Different Methods of Ankle-Brachial Index Calculation on the Prevalence of Peripheral Arterial Disease and Arterial Calcification in Subjects with Type 2 Diabetes Mellitus from a Public Hospital in Peru

Study Design and Clinical Setting

An observational, cross-sectional, and descriptive study was conducted at Hospital María Auxiliadora, Lima, Peru, utilizing the database from risk foot consultations for patients with type 2 diabetes (DM2) in the endocrinology department, covering the period from December 1^st of 2015 to January 31th of 2020. This hospital is part of the Lima Sur Health Directorate.

Population and Sample

The database included patients with all types of diabetes who had no issues performing the ankle-brachial index (ABI), which included not having lesions or wounds on lower limbs, severe pain in lower limbs that hindered the placement of the sphygmomanometer cuff, poorly controlled blood pressure, or severe edema in lower limbs. For this study, records with complete ABI data were included.

Records with four complete ABIs were included, taking the entire available population and employing a non-probabilistic, census sampling method.

Variables

Ankle-Brachial Index (ABI): This is the result of a mathematical calculation between the systolic pressures of the lower and upper limbs and was estimated using three methods applied to the numerator of the index. The lowest, highest, and average systolic pressure from the ipsilateral pedal or posterior tibial artery were used as the numerator. The highest brachial pressure was used as the denominator. PAD was defined as any ABI <0.9; arterial calcification was defined as an ABI >1.3; and the rest were categorized as normal. [24] (Fig 1).

• Download figure
• Open in new tab

Fig 1. Diagnostic flowchart for PAD according to ABI criteria.

Other Variables: Demographic variables described included sex (male/female), age group (<60 years/61 to 74.9/75 and older), and education level (none/primary/secondary/higher). History variables included: stroke (no/yes), coronary artery disease (absent/present), hypertension (absent/present), and chronic kidney disease (absent/present). Diabetes-related variables included: disease duration (0-30 years), diabetic neuropathy (absent/present), and diabetic retinopathy (absent/present). Laboratory results included: hypercholesterolemia (less than 100 mg/dl/more than 100 mg/dl), hypertriglyceridemia (triglycerides less than 150 mg/dl/more than 150 mg/dl), glycated hemoglobin (less than 7%/7% or more). Poor metabolic control was defined as a glycated hemoglobin level equal to or greater than 7%, following the recommendations of the American Diabetes Association [25]. Systolic blood pressure was considered high if ≥140 mmHg and/or diastolic pressure ≥90 mmHg for individuals at low cardiovascular risk (less than 15% over 10 years). Concerning cholesterol levels, the goal was LDL < 100 mg/dl, in addition to HDL ≥ 40 in men or ≥ 50 in women and triglycerides ≥ 150 mg/dl. [26]

At-Risk Foot Program

The At-Risk Foot Program of the Endocrinology Service at Hospital Nacional María Auxiliadora conducts clinical assessments based on the guidelines of the "International Working Group of Diabetic Foot Guideline on the Prevention of Foot Ulcers in Persons with Diabetes 2015." It assesses symptoms using the Total Symptom Score and foot care habits. It conducts foot inspections, assesses deformities, and evaluates protective, proprioceptive, and Achilles reflex sensitivity, while verifying pulse conditions. Systolic pressures in both arms and in the posterior tibial and pedal arteries were measured using a Yuwell mercury column sphygmomanometer and a Huntleigh pocket Doppler ultrasound device of 8 MHz.

The evaluation was carried out by endocrinologists with experience and training in diabetic foot care.

Procedures

Permission was requested from the Head of the Endocrinology Service for authorization to use the data. The database was downloaded without names or personally identifiable information. Data were evaluated and analyzed from March 1 to June 30, 2024. The authors filtered records according to the eligibility criteria established. A cleaning of the database was performed to search for extreme data. No imputation was conducted in the case of missing data.

Analysis Plan

The database was evaluated in Microsoft Excel to identify extreme values, incomplete data, or inconsistencies. Subsequently, STATA® V18 software (Stata Corp, College Station, Texas, USA) was utilized for data analysis.

The Ankle-Brachial Index (ABI) calculation was performed in a spreadsheet, following the algorithm in Figure 1. This process was carried out in the following order: first, patients with peripheral arterial disease (PAD) were identified, then those with arterial calcification among the remaining patients were selected, and finally, the unclassified cases were considered normal. These three categories were established for each of the three methods: highest, lowest and average ABI criterion. The prevalence of PAD, arterial calcification, and normal for each method was calculated along with its 95% confidence interval.

The absolute and relative frequencies of categorical variables were described. For quantitative variables, the mean and standard deviation were calculated if the data met the criteria for normality. Otherwise, the median and interquartile range were reported. The Shapiro-Wilk test was applied to determine the normality of the data.

A description of the ABI categorized into PAD, normal, and arterial calcification was performed according to demographic, clinical, and laboratory characteristics. This was done for each of the three methods of ABI calculation: highest pressure criterion, lowest pressure criterion, and average pressure criterion.

Using Poisson regression with robust variance, the crude prevalence ratio (PR) was calculated between cases of PAD and arterial calcification concerning the normal category for the variables of age, education level, duration of diabetes, sex, BMI, history of ulcer, peripheral neuropathy, and glycemic control. A multivariate analysis was conducted to obtain the adjusted prevalence ratios (aPR). Adjustments were made for variables that showed a p-value less than 0.2. Variables that were part of the definition of PAD (absence of pulse) were not adjusted. A significance level of 5% was used for all analyses.

We calculated the sample size power of 643 subjects to estimate the prevalence of peripheral arterial disease (PAD) for each criterion using the power command in STATA® V18. We employed the formula for estimating a proportion. We considered expected PAD prevalences based on Schröder’s study of 3.7%, 5.4%, and 14.6% for the highest, average, and lowest ABI criteria, respectively. [27] Additionally, we accounted for the proportions obtained in our study of 7.8%, 15.4%, and 28.2%. By configuring a two-tailed analysis and a 95% confidence level, we achieved powers of 99%, 100%, and 100%, respectively.

Ethical Aspects

Approval was obtained from the Institutional Ethics and Research Committee of Hospital María Auxiliadora under code HMA/CIEI/015/2024. The data did not contain any personally identifiable information.

General Characteristics

The sample presented a mean age of 61.4 years, of which 69.8% were women. Forty percent of the patients had more than 10 years since their diabetes diagnosis, and 31.9% used insulin alone or in combination with oral antidiabetic drugs as treatment. Among the most important comorbidities, 33.3% had hypertension, 8.6% had a history of ulcer, and 1.4% had a previous coronary artery disease.

Additionally, 28.9% presented painless peripheral neuropathy, while 61.4% experienced neuropathic pain. Regarding metabolic control targets, only 22.4% had a glycated hemoglobin level below 7%. When evaluating the integrated target of glycated hemoglobin lower than 7%, LDL below 100 mg/dl, and systolic blood pressure below 140 mmHg, only 4.4% met these criteria (Table 1).

Prevalence of PAD and Calcification

Peripheral arterial disease (PAD) was present according to the criteria of high, low, and average ABI in 7.8%, 28.2%, and 15.4%, respectively. Arterial calcification was present in 18.2%, 16.2%, and 11% for the same criteria (Table 2).

Normal ABI values were observed in 74% of cases using the highest pressure, 73.6% with the average pressure, and 55.7% with the lowest pressure.

Supplementary Tables S1, S2, and S3 detail the absolute and relative frequencies of each clinical-demographic variable according to the ABI results for each of the three methods. Supplementary Table S4 describes the means of each quantitative variable according to the ABI results.

Characteristics Associated with PAD

For PAD, age was the only factor associated across all three methods. For the high ABI method, in subjects over 75 years old, PAD increased by 7.1 times compared to those under 60 years old (aPR 8.2; 95% CI 1.11 – 60.1). Patients with a history of ulcers showed nearly three times higher prevalence of PAD (aPR 2.88; 95% CI 1.11 – 7.49; P < 0.05). Conversely, the prevalence of PAD decreased by 88% in obese patients compared to those of normal weight (aPR 0.11; 95% CI 0.03 -0.49; p < 0.05). For the low ABI method, a significant association with older age was also observed; in those over 75 years, PAD increased by 81% compared to those under 60 years (RP 1.81; 95% CI 1.11 – 2.95), in conjunction with patients with chronic kidney disease where the prevalence of PAD was 55% (aPR 1.55; 95% CI 1.08 - 2.23; p < 0.05). For the average ABI, the prevalence increased by 1.87 times (aPR 2.87; 95% CI 1.28 – 6.43), and the presence of hypertension increased the prevalence of PAD by 87% compared to those without this history (aPR 1.87; 95% CI 1.12 - 3.13; p < 0.05) (Table 3).

Characteristics Associated with Arterial Calcification (AC)

In arterial calcification (AC), diabetes duration was the only factor that was associated across all three methods of analysis. In the high ABI method, individuals with more than 20 years of diabetes had an AC prevalence 2.79 times higher compared to those with less than 10 years (aPR 3.79; 95% CI 2.29 – 6.25); in the average ABI, the prevalence increased by 2.12 times (aPR 3.12; 95% CI 1.54 - 6.30; p < 0.001), and for the low ABI, it increased by 2.01 times (aPR 3.01; 95% CI 1.88 - 4.82; p < 0.001).

Regarding elevated BMI (obesity), in the average ABI, the prevalence of AC more than doubled compared to subjects with normal weight (aPR 2.26; 95% CI 1.01 - 5.04; p < 0.05). Similarly, it increased in the low ABI (aPR 2.05; 95% CI 1.21 - 3.45; p < 0.05). For the high ABI, there was a 38% increase, although it was not significant (aPR 1.38; 95% CI 0.76 - 2.50; p=0.287).

In patients with a history of ulcers, only in the low ABI, there was double the prevalence of AC compared to those without this history (aPR 2.01, 95% CI 1.24 - 3.25, p < 0.01). In the other two methods, there was only a 40% increase, which was not significant (Table 4).

Peripheral Arterial Disease

There is strong evidence that an ABI less than 0.9 is associated with up to twice the risk of cardiovascular or all-cause mortality in subjects with diabetes. Furthermore, it is noted for its better specificity than sensitivity when using the classic highest ABI criterion. However, considering different ABI cut-off points depending on the method used could optimize risk identification. Our study’s findings on PAD prevalence using various methods are like other studies. In a US study with a multiethnic population, a prevalence of 3.7% was found with the highest ABI, 5.4% for the average ABI, and 14.6% with the lowest ABI [27]. A study in France found a prevalence of 22%, 23%, and 29% for the three measurement modes in the same order [28]. The same US study showed that using different calculation methods can slightly improve prognostic accuracy [27].

On the other hand, a recent clinical guideline agrees on using the lowest ABI criterion for risk stratification and the highest ABI for severity assessment and post-revascularization follow-up [18]. The lowest ABI increases prevalence and sensitivity for identifying high- risk patients, though it decreases specificity for diagnosing those with early-stage disease. In practice, diagnosing asymptomatic PAD with an ABI below 0.9 would elevate cardiovascular risk to higher categories. Considering the European cardiovascular risk score for dyslipidemia management (SCORE), with every diabetic patient having at least moderate cardiovascular risk, the presence of asymptomatic PAD would elevate to a high or very high level and intensify risk factor treatment [29].

Regarding characteristics associated with PAD, our study found an association with advanced age for any of the three methods, and depending on the method, factors varied. Increased prevalence at advanced ages reflects the natural progression of atherosclerosis and vascular risk in patients chronically exposed to diabetes [30]. We found no association with gender; it has been reported that women present a similar risk as men, contrary to previous assertions, requiring similar importance in screening [31]. Another risk factor is ethnicity, where Afro-descendants have a higher likelihood. In our study, the original data did not contain this information, but it wouldn’t be a limitation given the high degree of mixed ethnicity in our country, unlike other South American countries [32].

Arterial Calcification

Traditionally, AC was considered a generalized passive process typical of advanced age and metabolic alterations; however, it has recently been discovered as an active and regulated process similar to bone metabolism and thus modifiable. It can occur at the intima or media layer, with both coexisting to different extents [33]. An ABI greater than 1.3 demonstrates the existence of medial or Monckeberg calcification in lower limbs, causing vascular stiffness, possibly with low flow, in an artery that cannot be compressed. It was long considered an artifact, requiring other methods like the toe-brachial index, as finger arteries are less susceptible to AC [34]. However, it has been highlighted as a critical cardiovascular risk marker in the general population [35] and in patients with diabetes [36].

In our study, the frequency of subjects with an ABI greater than 1.3 in any limb was determined to be 18.3%, lower than that reported in China conducted in the general population, using an average ABI method finding a frequency of 24.4%. However, as the diagnostic algorithm usually prioritizes PAD, AC may be underreported in those who coexist. In our study, only 2.2% presented both findings. It’s important to note that the recommended cut-off point in clinical guidelines is 1.3 [24] or 1.4 [18,37]. Where a highest CA cut-off values reduce prevalence, it becomes more specific. One cut-off point can be established based on epidemiological criteria such as patient risk, disease prevalence, or costs and infrastructure to provide treatment to positive results.[38]

The most studied factors associated with AC include diabetes mellitus, chronic kidney disease, advanced age, and increased creatinine [39]. In our study, AC was associated with longer diabetes duration for any method. Age was not significant but showed a trend towards being more frequent in younger people, suggesting that in certain patients, AC begins very early. This is why, with the low ABI method, AC is associated with higher BMI. Chronic kidney disease did not increase AC prevalence, despite being present in a quarter of those with data for its calculation.

Being a generalized process, arterial calcification (AC) can be assessed using other diagnostic strategies, both invasive (arterial angiography, intravascular ultrasound) and non-invasive (carotid intima-media thickness, brachial artery flow-mediated dilation, or coronary calcification score), which are currently used as risk markers [40].

Importance of Findings

The simultaneous identification of PAD and AC in diabetic patients implies a dual approach to preventing amputations and reducing the risk of cardiovascular events. The use of the highest ankle pressure as the numerator in the ABI calculation is associated with increased specificity, minimizing the risk of overdiagnosis in healthy individuals and preventing unnecessary testing and treatment. However, this approach may lead to a higher rate of false negatives, particularly in patients with arterial calcification.[37,38] Utilizing the lowest ankle pressure would identify more individuals at risk of mortality, both from cardiovascular and other causes. These findings support considering different cut-off points for the ABI, as the traditional method (highest pressure) might overlook individuals with significant risk.[41]

Furthermore, it is necessary to report AC as an additional risk factor in regular screening to identify them early, and not only view it as a technical issue requiring additional imaging techniques to determine PAD. It would also be essential to locally validate its association with hard outcomes such as cardiovascular disease, major amputation or frailty. [42]

Limitations and Strengths of the Study

The use of a hospital-based study design may limit the generalizability of the findings, as the selected sample does not necessarily represent the entire population of type 2 diabetes mellitus patients in different care settings. This could affect the extrapolation of results to other patient groups outside the hospital setting. Additionally, the initial focus on determining PAD might have led to underreporting of AC, given that if a patient had PAD in one leg and AC in the other, PAD would take precedence. Moreover, nearly 40% of initial records had to be excluded due to incomplete ABIs; however, the number of included subjects provides good precision for estimating prevalences. Similarly, there was insufficient data on past medical history, body mass index, glycated hemoglobin, and chronic kidney disease, so associations may be limited by sample size. Some variables such as duration of diabetes may present as recall bias, and conditions like diabetic retinopathy may be underreported as they came from self-report and not active searches. A strength is that the ABI was performed by personnel trained in diabetic foot care and with a mercury sphygmomanometer, and peripheral neuropathy data were obtained through objective tests.

Conclusions

Using different ABI methods, we observed a prevalence of PAD ranging from 7.8% to 28%, and a prevalence of CA between 11% and 18% among patients with type 2 diabetes mellitus at a public hospital in Peru. The method with the lowest criterion showed the highest prevalence. It is more frequent with increasing age and longer duration of diabetes for any of the three methods. Further research is necessary to verify if the new ABI calculation methods provide improved accuracy in predicting complications