Ankle-brachial index (ABI) is conventionally derived as the ratio of higher of the 2 systolic ankle blood pressures to the higher brachial pressure (HABI method). Alternatively, ABI may be derived using the lower of the 2 systolic ankle pressures (LABI method). The objective of this study was to assess the utility and difference between 2 techniques in predicting peripheral artery disease (PAD). Participants who underwent both ABI measurement and arteriography from July 2005 to June 2010 were reviewed. Angiographic disease burden was scored semiquantitatively (0 = <50%, 1 = 50% to 75%, and 2 = >75% stenosis of any lower extremity arterial segment), and PAD by angiography was defined as >50% stenosis of any 1 lower extremity arterial segment. A combined PAD disease score was calculated for each leg. A total of 130 patients were enrolled (260 limbs). The ABI was <0.9 (abnormal) in 68% of patients by HABI method and in 84% by LABI. LABI method had higher sensitivity and overall accuracy to detect PAD compared with the HABI method. Regression analysis showed that an abnormal ABI detected by LABI method is more likely to predict angiographic PAD and total PAD burden compared with HABI. Moreover, abnormal ABI by LABI method had higher sensitivity and accuracy to detect PAD in patients with diabetes and below knee PAD compared with the HABI method. In conclusion, ABI determined by the LABI method has higher sensitivity and is a better predictor of PAD compared with the conventional (HABI) method.
Peripheral artery disease (PAD) is highly prevalent and is predicted to increase because of aging of the population. The ankle-brachial index (ABI) is as an easy, reliable noninvasive test used to screen patients for lower extremity PAD and has low interobserver variability. Conventionally, ABI is calculated as the ratio of the higher of the systolic blood pressures (SBPs) of the 2 ankle arteries of that limb (either the dorsalis pedis or posterior tibial) and the higher of the 2 SBPs of the upper limbs (HABI method). ABI calculated by HABI method underestimates the true prevalence of PAD, especially in the elderly population. This is important because patients with PAD are at increased risk for future cardiovascular events. The lack of awareness of signs and symptoms of PAD and ABI calculation by the HABI method may underdiagnose PAD. Alternatively, ABI can be calculated using the lower of the 2 ankle pressures (LABI method), which may improve sensitivity for detecting PAD. However, there are limited data on verification of diagnostic accuracy of this method using angiography. We hypothesized that LABI method would improve detection of PAD. Therefore, our study aimed to (1) study the diagnostic accuracy of these 2 methods in the detection of PAD and total PAD burden, (2) study the diagnostic accuracy in patients with diabetes in whom ABI determined by the HABI method is often falsely elevated because of medial calcinosis, and (3) assess the diagnostic utility for detecting below knee PAD that is unknown.
Methods
This was a single-center retrospective study performed at a major tertiary referral academic medical center. All patients who underwent both ABI measurement and arteriography of the lower extremities with digital subtraction angiography (DSA) performed from July 2005 to June 2010 at our institution were reviewed. Only patients who had an ABI done within 6 months before the angiogram were included in this study. Exclusion criterion included previous limb amputations proximal to the heads of metatarsals or proximal to the elbow in the upper limbs, previous bypass surgery, stenting, or prosthetic vascular reconstruction to the lower limbs or of the arteries of lower limb/abdominal aorta or subclavian or axillary arteries, an ABI >1.3 in both lower limbs, and any abdominal or lower extremity vascular surgery or intervention between the time of having the ABI measurement and the first available angiography. The study protocol was approved by the hospital ethics and human subjects committee.
For measurement of ABI, an Unetixs Vascular Incorporated Multilab Series 2-CP (Unetixs Inc, Rhode Island) with an 8- and 5-MHz bidirectional Doppler wave probe device was used. It also incorporated a calibrated dual-channel pulse volume recording for definitive quality waveform assessment. Appropriately sized cuffs were applied on the lower extremities, just above the malleoli, and on the arms. These were performed by an experienced examiner who was blinded to all clinical baseline parameters assessed. Measurements were performed, after a 5- to 10-minute rest, in the supine position with the upper body as flat as possible to minimize the effect of an increased tibial artery blood pressure because of sitting or semi-sitting position. ABI values were then calculated applying 2 different methods: the higher ankle pressure (either the posterior tibial or dorsalis pedis artery) was used as the numerator for the HABI method and the lower ankle pressure (either posterior tibial or dorsalis pedis artery) was used as the numerator for the LABI method. An abnormal ABI was defined as <0.9 for both methods.
Intra-arterial DSA was performed and assessed by consensus agreement by 2 experienced readers who were blinded to the clinical and ABI data. Arteriographies were performed within a 6-month period from the ABI measurements. Appropriate anteroposterior sequential views of the lower abdomen, pelvis, and lower extremities were obtained. Oblique views were obtained for the iliac and the proximal femoral arteries. Percentage stenosis was defined as a >50% diameter reduction determined by visual estimation and by quantitative measurement assessment. Stenosis was calculated as the ratio of the residual target vessel lumen diameter to the diameter of the reference segment of artery. Discrepant results between the 2 readers were assessed by a third experienced reader. All readers were blinded to the ABI data and measurements.
Angiographic disease was scored using the quantitative coronary assessment method (0 = <50%; 1 = 50% to 75%; 2 = >75% occlusion) of any lower extremity arterial segment. A combined PAD disease score was calculated based on the total number of segments affected in each leg separately. Below knee PAD was defined as involvement (>50%) of at least 1 segment of tibioperoneal trunk, peroneal artery, and anterior tibial and posterior tibial arteries. Above knee PAD was defined as involvement (>50%) of at least 1 segment of common iliac artery, external iliac artery, common femoral artery, superficial femoral artery, and popliteal artery. Patients with a fasting glucose ≥126 mg/dl, with an HbA1C level ≥6.5%, or on treatment with oral medications or insulin were defined as patients with diabetes in our study. Those with an SBP >140 mm Hg, with a diastolic blood pressure >85 mm Hg, or on treatment for hypertension were defined as patients with hypertension.
Statistical analysis was performed using the Statistical Program for Social Sciences, version 19.0 (SPSS Inc., Chicago, Illinois). Data were plotted (e.g., histograms and spaghetti plots linking variables) to examine for potential outliers and for the necessity of transformation before analysis. Summary statistics (e.g., mean, SD, minimum, maximum, proportions) was calculated for all variables. Continuous variables are expressed as mean ± SD. Pearson’s correlation analysis was performed to calculate correlation between continuous variables. Chi-square test was used to find association between categorical variables. An adjusted McNemar’s statistical test was used to compare the sensitivity and specificity of the 2 ABI methods, compared with DSA as the gold standard. Binary regression analysis was performed to identify predictors of PAD variables included in the model were age, gender, diabetes, hypertension, smoking history, coronary artery disease, LABI, and HABI methods. Linear regression analysis was used to identify predictors of total PAD burden. A p value <0.05 was accepted as indicating statistical significance.
Results
Baseline patient’s characteristics are noted in Table 1 . A total of 130 patients were enrolled (260 limbs). Patients who underwent angiography were older, were symptomatic, and had a higher prevalence of hypertension, tobacco use, dyslipidemia, coronary artery disease, and diabetes ( Table 1 ). Sixty-eight percent were diagnosed with an abnormal ABI using the HABI method compared with 84% with an abnormal ABI using the LABI method.
| Variable | All Limbs (n = 260) | HABI <0.9 (n = 173) | LABI <0.9 (n = 215) |
|---|---|---|---|
| Mean age (years) | 68 ± 9 | 69 ± 9 | 62 ± 6 |
| Body mass index (Kg/m 2 ) | 29 ± 6 | 29 ± 6 | 30 ± 8 |
| Male sex | 59% | 54% | 57% |
| Symptomatic | 94% | 95% | 94% |
| Tobacco use | 92% | 92% | 92% |
| Hypertension | 85% | 85% | 85% |
| Diabetes | 31% | 35% | 31% |
| Dyslipidemia | 75% | 74% | 74% |
| Chronic kidney disease | 17% | 14% | 15% |
| Cerebrovascular accident | 12% | 12% | 12% |
| Coronary artery disease | 81% | 77% | 79% |
| Renal artery stenosis | 8% | 9% | 9% |
| Carotid artery stenosis | 27% | 31% | 29% |
| LVEF | 56 ± 12 | 57 ± 12 | 61 ± 6 |
LABI method had a higher sensitivity and overall accuracy to detect >50% stenosis in one or more arterial segment compared with the HABI method ( Table 2 ). McNemar’s test demonstrated that differences seen between both methods were statistically significant (p <0.0001). Receiver operating characteristic (ROC) curve analysis showed that the area under curve was better for LABI method compared with the HABI method ( Figure 1 ). Binary logistic regression analysis indicated that an abnormal ABI by LABI method is more likely to predict PAD (>50% stenosis in one or more arterial segment) by angiography compared with the HABI method after adjusting for confounding factors ( Table 3 ). LABI method had higher sensitivity but lower specificity to detect >75% stenosis in one or more arterial segment compared with the HABI method ( Table 2 ). ROC analysis showed that the area under curve for LABI method was better than the HABI method ( Figure 2 ).
| At Least One Segment ≥50% | At Least One Segment ≥75% | |||
|---|---|---|---|---|
| HABI | LABI | HABI | LABI | |
| Sensitivity | 75% | 90% | 81 | 92 |
| Specificity | 63% | 47% | 57 | 34 |
| Positive predictive value | 90% | 88% | 79 | 74 |
| Negative predictive value | 36% | 52% | 60 | 69 |
| Overall accuracy | 73% | 83% | 73 | 73 |
| Odds Ratio | 95% CI | p Value | |
|---|---|---|---|
| LABI | 5.3 | 1.6–16 | 0.005 ∗ |
| HABI | 3.4 | 1.2–10 | 0.022 ∗ |
| Male gender | 6 | 2.5–14 | <0.001 ∗ |
| Age | 1.02 | 0.98–1.07 | 0.19 |
| Body mass index | 0.98 | 0.92–1.05 | 0.737 |
| Hypertension | 1.3 | 0.421–4.03 | 0.65 |
| Hyperlipidemia | 0.8 | 0.3–1.9 | 0.65 |
| Diabetes mellitus | 1.6 | 0.58–4.7 | 0.33 |
| Chronic kidney disease | 1.3 | 0.46–4.08 | 0.56 |
| Smoking | 1.4 | 0.34–6 | 0.61 |
∗ Final predictors of PAD after adjusting for variables age, BMI, DM, HTN, HLD, CKD, and smoking.
Both LABI and HABI methods had significant inverse correlation with total angiographic PAD burden score (LABI, r = −0.48, p <0.001; HABI, r = −0.41, p <0.001). Binary logistic regression analysis showed that an abnormal ABI detected by LABI is more likely to predict angiographic PAD in ≥3 segments compared with HABI after adjusting for confounding factors ( Table 4 ). Linear regression analysis also showed that an abnormal ABI detected by the LABI method is more likely to predict total PAD burden compared with the HABI method (B = −0.38, 95% confidence interval [CI] −5.4 to −2.7, p <0.001, vs B = −0.04, 95% CI −3.1 to 2.1, p = 0.7) after adjusting for confounding factors. ROC analysis showed that the area under curve for the LABI method was better than the HABI method for detecting PAD defined by >3 segment stenosis in at least 1 segment by angiography ( Figure 3 ).
| Odds Ratio | 95% CI | p Value | |
|---|---|---|---|
| LABI | 5.7 | 2.3–14.5 | <0.001 ∗ |
| HABI | 1.2 | 0.4–2.9 | 0.7 |
| Male gender | 1.7 | 0.9–3.2 | 0.1 |
| Age | 1.06 | 1.03–1.1 | <0.001 |
| Body mass index | 0.97 | 0.91–1.03 | 0.34 |
| Hypertension | 1.1 | 0.45–2.7 | 0.8 |
| Hyperlipidemia | 1.37 | 0.64–2.9 | 0.4 |
| Diabetes mellitus | 1.3 | 0.6–2.5 | 0.46 |
| Chronic kidney disease | 1.3 | 0.5–3.3 | 0.51 |
| Smoking | 4.3 | 0.9–21.2 | 0.06 |
| Left ventricular ejection fraction <50% | 1.6 | 0.6–3.8 | 0.3 |
∗ Final predictors of PAD after adjusting for variables age, BMI, DM, HTN, HLD, CKD, smoking, and LVEF.
A total of 41 patients had diabetes (82 limbs). Compared with the HABI method, the LABI method had better sensitivity and accuracy and similar specificity to detect PAD by angiography (McNemar’s test: p <0.01; Table 5 ). Binary logistic regression analysis identified an abnormal ABI by LABI method (odds ratio 12, 95% CI 1.9 to 76, p = 0.009) and male gender (odds ratio 6.8, 95% CI 1.1 to 41, p = 0.04) as independent predictors of PAD by angiography after adjusting for confounding variables and HABI method. Figure 4 shows that ROC was better for LABI method compared with the HABI method to detect ≥50% stenosis in at least 1 arterial segment by arteriography. The area under curve for LABI method was better than the HABI method when looking at severe lesions with >75% stenosis in at least 1 arterial segment by arteriography as shown in Figure 5 .
| At Least One Segment ≥50% | At Least One Segment ≥75% | |||
|---|---|---|---|---|
| HABI | LABI | HABI | LABI | |
| Sensitivity | 77% | 88% | 81% | 91% |
| Specificity | 40% | 40% | 42% | 35% |
| Positive predictive value | 90% | 91% | 80% | 80% |
| Negative predictive value | 20% | 33% | 45% | 58% |
| Overall accuracy | 72% | 82% | 71% | 77% |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
