Discordant Values in Lower Extremity Physiologic Studies Predict Increased Cardiovascular Risk.

Background Ankle-brachial indexes (ABI) are a noninvasive diagnostic tool for peripheral arterial disease and a marker of increased cardiovascular risk. ABI is calculated using the highest systolic blood pressure of the 4 ankle arteries (bilateral dorsalis pedis and posterior tibial). Accordingly, patients may be assigned a normal ABI when the result would be abnormal if calculated using one of the other blood pressure readings. Cardiovascular outcomes for patients with discordant ABIs are undescribed. Methods and Results We performed a retrospective study of patients who underwent ABI measurement for any indication between January 1996 and June 2018. Those with normal ABIs (1.00-1.39) were included. We compared patients with all 4 normal ABIs (calculated using all 4 ankle arteries; n=15 577, median age 64.0 years, 54.4% men) to those with discordant ABIs (at least 1 abnormal ABI ≤0.99; n=2095, median age 66.0 years, 47.8% men). The outcomes assessed were ischemic stroke, myocardial infarction, and all-cause mortality. Compared with patients with concordant normal ABIs, patients with discordant ABIs were older; women; smoked; and had chronic kidney disease, coronary artery disease, diabetes mellitus, chronic obstructive pulmonary disease, hypertension, or prior stroke. Patients with discordant ABIs had a greater risk of myocardial infarction (hazard ratio [HR], 1.31; 95% CI, 1.10-1.56), ischemic stroke (HR, 1.53; 95% CI, 1.37-1.72), and all-cause mortality (HR, 1.27; 95% CI, 1.16-1.39), including after adjustment for baseline comorbidities. Conclusions Discordant ABI results were associated with an increased risk of myocardial infarction, stroke, and all-cause mortality in the studied population. Clinicians should examine ABI calculations using all 4 ankle arteries to better characterize a patient's cardiovascular risk.

ankle‐brachial index

coronary artery disease

chronic kidney disease

hazard ratio

myocardial infarction

peripheral arterial disease

Clinical Perspective

What Is New?

This study evaluates the cardiovascular and mortality risks associated with a discordant ankle‐brachial index (ABI), referring to patients who had a normal ABI reported but had at least 1 abnormal ABI (<1.0) using the 4 ankle arteries.

We show that discordant ABI is associated with incident myocardial infarction, stroke, and all‐cause mortality in the studied population, including in models adjusted for baseline cardiovascular risk factors.

What Are the Clinical Implications?

We suggest that clinicians review normal lower extremity physiologic studies for the presence of discordant ABI measurements to better characterize a patient's cardiovascular risk.

Our results may encourage more aggressive screening and treatment of traditional modifiable cardiovascular risk factors in patients with discordant ABIs.

The ankle‐brachial index (ABI) is a simple, noninvasive study to assess for peripheral arterial disease (PAD). The index is calculated by dividing the highest systolic blood pressure from the dorsalis pedis or posterior tibial artery by that of the brachial artery. ABIs define PAD with reasonable accuracy and have been shown to predict cardiovascular events and mortality.1, 2, 3, 4, 5 Abnormal ABIs in asymptomatic patients have been associated with a higher incidence of death, myocardial infarction (MI), and stroke.6 Patients may be classified as having a normal ABI when 1 ankle artery systolic blood pressure is normal while the other ipsilateral ankle artery pressure is reduced or abnormal—a condition hereafter referred to as discordant ABI. Little attention is given to discordant ABI calculations, again referring to those patients who had a normal ABI reported, but also had an abnormal ABI calculated using the lower of the 2 systolic blood pressures from the ankle arteries for that extremity. We question whether discordant ABIs are a marker of increased cardiovascular and mortality risks and as such should be given consideration when making clinical decisions.

Methods

We conducted a retrospective review of patients who had an ABI performed at Mayo Clinic. We identified all patients who had ABIs performed for any indication between January 1996 and June 2018, excluding patients younger than 18 years of age and those who did not provide research authorization. The cohort was then limited to those with normal ABIs of 1.00 to 1.39. We manually reviewed the ABI reports to assess the 4 ABI calculations performed to derive the reported ABI (the highest of the 2 calculated ABIs for each leg is standardly reported). The 4 ABIs correspond to the 4 ankle arteries (bilateral tibial and dorsalis pedis), specifically the systolic blood pressure at each ankle artery over the corresponding brachial artery systolic blood pressure. Those with discordant ABIs, meaning at least 1 abnormal ABI calculated using the 4 ankle arteries (whereas the ABI reported for each leg was normal), were identified. Abnormal was defined as ABI ≤0.99. Specifically, those with 1 abnormal ABI (other ipsilateral ABI and 2 ABIs in the opposite leg were normal) or 1 abnormal ABI in both legs were classified as discordant (whereas patients with 2 normal ABIs for each leg were considered normal).

The research complies with the guidelines for human studies and was conducted ethically. The Mayo Clinic institutional review board approved this study. The review type was classified as minimal risk; therefore, the requirement for informed consent was waived. The data that support the findings of the study are available from the corresponding author upon reasonable request.

Variables Assessed

Baseline characteristics for the cohort were assessed and included age, sex, smoking history (current or prior smoker), coronary artery disease (CAD), chronic kidney disease (CKD), chronic obstructive pulmonary disease, diabetes mellitus, atrial fibrillation, hyperlipidemia, hypertension, and history of prior MI or ischemic stroke. The CKD variable was limited to CKD stage ≥3, including end‐stage renal disease or dialysis. There were no missing data for the baseline variables assessed. Baseline (prior to ABI) use of the following medications was also assessed: aspirin, other antiplatelet agent (clopidogrel, ticagrelor, prasugrel), anticoagulant (warfarin or nonvitamin K anticoagulant), and statin. The outcomes assessed were ischemic stroke, MI, and all‐cause mortality.

International Classification of Diseases, Ninth Revision (ICD‐9) and International Classification of Diseases, Tenth Revision (ICD‐10) diagnostic codes were used to define the baseline medical conditions and outcomes for extraction from inpatient and outpatient medical records (Table S1). Mortality data were obtained from the electronic medical records. Random samples of both stroke and MI outcomes were manually verified through a review of the electronic medical record (Data S1).

Statistical Analysis

Mean and SD, median and range, or absolute number and percentage were used to describe the continuous and categorical patient characteristics. Wilcoxon rank‐sum test and chi‐square tests of significance were employed to compare those with normal and discordant ABIs. Bonferroni correction (α=0.05/3) was used to adjust for multiple testing. A P value <0.05 was considered statistically significant. Hazard regression analysis was conducted to calculate hazard ratios (HRs) with 95% CIs for the outcomes of interest, comparing the group with discordant ABIs versus those with normal ABIs. This analysis was also completed with 2 adjusted models. The first adjusted for age and sex, and the second adjusted for age, sex, diabetes mellitus, CKD stage 3 or greater including end‐stage renal disease or dialysis, CAD, hypertension, hyperlipidemia, and smoking history. Multivariate hazard regression analysis was performed to assess the baseline variables associated with each outcome for those with normal and those with discordant ABIs. Variables significant in the univariate regression were included in the multivariate analysis. All statistical analyses were performed with SAS version 9.4 software (SAS Institute Inc., Cary, NC).

Results

A total of 40 174 patients were identified who underwent ABI measurement, and 17 672 of these patients had an overall normal ABI reported and were included in the study. Of those, 2095 (11.9%) had discordant ABIs and were compared with those who did not have discordant ABIs (n=15 577). The median ABI value for those with discordant ABIs was 0.97 (range, 0–0.99). The median ABI for those with normal ABIs was 1.10 (range, 1.00–1.39). The median follow‐up for the cohort was 36.3 months (interquartile range, 95.0 months).

Baseline Characteristics

The baseline characteristics for the groups of patients with normal and discordant ABIs are shown in Table 1. There were no significant differences in race or ethnicity between the groups. Most of the assessed baseline comorbidities were significantly more prevalent in the discordant ABI group. Patients with discordant ABIs were more likely to be older (median age 66.0 versus 64.0 years), women (52.2% versus 45.6%; P<0.0001), smokers (44.2% versus 40.2%; P<0.0001), and have a history of CKD (6.3% versus 5.0%; P=0.01), CAD (32.4% versus 28.2%; P<0.0001), chronic obstructive pulmonary disease (3.9% versus 2.4%; P<0.0001), diabetes mellitus (24.2% versus 21.6%; P=0.007), hypertension (51.7% versus 49.3%; P=0.04), and prior ischemic stroke (4.2% versus 2.9%; P<0.0001). There were no significant differences in the prevalence of atrial fibrillation, hyperlipidemia, prior MI, or prior stroke between the groups. There were no significant differences in the baseline drugs assessed (aspirin or antiplatelet, anticoagulant, or statin therapies).

Table 1

Baseline Characteristics of Patients With Normal and Discordant ABIs

	Normal ABI (N=15 577)	Discordant ABI (N=2095)	P Value
Median age, y (range in y)	64.0 (18.0–102.0)	66.0 (18.0–96.0)	<0.0001
Male sex, n (%)	8472 (54.4)	1001 (47.8)	<0.0001
Smoking history, n (%)	6173 (39.6)	927 (44.2)	<0.0001
CAD, n (%)	4388 (28.2)	679 (32.4)	<0.0001
CKD, n (%)	777 (5.0)	131 (6.3)	0.01
COPD, n (%)	370 (2.4)	81 (3.9)	<0.0001
Diabetes mellitus, n (%)	3369 (21.6)	508 (24.2)	0.007
Atrial fibrillation, n (%)	1622 (10.4)	243 (11.6)	0.10
Hyperlipidemia, n (%)	6447 (41.4)	870 (41.5)	0.90
Hypertension, n (%)	7676 (49.3)	1083 (51.7)	0.04
MI, n (%)	603 (3.9)	99 (4.7)	0.06
Stroke, n (%)	450 (2.9)	87 (4.2)	<0.0001
Aspirin, n (%)	3848 (24.7)	532 (25.4)	0.49
Antiplatelet, n (%)	701 (4.5)	101 (4.8)	0.51
Anticoagulant, n (%)	1256 (8.1)	180 (8.6)	0.41
Statin, n (%)	3382 (21.7)	449 (21.4)	0.77

P values were calculated by 2‐sample t tests and chi‐square tests for continuous and categorical variables, respectively. ABI indicates ankle‐brachial index; CAD, coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; and MI, myocardial infarction.

Outcomes

There was a total of 1007 ischemic strokes (5.7%), 1068 MIs (6.0%), and 4033 deaths (23.5%, n=489 missing data) in the cohort after the index date (ABI measurement). The combined outcome count in the cohort was 6167 (34.9%). Compared with those with normal ABIs, the discordant ABI group had a high percentage of ischemic stroke (6.6% [n=139] versus 5.6% [n=868]), MI (7.1% [n=149] versus 5.9% [n=919]), and death (26.4% [n=544] versus 23.1% [n=3489]).

Compared with those with normal ABIs, those with discordant ABIs had a significantly increased risk of MI (HR, 1.31; 95% CI, 1.10–1.56), stroke (HR, 1.26; 95% CI, 1.05–1.51), and all‐cause mortality (HR, 1.27; 95% CI, 1.16–1.39) as shown in Table 2. The analysis remained significant with similar HRs when adjusting for baseline comorbidities (Table 2).

Table 2

HRs for the Outcomes of Interest Comparing Patients With Discordant and Normal Ankle‐Brachial Indexes

	Unadjusted HR (95% CI)	Adjusted HR (95% CI)a	Adjusted HR (95% CI) b
Ischemic stroke	1.26 (1.05–1.51)	1.23 (1.03–1.47)	1.20 (1.00–1.44)
Myocardial infarction	1.31 (1.10–1.56)	1.29 (1.09–1.54)	1.21 (1.01–1.44)
All‐cause mortality	1.27 (1.16–1.39)	1.19 (1.08–1.30)	1.15 (1.05–1.26)

HR indicates hazard ratio.

Adjusted for age and sex.

Adjusted for age, sex, diabetes mellitus, chronic kidney disease stage 3 or greater, coronary artery disease, hypertension, hyperlipidemia, and smoking history.

Predictors of MI, Ischemic Stroke, and All‐Cause Mortality

Multivariate modeling for both normal and discordant ABI groups evaluating baseline risk factors for the outcomes of MI, ischemic stroke, and all‐cause mortality is shown in Tables 3, 4 through 5, respectively. Univariate modeling for the 3 outcomes is shown in Tables S2 through S4. The most predictive risk factors for the 3 outcomes were similar among the discordant and ABI cohorts, although the HR magnitude varied as did the full array of predictive variables.

Table 3

Multivariate Regression Analysis for Myocardial Infarction Comparing Patients With Discordant and Normal ABIs

Variable	Normal ABI, HR (95% CI)	Discordant ABI, HR (95% CI)
Age	1.01 (1.01–1.02)a	1.02 (1.01–1.04)a
Male	1.07 (0.93–1.24)a	1.02 (0.72–1.46)a
Diabetes mellitus	1.23 (1.06–1.41)a	1.47 (1.04–2.09)a
CKD	1.74 (1.39–2.17)a	1.64 (0.96–2.80)
CAD	2.02 (1.72–2.36)a	1.69 (1.17–2.45)a
Hypertension	1.29 (1.09–1.52)a	0.90 (0.60–1.35)
Hyperlipidemia	0.76 (0.65–0.90)a	0.93 (0.62–1.37)
Smoker	1.16 (1.02–1.33)a	0.89 (0.63–1.26)
COPD	1.08 (0.75–1.55)	0.63 (0.23–1.78)
Atrial fibrillation	1.26 (1.05–1.51)a	1.26 (0.81–1.95)

Variables significant in the univariate regression were included in the multivariate analysis. ABI indicates ankle‐brachial index; CAD, coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; and HR, hazard ratio.

Significant P value (<0.05), calculated by type 3 Wald.

Table 4

Multivariate Regression Analysis for Ischemic Stroke Comparing Patients With Discordant and Normal ABIs

Variable	Normal ABI, HR (95% CI)	Discordant ABI, HR (95% CI)
Age	1.02 (1.01–1.02)a	1.02 (1.00–1.03)a
Male	1.10 (0.95–1.27)	…
Diabetes mellitus	1.27 (1.09–1.48)a	1.03 (0.70–1.50)
CKD	1.05 (0.79–1.40)	…
CAD	1.28 (1.10–1.50)a	0.94 (0.65–1.35)
Hypertension	1.30 (1.11–1.53)a	1.58 (1.02–2.43)a
Hyperlipidemia	1.13 (0.97–1.31)	1.83 (1.22–2.73)a
Smoker	…	…
COPD	1.43 (0.98–2.07)	1.38 (0.63–3.01)
Atrial fibrillation	1.43 (1.19–1.73)a	2.04 (1.33–3.12)a

Variables significant in the univariate regression were included in the multivariate analysis. Empty cells indicate that the variable was not significant in the univariate analysis. ABI indicates ankle‐brachial index; CAD, coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; and HR, hazard ratio.

Significant P value (<0.05), calculated by type 3 Wald.

Table 5

Multivariate Regression Analysis for All‐Cause Mortality Comparing Patients With Discordant and Normal ABIs

Variable	Normal ABI, HR (95% CI)	Discordant ABI, HR (95% CI)
Age	1.06 (1.05–1.06)a	1.04 (1.04–1.05)a
Male	1.31 (1.22–1.41)a	1.24 (1.04–1.49)a
Diabetes mellitus	1.43 (1.32–1.55)a	1.52 (1.26–1.83)a
CKD	2.23 (1.97–2.52)a	2.34 (1.77–3.09)a
CAD	1.36 (1.26–1.47)a	1.39 (1.15–1.68)a
Hypertension	0.99 (0.91–1.06)	0.99 (0.81–1.21)
Smoker	1.12 (1.05–1.20)a	…
COPD	1.58 (1.31–1.89)a	1.36 (0.87–2.13)
Atrial fibrillation	1.49 (1.36–1.62)a	1.55 (1.25–1.94)a

Variables significant in the univariate regression were included in the multivariate analysis. Hyperlipidemia did not remain significant and was not included in the model for either ABI group. Empty cells indicate that the variable was not significant in the univariate analysis. ABI indicates ankle‐brachial index; CAD, coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; and HR, hazard ratio.

Significant P value (<0.05), calculated by type 3 Wald.

Discussion

Study Outcomes

Our study in a large cohort found that those with discordant ABIs compared with normal ABIs had a significantly higher risk of MI, ischemic stroke, and all‐cause mortality. Discordant ABI measurements independently predicted these outcomes when adjusted for baseline comorbidities. A substantial percentage (12%) of patients reported as having normal ABIs had discordant indexes. PAD has been strongly associated with cardiovascular outcomes. Patients with PAD, including PAD defined by abnormal ABI, have increased risk of MI, stroke, cardiovascular mortality, and all‐cause mortality compared with those without PAD.7, 8, 9 Our results suggest that a similar relationship exists for those with discordant ABIs.

There was a fairly high percentage of patients who experienced the outcomes of interest in the cohort, likely reflecting the high baseline comorbidity of patients who are referred for ABI testing. However, mortality and cardiovascular event rates were less than that seen in patients with PAD, especially those with prior CAD or stroke.4, 10

Baseline Status of Patients With Discordant and Normal ABIs

We noted that baseline comorbidities were higher in the group of patients with discordant ABIs. Many of these risk factors are associated with PAD. Age, diabetes mellitus, and smoking have most commonly been associated with PAD of the identified risk factors.11, 12 Additional traditional risk factors include hyperlipidemia, hypertension, and metabolic syndrome.13 Renal disease has also been associated with incident PAD.13, 14 CAD is a risk factor for PAD given both conditions are manifestations of atherosclerosis, although PAD is often unrecognized in those with CAD.15, 16, 17

Our study interestingly found that women had a higher prevalence of discordant ABIs than men, which may support that women are more likely to have underrecognized cardiovascular risk. The women in the study were younger than the men in both the normal and discordant ABI groups (mean age in the discordant group, 60.32±17.56 versus 64.16±13.68). Research has shown a similar prevalence of PAD between men and women overall, with a higher prevalence in men at younger ages.18, 19 Women appear to present later, have higher rates of asymptomatic and unrecognized disease, and continue to be underrepresented in PAD studies.16, 19

There was no significant difference in the prevalence of atrial fibrillation between the groups. Studies offer conflicting results regarding the association between atrial fibrillation and PAD.20, 21 The lack of difference between groups in the baseline use of aspirin and antiplatelet agent, anticoagulation, and statin therapies likely reflects the similar prevalence of hyperlipidemia, atrial fibrillation and prior major bleeding (8.4% in those with normal ABI versus 7.9% in those with discordant ABI, P=0.46). However, both groups may have been undertreated given the high prevalence of baseline cardiovascular disease or risk, which may be especially relevant in those with discordant ABIs given our study findings.

Predictors of Study Outcomes

Traditional cardiovascular risk factors were associated with MI, ischemic stroke, and all‐cause mortality in both the normal and discordant ABI groups. However, there were several differences in predictive variables between groups, which may point to the importance of particular comorbidities in predicting adverse cardiovascular events in those with discordant ABIs. For instance, diabetes mellitus may be an especially impactful risk factor for MI in patients with discordant ABIs compared with those with normal ABIs. PAD patients with diabetes mellitus are at higher risk for rapid progression of their PAD and development of cardiovascular outcomes than PAD patients without diabetes mellitus.22

Study Implications

We found that patients with discordant ABIs at our institutions had a higher risk of MI, ischemic stroke, and death, even after adjusting for baseline comorbidities. Our results may encourage more aggressive traditional risk factor modifications in patients with discordant ABIs. Tailored to individual patients, clinicians could consider more intensive lifestyle modifications, therapies to promote smoking cessation, more potent lipid‐lowering therapy, tighter hypertension and glycemic controls, and aspirin use for the primary prevention of cardiovascular events if the benefits were felt to outweigh the harms given that this population represents a higher risk group. We do not suggest regarding a discordant ABI as a diagnosis of PAD (to warrant treatment as such) or as a finding to provoke invasive testing. A discordant ABI measurement may be a precursor to PAD. Future research could assess for incident PAD in this population as well as claudication and limb ischemia. Future research may also evaluate noncompressible ABIs (>1.4) and abnormal ABIs (≤0.9) as comparators. Another comparator of interest is discordant ABIs with a greater degree of abnormality (ie, ≤0.9); we hypothesize that more abnormal discordant ABIs would convey greater risk for the outcomes studied.

Limitations

Limitations of this study include that it is retrospective with variable follow‐up. All patients who underwent ABI testing were included. Whether testing was done for screening purposes was not noted. Testing was limited to resting ABI. Exercise ABI and toe‐brachial index were not assessed, perhaps leading to a misclassification of disease in some patients. We did not assess functional capacity or its decline in the cohort. ICD‐9 and ICD‐10 diagnosis codes were used to define baseline variables and outcomes, although manual validation indicated comparable accuracy with that of previous studies. Outcomes that occurred outside of the Mayo Clinic healthcare system would not have been included. The cohort was predominantly white, which may reduce the generalizability of our results. Despite these limitations, the study is unique, examining a previously unstudied variable that appears to represent a significant and independent marker of cardiovascular risk.

Conclusions

PAD remains a common but underdiagnosed and undertreated disease, including in patients with known ischemic heart disease.13, 16 PAD confers a high risk of cardiovascular events and mortality, with similar risks for both symptomatic and asymptomatic diseases.23 To our knowledge, this is the first study to examine cardiovascular risk associated with discordant ABIs in a large cohort referring to studies that reported an overall normal ABI but had at least 1 abnormal reading based on an assessment of the 4 ankle arteries. Discordant ABIs appear to better characterize a patient's risk for MI, ischemic stroke, and all‐cause mortality in our population. The prevalence of discordant ABIs in the cohort was sizable, indicating a potentially large, unrecognized fraction of the community that may benefit from more intensive cardiovascular risk factor modification. Accordingly, discordant values should be reported for all ABI studies. Given our study findings, clinicians should be cognizant of discordant ABI values and their potential clinical implications.

Sources of Funding

None.

Disclosures

None.

Data S1 Tables S1–S4 Reference 24

Click here for additional data file.