Abstract
Introduction:
Amputations among younger patients with chronic limb threatening ischemia (CLTI) may carry higher personal and societal costs, but younger patients are often not included in CLTI research because of dataset limitations. We aimed to characterize and compare outcomes between younger (<65 years old) and older patients with CLTI.
Methods:
This retrospective cohort study identified patients with CLTI between July 1, 2014 and December 31, 2017 in the MarketScan commercial claims database, a proprietary set of claims for over 50 million patients with private insurance in the United States. The primary outcome was major adverse limb events (MALE); secondary outcomes included amputations, major adverse cardiovascular events, and statin prescription fills.
Results:
The study cohort included 64,663 people with CLTI, of whom 25,595 (39.6%) were <65 years old. Younger patients were more likely to have diabetes mellitus (54.1% versus 49.9%, p<.001) but less likely to have other comorbidities. A higher proportion of younger patients suffered MALE (31.7% versus 30.2%, p=.002), specifically amputation (11.5% versus 9.3%, p<.001). After adjustment, age <65 years old was associated with a 24% increased risk of amputation (HRadj 1.24, 95%CI 1.18–1.32, p<.001) and a 10% increased risk of MALE (HRadj 1.10, 95%CI 1.07–1.14, p<.001).
Conclusions:
A significant proportion of commercially insured patients with CLTI are under the age of 65, and younger patients have worse limb-related outcomes. These findings highlight the importance of aggressively treating risk factors for atherosclerosis and intentionally including younger patients with CLTI in future analyses to better understand their disease patterns and outcomes.
Keywords: outcomes research, chronic limb threatening ischemia, critical limb ischemia, peripheral artery disease
Introduction:
The incidence and prevalence of peripheral artery disease (PAD) increase with older age, and it is estimated to affect 3.41% of people aged 50–59 versus 7.77% of people aged 60–69 in the United States.(1) While important risk factors for PAD such as smoking history, chronic kidney disease (CKD), and diabetes mellitus (DM) are more common among older patients, their prevalence continues to rise among younger patients.(2,3) This suggests that PAD may begin to affect younger patients in larger numbers over the coming years. Furthermore, research among patients with DM and CKD has shown that younger patients now suffer amputations at a higher rate than do older patients: while the overall yearly amputation rate among patients with DM and CKD has decreased over time, this decline has disproportionately benefitted older patients while amputation rates in younger patients have stagnated or increased.(4,5)
Age-based differences in patient presentations, treatment patterns, and outcomes have not been adequately explored on a national scale among patients with PAD. Characterizing these differences, if any, is the first step towards improving the care and outcomes of younger patients. Unfortunately, most large PAD analyses rely on Centers for Medicare and Medicaid Services (CMS) data, which excludes most patients under the age of 65, on procedural datasets that yield cohorts of post-revascularization patients, or on inpatient databases without mid- to long-term follow-up.(6–8) A recent Vascular Quality Initiative (VQI) analysis reported that the average age of patients undergoing amputation in the database was 66 years old, indicating that a large number of vascular amputees are under the age of Medicare eligibility.(9) We aimed to use the MarketScan database, which includes commercially-insured patients (including fee-for-service Medicare) of all ages from across the United States with long-term follow-up, to characterize and compare patient factors, treatment, and outcomes between patients younger than 65 with those aged ≥ 65. We hypothesized that patients younger than 65 would have different baseline characteristics and worse limb-related outcomes.
Methods:
Data source
We used the MarketScan commercial claims database (IBM Research, Yorktown Heights, NY), which is a proprietary dataset consisting of inpatient admissions and services claims, outpatient services and pharmaceutical claims, facility headers, and enrollment details for over 50 million patients with private insurance (including Medicare fee-for-service coverage) across the United States. Patients with certain participating employer-sponsored and private insurance plans, as well as other household members on the same insurance plans, are included in the dataset. We used claims from July 1, 2014 through December 31, 2017 for cohort identification and July 1, 2013 through December 31, 2016 for comorbidity ascertainment.
Patient population
Individuals were included if they had International Classification of Diseases (ICD) 9 or 10 Clinical Modification (CM) diagnosis codes for CLTI, with ICD-9-CM codes used from July 2014 through September 2015 and ICD-10-CM codes used from October 2015 through December 2017. ICD-9-CM codes included 440.22 (atherosclerosis with rest pain), 440.23 (atherosclerosis with ulceration), and 440.24 (atherosclerosis with gangrene); ICD-10-CM codes included i70.(2–7)2 (atherosclerosis with rest pain); i70.(2–7)(3–5) (atherosclerosis with ulceration); and i70.(2–7)6 (atherosclerosis with gangrene). Though diagnosis codes perform poorly to identify PAD without CLTI,(10) codes for PAD with CLTI have been shown to have a much higher positive predictive value.(11) For the purposes of disease severity classification (rest pain, ulceration, or gangrene), patients were classified based off the ICD-9 or −10 code at the time of cohort inclusion. Patients with <1 year of enrollment prior to the date of the qualifying claim were excluded in order to facilitate comorbidity ascertainment. Claims without an enrollment ID were excluded in order to facilitate outcome ascertainment. Patients under the age of 40 were excluded because the rarity of CLTI in patients under the age of 40 raises the relative likelihood of a coding error.(12)
Outcomes and covariates
The index encounter was the first encounter with a CLTI code in any position during the study period. The primary outcome was a composite of major adverse limb events (MALE) consisting of acute limb ischemia (ALI), revascularization, or major or minor amputation. Secondary outcomes included modified major adverse cardiovascular events (MACE, consisting of myocardial infarction (MI) and stroke); the presence of statin prescription fills in the year following index encounter for patients whose insurance plans included prescription coverage; and the individual components of the MALE composite. Diagnosis and procedure codes used to ascertain these outcomes are available in Supplemental Table I. Statins included in the prescription fill outcome are available in Supplemental Table II. Post-index fills of statin prescriptions was chosen as an endpoint rather than fills of prescriptions for antiplatelet or anticoagulant medications because statins are unequivocally recommended for all patients with CLTI while anticoagulant and prescription antiplatelet medications (versus aspirin, whose use would not be captured using claims) are recommended for only certain subsets of CLTI patients.(13–15)
Other variables of interest included age, sex, and US Census Division of residence. Patient race and ethnicity are not available in MarketScan. Comorbidities included cerebrovascular disease (CBVD), ischemic heart disease, congestive heart failure (CHF), hypertension (HTN), diabetes mellitus (DM), renal disease, dementia, and cancer; these were ascertained using previously-validated coding algorithms.(16,17) The presence or absence of fills of prescriptions for statins, antiplatelet, and anticoagulant medications over the year prior to the index encounter was also collected from among patients whose insurance plans included prescription coverage. Included medications are listed in Supplemental Table II. Patients whose health plans included prescription coverage (89.6% of the cohort) are described in Supplemental Table III.
Statistical analysis
The cohort’s baseline characteristics were described overall and divided into age groups of 40–64 years old and ≥ 65 years old. Age was dichotomized in order to more easily convey differences between patients who would be captured in CMS analyses and those who would be less likely to be included in CMS analyses. In addition, the 40–64 year old group was not subdivided further because doing so would have been arbitrary and resulted in multiple smaller groups, weakening our statistical analysis. Differences in baseline characteristics between the groups were tested using Pearson’s chi-square tests for categorical variables and the Wilcoxon rank-sum test for continuous variables.
The observed cumulative incidence of each primary and secondary outcome at one year was described using Kaplan-Meier curves. Mortality was not used as a competing risk for repair because mortality was not available in the MarketScan database during our study period, but patients were censored when their enrollment ID was removed from the database (which may indicate death, loss of health insurance, or discontinuation of an insurance’s participation in the MarketScan database). Separate unadjusted univariable and adjusted multivariable Cox proportional hazard regression models were used to examine the associations between patient characteristics and the primary outcome, MALE, as well as between patient characteristics and the secondary outcomes.
Candidate variables for the Cox proportional hazards models included age group, sex, CLTI severity (rest pain, ulceration, or gangrene), and comorbidities. US Census Division of residence was not included as a candidate variable because the representativeness of MarketScan enrollment across Census Divisions is undefined. Fills of prescriptions for antiplatelet or anticoagulant medications were not included as candidate variables because there are multiple reasons a patient might or might not be prescribed those medications that each interact differently with other comorbidities and medical history. Statistics were carried out using SAS (version 9.4, Cary, NC) by CBF. As this was an analysis of the MarketScan database, patients did not consent specifically for this study, which was determined exempt by our institutional review board (Pro00103946). No extramural funding was used to support this work, though some authors receive extramural funding for other work.
Results:
Baseline characteristics
There were 64,663 people in the MarketScan database with CLTI-coded encounters during the study period. Of these, 25,595 patients were aged 40–64 years old (39.6%) while 39,068 were ≥ 65 years old (60.4%). Younger patients were more likely to be male (56.8% versus 51.9%, p<.001, Table I). The distribution of CLTI severity differed significantly between groups (p<.001): younger patients were more likely to have rest pain (59.4% versus 52.5%) and less likely to have ulcerations (31.8% versus 39.9%). Gangrene was present in 8.9% of younger and 7.6% of older patients.
Table I:
Baseline characteristics
| Variable | Overall, N=64,663 N (%) | 40–64 years old, N=25,595 N (%) | ≥65 years old, N=39,068 N (%) | p-value |
|---|---|---|---|---|
| CLTI severity | < .001 | |||
| Rest pain | 35,720 (55.2%) | 15,199 (59.4%) | 20,521 (52.5%) | |
| Ulceration | 23,706 (36.7%) | 8,130 (31.8%) | 15,576 (39.9%) | |
| Gangrene | 5,237 (8.1%) | 2,266 (8.9%) | 2,971 (7.6%) | |
| Demographics | ||||
| Age (years), Mean (SD) | 69.65 (12.94) | 56.53 (5.94) | 78.25 (8.21) | < .001 |
| Male sex | 34,804 (53.8%) | 14,545 (56.8%) | 20,259 (51.9%) | < .001 |
| Census Division | < .001 | |||
| Missing | 4,151 (.) | 1,327 (.) | 2,824 (.) | |
| New England | 1,628 (2.7%) | 617 (2.5%) | 1,011 (2.8%) | |
| Mid-Atlantic | 15,666 (25.9%) | 6,348 (26.2%) | 9,318 (25.7%) | |
| East North Central | 15,065 (24.9%) | 4,341 (17.9%) | 10,724 (29.6%) | |
| West North Central | 1,575 (2.6%) | 596 (2.5%) | 979 (2.7%) | |
| South Atlantic | 12,144 (20.1%) | 5,765 (23.8%) | 6,379 (17.6%) | |
| East South Central | 3,254 (5.4%) | 1,843 (7.6%) | 1,411 (3.9%) | |
| West South Central | 5,852 (9.7%) | 2,729 (11.2%) | 3,123 (8.6%) | |
| Mountain | 1,733 (2.9%) | 723 (3.0%) | 1,010 (2.8%) | |
| Pacific | 3,595 (5.9%) | 1,306 (5.4%) | 2,289 (6.3%) | |
| Comorbidities | ||||
| CBVD | 18,815 (29.1%) | 5,281 (20.6%) | 13,534 (34.6%) | < .001 |
| Ischemic heart disease | 28,581 (44.2%) | 8,585 (33.5%) | 19,996 (51.2%) | < .001 |
| CHF | 15,926 (24.6%) | 3,814 (14.9%) | 12,112 (31.0%) | < .001 |
| Hypertension | 53,901 (83.4%) | 19,553 (76.4%) | 34,348 (87.9%) | < .001 |
| DM | 33,331 (51.5%) | 13,843 (54.1%) | 19,488 (49.9%) | < .001 |
| Renal disease | 16,432 (25.4%) | 4,733 (18.5%) | 11,699 (29.9%) | < .001 |
| Dementia | 4,366 (6.8%) | 160 (0.6%) | 4,206 (10.8%) | < .001 |
| Cancer | 8,336 (12.9%) | 1,876 (7.3%) | 6,460 (16.5%) | < .001 |
| Prior MI | 2,369 (9.3%) | 2,369 (9.3%) | 4,953 (12.7%) | < .001 |
| Medication History | ||||
| RX Eligible | 57,937 (89.6%) | 22,561 (88.1%) | 35,376 (90.5%) | < .001 |
| Statins | 32,699 (56.4%) | 11,554 (51.2%) | 21,145 (59.8%) | < .001 |
| Antiplatelet medications | 13,255 (22.9%) | 4,937 (21.9%) | 8,318 (23.5%) | < .001 |
| Anticoagulant medications | 9,901 (17.1%) | 2,380 (10.5%) | 7,521 (21.3%) | < .001 |
RX: prescription
Younger patients were more likely to have DM (54.1% versus 49.9%, p<.001) but were less likely to have other comorbidities. Younger patients were slightly less likely to have prescription coverage (88.1% versus 90.5%, p<.001) when compared with older patients. Among patients with prescription coverage, fewer younger patients had filled a prescription for statins (51.2% versus 59.8%, p <.001), antiplatelet medications (21.9% versus 23.5%, p <.001), or anticoagulant medications (10.5% versus 21.3%, p <.001) in the year prior to index CLTI encounter compared to older patients.
Outcomes at one year following index encounter
Younger patients more commonly experienced MALE by one year (31.7% versus 30.2%, p=.002, Table II and Figure 1). Among the individual components of MALE, amputation was significantly more common in younger patients (11.5% versus 9.3%, p<.001) as was ALI (1.5% versus 1.2%, p<.001). There was no significant difference in major amputations between younger and older patients (4.7% versus 4.3%, respectively, p=.08) but younger patients were more likely to undergo minor amputation (8.1% versus 6.1%, p<.001). There was no difference in proportion of patients undergoing revascularization by one year (25.5% versus 25.7%, p=.42). Younger patients were significantly less likely to suffer MACE by one year (2.5% versus 3.3%, p<.001). Younger patients were also significantly less likely to fill a prescription for a statin over the year following index encounter (45.9% versus 50.7%, p<.001).
Table II:
Unadjusted event rates at one year
| Variable | 40–64 years old, N=25,595 N (%) | ≥65 years old, N=39,068 N (%) | p-value |
|---|---|---|---|
| Modified MACE | 501 (2.5%) | 979 (3.3%) | < .001 |
| MALE | 7379 (31.7%) | 10704 (30.2%) | .002 |
| Revascularization | 5976 (25.5%) | 9156 (25.7%) | .42 |
| Amputation | 2550 (11.5%) | 3147 (9.3%) | < .001 |
| Major amputation | 1024 (4.7%) | 1440 (4.3%) | .08 |
| Minor amputation | 1800 (8.1%) | 2049 (6.1%) | < .001 |
| Acute Limb Ischemia | 357 (1.5%) | 409 (1.2%) | < .001 |
| Statins, post-diagnosisa | 9355 (45.9%) | 16084 (50.7%) | < .001 |
Among prescription-eligible patients: 22,561 patients aged 40–64 years old and 35,376 patients aged ≥ 65 years old
Figure 1: Event rates in younger and older patients at one year.
All depicted differences in endpoints are statistically significant (MALE p = .002; amputation p<.001; MACE p<.001; statin prescription p<.001
Factors associated with MALE
On unadjusted analysis, younger age was associated with a 5% higher risk of MALE (unadjusted HR= 1.05 95% CI 1.02–1.08, p=.002, Table III), and following adjustment, age remained significantly associated with a 10% higher risk of MALE (adjusted HR= 1.10 95% CI 1.07–1.14, p<.001). Gangrene, ulceration, male sex, CBVD, ischemic heart disease, HTN, DM, renal disease, and prior MI were also all significantly associated with higher risk of MALE following adjustment. CHF, dementia, and cancer were each associated with a significantly lower likelihood of MALE following adjustment.
Table III:
Unadjusted and adjusted associations between patient characteristics and MALE and amputation at one year
| MALE | Amputation | |||||||
|---|---|---|---|---|---|---|---|---|
| Parameter | Unadjusted HR (95% CI) | p | Adjusted HR (95% CI) | p | Unadjusted HR (95% CI) | p | Adjusted HR (95% CI) | p |
| Age Group (Ref ≥ 65) | ||||||||
| 40–64 | 1.05 (1.02, 1.08) | .002 | 1.10 (1.07, 1.14) | < .001 | 1.24 (1.18, 1.32) | < .001 | 1.24 (1.18, 1.32) | < .001 |
| Disease severity (Ref = Rest Pain) | ||||||||
| Ulceration | 1.61 (1.56, 1.66) | < .001 | 1.50 (1.45, 1.55) | < .001 | 4.39 (4.07, 4.74) | < .001 | 4.39 (4.07, 4.74) | < .001 |
| Gangrene | 2.83 (2.71, 2.96) | < .001 | 2.41 (2.30, 2.52) | < .001 | 13.68 (12.57, 14.89) | < .001 | 13.68 (12.57, 14.89) | < .001 |
| Demographics | ||||||||
| Male sex | 1.56 (1.51, 1.60) | < .001 | 1.33 (1.29, 1.37) | < .001 | 1.55 (1.46, 1.64) | < .001 | 1.55 (1.46, 1.64) | < .001 |
| Comorbidities | ||||||||
| CBVD | 1.33 (1.29, 1.37) | < .001 | 1.22 (1.18, 1.26) | < .001 | 1.01 (0.95, 1.07) | .73 | 1.01 (0.95, 1.07) | .73 |
| IHD | 1.63 (1.58, 1.68) | < .001 | 1.33 (1.29, 1.38) | < .001 | 1.07 (1.01, 1.14) | .03 | 1.07 (1.01, 1.14) | .03 |
| CHF | 1.35 (1.31, 1.39) | < .001 | 0.93 (0.90, 0.96) | < .001 | 1.16 (1.09, 1.24) | < .001 | 1.16 (1.09, 1.24) | < .001 |
| HTN | 1.58 (1.51, 1.65) | < .001 | 1.21 (1.16, 1.27) | < .001 | 1.16 (1.06, 1.28) | .002 | 1.16 (1.06, 1.28) | .002 |
| DM | 1.54 (1.50, 1.59) | < .001 | 1.19 (1.15, 1.23) | < .001 | 2.01 (1.88, 2.14) | < .001 | 2.01 (1.88, 2.14) | < .001 |
| Renal disease | 1.57 (1.52, 1.62) | < .001 | 1.18 (1.14, 1.22) | < .001 | 1.43 (1.35, 1.52) | < .001 | 1.43 (1.35, 1.52) | < .001 |
| Dementia | 0.76 (0.71, 0.81) | < .001 | 0.67 (0.63, 0.72) | < .001 | 0.83 (0.74, 0.93) | .001 | 0.83 (0.74, 0.93) | .001 |
| Cancer | 0.94 (0.90, 0.98) | .006 | 0.89 (0.85, 0.93) | < .001 | 0.85 (0.79, 0.93) | < .001 | 0.85 (0.79, 0.93) | < .001 |
| Prior MI | 1.64 (1.58, 1.71) | < .001 | 1.16 (1.11, 1.21) | < .001 | 1.14 (1.05, 1.23) | .001 | 1.14 (1.05, 1.23) | .001 |
Factors associated with amputation
Younger age was associated with a 24% higher unadjusted risk of amputation (unadjusted HR= 1.24 95% CI 1.18–1.32, p<.001, Table III), which remained unchanged after adjustment (adjusted HR= 1.24 95% CI 1.18–1.32, p<.001). Following adjustment, gangrene, ulceration, male sex, ischemic heart disease, HTN, DM, renal disease, and prior MI were also all associated with a significantly higher risk of amputation while dementia and cancer were associated with significantly lower adjusted risk of amputation.
Factors associated with MACE
Younger age was significantly associated with 23% lower risk of MACE on unadjusted analysis (unadjusted HR= 0.77 95% CI 0.69–0.85, p<.001, Table IV) but this did not remain true after adjustment (adjusted HR= 0.93 95% CI 0.83–1.04, p=.23). After adjustment, the presence of ulceration and all comorbidities except CHF, dementia, and cancer were significantly associated with higher risk of MACE (dementia was associated with a significantly lower risk of MACE).
Table IV:
Unadjusted and adjusted associations between patient characteristics and MACE at one year
| Parameter | Unadjusted HR (95% CI) | p | Adjusted HR (95% CI) | p |
|---|---|---|---|---|
| Age Group (Ref ≥ 65) | ||||
| 40–64 | 0.77 (0.69, 0.85) | < .001 | 0.93 (0.83, 1.04) | .23 |
| Diagnosis (Ref = Rest Pain) | ||||
| Ulceration | 1.35 (1.21, 1.50) | < .001 | 1.17 (1.05, 1.32) | .006 |
| Gangrene | 1.42 (1.18, 1.71) | < .001 | 1.12 (0.92, 1.35) | .25 |
| Demographics | ||||
| Male sex | 1.27 (1.15, 1.41) | < .001 | 1.09 (0.98, 1.21) | .12 |
| Comorbidities | ||||
| CBVD | 2.15 (1.94, 2.38) | < .001 | 1.78 (1.60, 1.98) | < .001 |
| IHD | 2.10 (1.89, 2.33) | < .001 | 1.35 (1.19, 1.53) | < .001 |
| CHF | 1.79 (1.61, 1.99) | < .001 | 1.09 (0.96, 1.23) | .20 |
| HTN | 2.37 (1.97, 2.86) | < .001 | 1.53 (1.25, 1.86) | < .001 |
| DM | 1.65 (1.49, 1.84) | < .001 | 1.29 (1.16, 1.44) | < .001 |
| Renal disease | 1.81 (1.63, 2.01) | < .001 | 1.24 (1.10, 1.40) | < .001 |
| Dementia | 1.06 (0.85, 1.31) | .60 | 0.80 (0.64, 1.00) | .049 |
| Cancer | 1.07 (0.92, 1.25) | .37 | 0.95 (0.82, 1.11) | .53 |
| Prior MI | 2.36 (2.08, 2.67) | < .001 | 1.52 (1.32, 1.74) | < .001 |
Factors associated with filling a statin prescription in the year following index encounter
Younger age was significantly associated with a 13% lower likelihood of a post-index statin fill (unadjusted HR= 0.87, 95% CI 0.84–0.89, p<.001, Table V), which was unchanged following adjustment (adjusted HR= 0.86, 95% CI 0.84–0.88, p<.001). Following adjustment, gangrene, ulceration, male sex, and all other comorbidities except CHF and prior MI were associated with a significantly higher adjusted likelihood of a post-index statin fill (CHF was associated with a significantly lower likelihood of a post-index statin fill).
Table V:
Unadjusted and adjusted hazard ratios for the associations between patient characteristics and filling a statin prescription at one year
| Parameter | Unadjusted HR (95% CI) | p | Adjusted HR (95% CI) | p |
|---|---|---|---|---|
| Age Group (Ref Ref ≥ 65) | ||||
| 40–64 | 0.87 (0.84, 0.89) | < .001 | 0.86 (0.84, 0.88) | < .001 |
| Diagnosis (Ref = Rest Pain) | ||||
| Ulceration | 0.95 (0.93, 0.98) | < .001 | 0.91 (0.88, 0.93) | < .001 |
| Gangrene | 0.91 (0.86, 0.95) | < .001 | 0.83 (0.79, 0.88) | < .001 |
| Demographics | ||||
| Male Sex | 1.19 (1.16, 1.22) | < .001 | 1.13 (1.10, 1.16) | < .001 |
| Comorbidities | ||||
| CBVD | 1.15 (1.12, 1.19) | < .001 | 1.07 (1.04, 1.10) | < .001 |
| Ischemic heart disease | 1.30 (1.27, 1.33) | < .001 | 1.19 (1.15, 1.22) | < .001 |
| CHF | 0.96 (0.93, 0.99) | .008 | 0.81 (0.79, 0.84) | < .001 |
| HTN | 1.61 (1.55, 1.67) | < .001 | 1.41 (1.36, 1.47) | < .001 |
| DM | 1.43 (1.40, 1.47) | < .001 | 1.37 (1.34, 1.41) | < .001 |
| Renal disease | 1.07 (1.04, 1.10) | < .001 | 0.95 (0.92, 0.98) | .00 |
| Dementia | 0.81 (0.77, 0.86) | < .001 | 0.79 (0.74, 0.83) | < .001 |
| Cancer | 1.01 (0.97, 1.05) | .64 | 0.96 (0.93, 1.00) | .03 |
| Prior MI | 1.16 (1.12, 1.20) | < .001 | 1.02 (0.98, 1.06) | .41 |
Discussion:
Our findings suggest that people <65 years old are a significant minority and distinct subgroup of CLTI patients who have an increased risk of adverse limb-related outcomes and lower likelihood of receiving guideline-directed medical therapy. Nearly 40% of the patients with CLTI in this MarketScan analysis cohort were under the age of 65. Differences in their baseline characteristics, treatment patterns, and outcomes highlight the importance of including patients <65 years old in CLTI analyses and of vigorously treating this aggressive manifestation of atherosclerosis. Younger CLTI patients were less likely to have filled a statin prescription in the year prior to index encounter and remained less likely to fill a statin prescription in the year following index encounter. Finally, younger patients were at increased risk for amputations even after controlling for disease severity and comorbidities. These findings have significant ramifications for future CLTI research and care-improvement initiatives.
Most research about CLTI in large datasets either uses CMS data, in which few patients <65 years old are represented, or uses cohorts of patients undergoing certain interventions.(6,18–20) While the National Inpatient Sample (NIS) and National Readmissions Database (NRD) both include patients <65 years old and patients not undergoing intervention, neither has the ability to follow patients over time for mid- to long-term outcomes.(7,8) We chose to use MarketScan because it yields a geographically-diverse patient cohort without age or procedural limitations while also facilitating outcome ascertainment over time. Armstrong et. al. previously used the MarketScan dataset to examine risk factors for amputation, showing that patients who were treated conservatively tended to be older and those treated with minor amputation or surgical bypass tended to be younger.(21) However, as the average age of all treatment groups was ≥ 66 years, the analysis shed no light specifically on the presentation, treatment, and outcome patterns of patients under the age of 65. Our investigation is focused on younger CLTI patients as a distinct group and adds to this prior analysis by showing that younger CLTI patients are different from older CLTI patients both at presentation and over follow-up, suggesting that future CLTI analyses should not rely on CMS, NIS, NRD, or VQI data alone.
Prior research about the amputation risk among younger versus older people has focused on patients with DM or with CKD and showed a higher amputation risk among younger patients,(4,5) but adjustment for those comorbidities did not attenuate the 24% increased risk of amputation in our analysis. Even though the absolute difference in amputation rates was relatively small, there were nearly 600 additional amputations in this subset of insured American adults due to the 2% absolute difference in amputation rates (MarketScan includes only people with specific forms of commercial health insurance). Most of the difference in amputation rates was due to a higher rate of minor amputations among younger patients. Though minor amputations are understood to be less morbid than major amputations, evidence suggests that minor amputations carry significant implications for survival, functional status, and subsequent major amputation.(22)(23) Minor amputations are also clinically meaningful on a national scale given that patients <65 years old are more likely to be employed and may therefore suffer added economic effects of even minor amputation and its associated recovery. The fact that the higher risk of amputation persisted in younger patients despite adjustment for demographics, comorbidities, and CLTI severity suggest that there are other, unaccounted for, risk factors experienced by younger patients. These may include diabetes severity, the lower rates of statin treatment, race/ethnicity, a family history of early atherosclerotic disease, or other genetic factors.
Despite this elevated risk of amputation, younger CLTI patients in Marketscan were not any more likely to undergo revascularization after adjustment. We hypothesized that younger patients would be more likely to undergo revascularization because of the increased potential benefit of limb salvage in the face of longer expected survival as well as because younger patients may be less frail and more likely to safely undergo revascularization. Therefore, the lack of elevated risk for revascularization among younger patients may actually represent missed opportunities to alter the course of disease in these younger patients. Though the reasons revascularization was not pursued are impossible to ascertain in a retrospective database review, it may be that younger patients had more diffuse infrapopliteal disease that was less amenable to revascularization. The overall rate of revascularization was relatively low for both younger and older patients in this analysis, which may be due to some patients undergoing revascularization prior to the study period.
Similarly, despite Level 1A recommendations for statin use in PAD patients,(13–15) statin prescription fill rates overall were very low (<50%) among the patients with prescription drug coverage (89.6% of the cohort). This is lower than an analysis of patients in the Veterans Affairs system with incident PAD, of whom 72% were taking some form of statin, though statin use was only 57.9% among patients with PAD as their only form of atherosclerosis.(24) Younger patients in our cohort were less likely to fill a prescription for statins over the year following index encounter even after controlling for other comorbidities that might influence statin prescription. Though we could not assess out-of-pocket costs, statins are available in relatively inexpensive generic formulations. There is limited evidence suggesting that younger patients may be less adherent to prescribed medications,(25) but the possibility that the age-related disparity in statin prescription fills is related to prescribing habits cannot be dismissed without further investigation, especially since our analysis was limited to patients with prescription drug coverage. Statin prescription for PAD patients has previously been used as a quality metric in the VQI Registry. The differential in statin prescription fills between younger and older patients suggests that younger patients may not be receiving adequate PAD guideline-directed care and points to a possible targeted intervention to improve limb outcomes in younger patients. This is especially problematic in light of evidence suggesting that atherosclerosis manifesting at a younger age follows a more aggressive natural history, meaning that failure to use evidence-based therapies among younger patients may expose them to even higher risks of thrombotic events.(26,27)
This analysis has some limitations. Most notably, mortality was not available in the MarketScan dataset during this study period. While we censored patients whose enrollment ID dropped out of the dataset, there are multiple reasons this might occur in addition to death (such as individual loss of insurance or change in the complement of insurances included in the dataset). This shortcoming is reflected in the fact that dementia and cancer were associated with lower likelihoods of MALE and amputation, when it is more likely that dementia and cancer marked an increased risk of dying prior to suffering MALE or amputation. Also, race/ethnicity is not available in MarketScan, eliminating our ability to understand and account for the role of this important factor in treatment and outcomes. Furthermore, this cohort included both prevalent and incident patients with CLTI. Because of the limited number of years of MarketScan data we had available, adequate look-back to confidently exclude prevalent cases was not feasible. In addition, the relative age composition of the analytic cohort versus all insured Americans is unknown, and it may be that younger patients were disproportionately represented in the cohort due to older patients participating in other Medicare plans or receiving care from the Veterans Affairs Health System.
Finally, the fact that MarketScan includes only commercially-insured individuals is significant limitation. The rates of MACE and MALE are likely even higher among patients without commercial insurance, including uninsured people and those with Medicaid. Given that Americans under the age of 65 are more likely to be uninsured (14.7% of American civilians aged 18–64 were uninsured in 2019 compared to <1% of American civilians aged ≥ 65),(28) the relationship between age and insurance status specifically among patients with CLTI requires further investigation. Age-related disparities in insurance coverage could exacerbate the age-related differences in outcomes we demonstrated in this insured MarketScan cohort. For instance, uninsured younger CLTI patients may be less likely to undergo revascularization procedures, leading to an even higher risk of amputation – and this may compound socioeconomic and racial disparities in younger patients, as poor and non-white younger Americans are more likely to be uninsured.(28) While the MarketScan dataset has many positive aspects, including its size, geographic diversity, age inclusiveness, longitudinal follow-up, and inclusion of patients regardless of whether they underwent a procedure or not, analysis of a nationally-representative, insurance-blind dataset is needed to further elucidate the ways in which the relationship between age and insurance status affect CLTI outcomes.
Conclusions:
Our analysis shows that commercially insured younger CLTI patients are a distinct subgroup with different comorbidity patterns, receiving different treatments, and suffering worse limb-related outcomes. These findings suggest that younger CLTI patients are deserving of additional study in order and targeted approaches to understand how to better care for them and improve their disease-related outcomes. Analyses of CLTI patient presentation, treatment, and outcome patterns should not extrapolate from CMS or procedural datasets to younger patients or patients not undergoing interventions.
Supplementary Material
Funding/support:
EHW was supported by the National Heart, Lung, And Blood Institute of the National Institutes of Health under award number F32HL151181.
Footnotes
Conflict of interest disclosures: EHW, CBF, PG, AC, and CL have no disclosures. MRP reports research support from the Agency for Healthcare Research and Quality, AstraZeneca, Bayer, Jansen, Procyrion, and Heartflow outside of the submitted work; Honoraria/advisory board participation with Bayer, Janssen Pharmaceuticals, and AstraZeneca outside of the submitted work. WSJ reports research support from Boehringer Ingelheim, Doris Duke Charitable Foundation, National Institutes of Health, and Patient-Centered Outcomes Research Institute outside of the submitted work; Advisory board participation with Bayer, Bristol-Myers Squibb, and Janssen Pharmaceuticals outside of the submitted work.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References:
- 1.Eraso LH, Fukaya E, Mohler ER 3rd, Xie D, Sha D, Berger JS. Peripheral arterial disease, prevalence and cumulative risk factor profile analysis. Eur J Prev Cardiol 2014;21:704–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Murphy D, McCulloch CE, Lin F et al. Trends in Prevalence of Chronic Kidney Disease in the United States. Ann Intern Med 2016;165:473–481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.System USDS. Diagnosed diabetes stratified by age. Centers for Disease Control and Prevention, 2020. [Google Scholar]
- 4.Franz D, Zheng Y, Leeper NJ, Chandra V, Montez-Rath M, Chang TI. Trends in Rates of Lower Extremity Amputation Among Patients With End-stage Renal Disease Who Receive Dialysis. JAMA Intern Med 2018;178:1025–1032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Geiss LS, Li Y, Hora I, Albright A, Rolka D, Gregg EW. Resurgence of Diabetes-Related Nontraumatic Lower-Extremity Amputation in the Young and Middle-Aged Adult U.S. Population. Diabetes Care 2019;42:50–54. [DOI] [PubMed] [Google Scholar]
- 6.Jones WS, Patel MR, Dai D et al. Temporal trends and geographic variation of lower-extremity amputation in patients with peripheral artery disease: results from U.S. Medicare 2000–2008. J Am Coll Cardiol 2012;60:2230–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kohn CG, Alberts MJ, Peacock WF, Bunz TJ, Coleman CI. Cost and inpatient burden of peripheral artery disease: Findings from the National Inpatient Sample. Atherosclerosis 2019;286:142–146. [DOI] [PubMed] [Google Scholar]
- 8.Brahmandam A, Gholitabar N, Cardella J et al. Discrepancy in Outcomes after Revascularization for Chronic Limb-Threatening Ischemia Warrants Separate Reporting of Rest Pain and Tissue Loss. Ann Vasc Surg 2020. [DOI] [PubMed] [Google Scholar]
- 9.Gabel J, Jabo B, Patel S et al. Analysis of Patients Undergoing Major Lower Extremity Amputation in the Vascular Quality Initiative. Ann Vasc Surg 2018;46:75–82. [DOI] [PubMed] [Google Scholar]
- 10.Mell MW, Pettinger M, Proulx-Burns L et al. Evaluation of Medicare claims data to ascertain peripheral vascular events in the Women’s Health Initiative. J Vasc Surg 2014;60:98–105. [DOI] [PubMed] [Google Scholar]
- 11.Fan J, Arruda-Olson AM, Leibson CL et al. Billing code algorithms to identify cases of peripheral artery disease from administrative data. J Am Med Inform Assoc 2013;20:e349–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kalbaugh CA, Loehr L, Wruck L et al. Frequency of Care and Mortality Following an Incident Diagnosis of Peripheral Artery Disease in the Inpatient or Outpatient Setting: The ARIC (Atherosclerosis Risk in Communities) Study. J Am Heart Assoc 2018;7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gerhard-Herman MD, Gornik HL, Barrett C et al. 2016 AHA/ACC Guideline on the Management of Patients With Lower Extremity Peripheral Artery Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol 2017;69:1465–1508. [DOI] [PubMed] [Google Scholar]
- 14.Conte MS, Bradbury AW, Kolh P et al. Global Vascular Guidelines on the Management of Chronic Limb-Threatening Ischemia. European journal of vascular and endovascular surgery : the official journal of the European Society for Vascular Surgery 2019;58:S1–S109.e33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Aboyans V, Ricco JB, Bartelink MEL et al. Editor’s Choice - 2017 ESC Guidelines on the Diagnosis and Treatment of Peripheral Arterial Diseases, in collaboration with the European Society for Vascular Surgery (ESVS). European journal of vascular and endovascular surgery : the official journal of the European Society for Vascular Surgery 2018;55:305–368. [DOI] [PubMed] [Google Scholar]
- 16.Birman-Deych E, Waterman AD, Yan Y, Nilasena DS, Radford MJ, Gage BF. Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors. Med Care 2005;43:480–5. [DOI] [PubMed] [Google Scholar]
- 17.Quan H, Sundararajan V, Halfon P et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005;43:1130–9. [DOI] [PubMed] [Google Scholar]
- 18.Torbjornsson E, Blomgren L, Fagerdahl AM, Bostrom L, Ottosson C, Malmstedt J. Risk factors for amputation are influenced by competing risk of death in patients with critical limb ischemia. J Vasc Surg 2020;71:1305–1314 e5. [DOI] [PubMed] [Google Scholar]
- 19.Heikkila K, Loftus IM, Mitchell DC, Johal AS, Waton S, Cromwell DA. Population-based study of mortality and major amputation following lower limb revascularization. Br J Surg 2018;105:1145–1154. [DOI] [PubMed] [Google Scholar]
- 20.Dermody M, Homsy C, Zhao Y, Goodney PP, Estes JM, Vascular Study Group of New E. Outcomes of infrainguinal bypass determined by age in the Vascular Study Group of New England. J Vasc Surg 2015;62:83–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Armstrong EJ, Ryan MP, Baker ER, Martinsen BJ, Kotlarz H, Gunnarsson C. Risk of major amputation or death among patients with critical limb ischemia initially treated with endovascular intervention, surgical bypass, minor amputation, or conservative management. J Med Econ 2017;20:1148–1154. [DOI] [PubMed] [Google Scholar]
- 22.Chiang N, Wang J, Marie N, Wu A, Ravindra R, Robinson D. Evaluation of Clinical Outcomes Following Minor Amputations in Australia - An Important Consideration for Timing of Revascularisation. Ann Vasc Surg 2021. [DOI] [PubMed] [Google Scholar]
- 23.Suckow BD, Goodney PP, Cambria RA et al. Predicting functional status following amputation after lower extremity bypass. Ann Vasc Surg 2012;26:67–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Arya S, Khakharia A, Binney ZO et al. Association of Statin Dose With Amputation and Survival in Patients With Peripheral Artery Disease. Circulation 2018;137:1435–1446. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kripalani S, Gatti ME, Jacobson TA. Association of age, health literacy, and medication management strategies with cardiovascular medication adherence. Patient Educ Couns 2010;81:177–81. [DOI] [PubMed] [Google Scholar]
- 26.McCready RA, Vincent AE, Schwartz RW, Hyde GL, Mattingly SS, Griffen WO Jr. Atherosclerosis in the young: a virulent disease. Surgery 1984;96:863–9. [PubMed] [Google Scholar]
- 27.Hallett JW Jr., Greenwood LH, Robison JG. Lower extremity arterial disease in young adults. A systematic approach to early diagnosis. Ann Surg 1985;202:647–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Cohen RA CA, Martinez ME, Terlizzi EP. Health insurance coverage: Early release of estimates from the National Health Interview Survey, 2019. In: Statistics DoHI, editor: National Center for Health Statistics, 2020. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.