception that minor amputations provide durable limb salvage for at-risk patients with DM/PAD.

The present study expands upon recent work published by our group—which investigated the interaction of tissue loss type with amputation risk⁵—and characterizes the natural history of minor amputation among Medicare patients with DM/PAD. While recent studies such as Birmpili et al. and Gyldenkerne et al. have begun to study the association between minor and major amputation among patients with DM/PAD, similar cohorts examining outcomes in the United States are lacking.^7,10 It is important to note that the groups with and without prior minor amputation in our study differed significantly in their comorbidities, and the purpose of this study was not to compare these two very dissimilar groups. In our study, all-comers with DM/PAD without a prior minor amputation were at a low absolute risk of progressing to major amputation within five years of DM/PAD diagnosis. While patients with DM/PAD are known to be at a higher risk of limb loss than patients with either comorbidity alone,^1,28 it is unsurprising that the risk of amputation is low within five years of diagnosis as atherosclerotic occlusive disease and subsequent limb ischemia develop over years to decades. This baseline rate does, however, contextualize the observed “survival experience” of DM/PAD patients with prior minor amputation progressing to major amputation.

Our multivariable Cox model identified male sex, rurality, Black non-Hispanic or Native American race, lower socioeconomic status, revascularization procedures, and preventive testing as important independent predictors of major amputation. The increased risk of major amputation among men with DM/PAD has been extensively reported,^6,7,10,28 and is contrary to the existing data on increased CLTI-related morbidity among women.^29,30 As prior studies have also demonstrated an increased amputation risk among men with diabetes alone, it has been proposed that patients with DM/PAD have a different arterial disease pattern than those with CLTI alone.²⁸ The impact of socioeconomic status was confirmed both on the individual and aggregate level, as patients eligible for both Medicare and Medicaid and patients with less than $40,000 annual median household income had a higher likelihood of progression to major amputation. Our findings are consistent with prior studies reporting racial/ethnic and socioeconomic disparities in outcomes after lower limb amputation, which are likely driven by differential access to care, regional clustering, and bias at the provider and system levels.^19,25,31 Additionally, patients undergoing foot examination, vascular imaging, or hemoglobin A1c testing had an increased likelihood of major amputation in this cohort. This has been described in prior work,¹⁸ and may be because healthier patients are less likely to undergo monitoring and work-up than patients with multiple medical comorbidities. The aforementioned subgroups of patients are likely at differentially higher risk for the harms of minor amputation, which include increased costs, rehabilitation time, readmissions, and reinterventions.^32–34

The longitudinal rate of death was sobering among the patients in this cohort. Patients without versus with a prior minor amputation had a five-year Kaplan-Meier estimated mortality of 41% and 48%, respectively. This prompted our assessment of major amputation and death as a composite secondary outcome, to examine the impact of death as a competing risk. Despite the high rate of mortality in patients with DM/PAD, a significant number remained at-risk for major amputation over the study period, and a large fraction eventually progressed to major amputation. Patients with diabetes and PAD fear major amputation more than death as an outcome,^35,36 and a higher index of caution is likely warranted in those with minor amputations—even considering the high baseline risk of death among these patients—given the frequency of future major amputation within five years.

Importantly, we considered the impact of factors that could confound the association between prior minor amputation and subsequent major amputation. Firstly, the established impact of dialysis dependence as a risk factor for amputation^26,27 prompted our subgroup analysis of DM/PAD patients on dialysis, to ensure that our findings were not being disproportionately driven by this high-risk subgroup. Unsurprisingly, patients with DM/PAD on dialysis had higher cumulative incidences of major amputation, combined major amputation and death, and death than their counterparts not on dialysis. Nevertheless, the association of prior minor amputation with our primary and secondary outcomes was similar in magnitude and direction in our subgroup (dialysis) and primary (all-comers) analyses. Secondly, even though patients with prior minor amputation underwent open and endovascular revascularization procedures more frequently than patients without prior minor amputation, revascularization itself did not fully account for the increased hazard of major amputation. It appears that prior minor amputation, which serves as a marker of severe underlying atherosclerotic occlusive disease, portends a substantial risk for future limb events irrespective of dialysis-dependent status or prior revascularization status.

Our finding that only 31% of DM/PAD patients with a CPT code for chronic limb-threatening ischemia and a prior minor amputation underwent revascularization prior to major amputation, death, or loss to follow-up is striking but consistent with prior work. In a statewide analysis from 2005–2013 among patients with DM/PAD, a lower extremity ulcer and prior minor amputation, 64% of these patients did not have a revascularization attempt prior to subsequent major amputation being performed.⁶ We also found that between 7.8% to 13.3% of all patients with DM/PAD underwent a revascularization procedure after DM/PAD claim date and before major amputation, death, or loss to follow-up depending on PAD severity. This is consistent with rates observed in a recently published study using the Nationwide Inpatient Sample between 2003–2017, where 8,654 to 12,627 revascularization procedures per 100,000 hospitalizations per year were observed.⁴ The two-thirds of DM/PAD patients with CLTI and prior minor amputation that do not undergo revascularization prior to major amputation has been described,⁶ and our work suggests that this continues to be an opportunity for system-level improvement.

There are several limitations to this work. Firstly, claims-based analyses are unable to identify the laterality of amputation. While this limitation could be addressed by future studies using longitudinal clinical databases, this work is still able to characterize subsequent amputation procedures and identify patient groups at risk for future major amputation after an initial minor amputation. Secondly, the severity of diabetes is unknown in this dataset, and it is likely that patients who are diet controlled, on oral medications or insulin-dependent have different risks of amputation. This could be improved upon by considering the impact of HbA1c on amputation risk, and we lacked sufficient granularity of the actual HbA1c values per patient to perform this analysis in this study. Thirdly, we lacked granularity on the indication for prior minor amputation as well as information on vascular anatomy of the limb, ambulatory status, and patient frailty. While we are limited in our ability to characterize the appropriateness of index and subsequent amputation, revascularization procedures etc., this work is able to establish minor amputation as a powerful marker of future major amputation. Lastly, undergoing a minor amputation is a time dependent exposure, which may have introduced bias into our estimations of the outcome. To address this, we modeled minor amputation in the Kaplan-Meier estimators both with the starting time being the diagnosis of DM/PAD, but also with the starting time being the claim for the minor amputation. Both models showed similar findings.

Conclusion:

Approximately 25% of patients with DM/PAD and a prior minor amputation will require a major amputation within five years. It appears that a prior minor amputation is not a panacea against limb loss over time, but rather represents a risk factor similar to tissue loss. Additionally, a prior minor amputation remains a stronger predictor for major amputation than any demographic or socioeconomic exposure. These findings should help to calibrate clinical expectations for both patients and physicians, to help inform evidence-based care delivery. Lastly, policymakers should anticipate that DM/PAD patients with prior minor amputation represent an at-risk healthcare resource intensive population, with implications for hospitals and healthcare systems alike.

ARTICLE HIGHLIGHTS.

Type of Research:

Retrospective cohort study of Medicare claims data.

Key Findings:

Among 274,225 Medicare beneficiaries with concomitant diabetes and peripheral arterial disease and a prior minor amputation, 27% progressed to a major amputation within five years. Compared to 11,982,949 beneficiaries without prior minor amputation, patients with prior minor amputations were six-fold more likely to require a subsequent major amputation after risk-adjustment.

Take home Message:

Healed minor amputations remain a significant longitudinal risk factor for major amputation and carry approximately a one-in-four risk of major amputation over a five year period.