Ambulatory Arterial Stiffness Index: Regression Method Comparison and Its Association With Pulse Pressure and Circadian Patterns

ABSTRACT

The Arterial Stiffness Index (AASI) is a calculation obtained through Ambulatory Blood Pressure Monitoring (ABPM), and is an indirect measure of the elastic properties of the arterial wall; but there is heterogeneity in its scope as a predictor of vascular wall health. A comparison is made between linear regression and exponential regression of the AASI, as well as an analysis of variance, according to circadian patterns and pulse pressure (PP) values. This work is an analytical observational study in 106 individuals, most of them women (63%) with a mean age of 53 ± 17.32 years. The coefficient of determination (r2) for the linear relationship was 0.53 ± 0.17, similar to the exponential relationship with an r ² of 0.52 ± 0.17 (p = 0.7032). Patients with PP < 52 mmHg had an AASI of 0.3839 ± 0.1428 and for PP > 53 mmHg an AASI of 0.5330 ± 0.1108 (p < 0.0001). When comparing the AASI between Dipper vs. Riser circadian patterns, there was homoscedasticity (p = 0.3717); on the contrary, in the intergroup evaluation with Non‐Dippers, heteroscedasticity was observed (Dipper vs. Non‐Dipper; p = 0.0316 and Non‐Dipper vs. Riser; p = 0.01978). This study concludes that the best determination of AASI is linear regression, robustly correlating with the values of PP > 53 mmHg and AASI > 0.5 (r = 0.9628). The behavior of the data in the Non‐Dipper group is heterogeneous, probably due to their own physiological characteristics. In addition, AASI could be an indirect measure of arterial stiffness and be more directly associated with arterial elasticity and its deformation capacity.

1. Introduction

High blood pressure is a pathology with a high impact on global public health, with a great influence on morbidity and mortality rates. Worldwide, there were around 1390 million hypertensive patients in 2010, with a great disparity in prevalence depending on the income level of each region [1]. High blood pressure is a highly prevalent condition in the global and national population. According to the Venezuelan Study of Cardiometabolic Health (EVESCAM), conducted by the Venezuelan Society of Internal Medicine in 2017, it was estimated that more than a third of the Venezuelan population was hypertensive [2].

With the evolution of the different cohorts of the Framingham study, additional parameters that were previously considered physiological and even essential in hemodynamic functioning were evaluated. Among the first observations, the trend toward isolated systolic hypertension stood out, with a higher prevalence in older adults [3, 4, 5].

Based on these findings, arterial stiffness is evaluated as a cardiovascular risk factor to be considered. However, the difficulties in obtaining reliable data on the arterial wall, with simple and reproducible clinical methods within the reach of physicians, have hindered the progress toward possible therapeutic approaches to this condition [6, 7].

Therefore, being able to use a methodology that allows us to emulate the physiological phenomena involved in arterial stiffness is, at least, a goal of great interest for the clinical physician. Ambulatory blood pressure monitoring (ABPM) is a simple and reproducible method that is readily available to clinicians and which in recent years has been included in various international guidelines for the diagnosis and treatment of arterial hypertension [8].

In 2006, Li et al. introduced the Ambulatory Arterial Stiffness Index (AASI), defined as one minus the slope of a linear regression of diastolic blood pressure (DBP) data on systolic blood pressure (SBP) data derived from ABPM. They hypothesized that AASI might reflect arterial stiffness [9]. Evidence from the last decade supports the notion that AASI is capable of indirect measures of the elastic properties of the arterial wall and its changes. In different trials, AASI has been reported to be a predictor of target organ damage, cardiovascular mortality, and cerebrovascular events, even better than pulse pressure (PP) [3, 6, 7, 10].

Palencia et al. conducted a study in which the different parameters of the ABPM were related to the AASI. They found that AASI was positively correlated with PP, load, age, and SBP. However, it is important to note that the relationship obtained between these parameters was moderate, at best [11].

Gavish et al. argue that the regression method used to determine the slope and the AASI may be inappropriate due to an overestimation in its calculation, which conceals its true dependency on clinical characteristics. Additionally, Schillaci et al. examine AASI's apparent dependency on the nocturnal reduction in blood pressure, pointing out that a narrow variation in diastolic values—common among Non‐Dipper patients—flattens the regression slope and artificially raises the AASI. Both perspectives converge in suggesting that the traditional method of calculating AASI makes it vulnerable to influences not directly related to arterial stiffness, potentially limiting its actual predictive value [12, 13, 14].

Considering that the arterial pressure‐diameter curve (p‐d) is nonlinear (it follows an exponential form); and that the greater the arterial stiffness (for a higher AASI value), the more pronounced the nonlinearity of p‐d should be, it could be inferred that the relationship in the different parameters of ABPM could be nonlinear [10]. To date, no studies have addressed whether the most appropriate regression analysis for AASI should differ from the linear model, beyond examining the influence of confounding clinical variables. This study performs a comparative analysis of linear and exponential regression models between diastolic and systolic blood pressure (DBP and SBP), aiming to determine whether the traditional linear fit used to calculate AASI is appropriate. The underlying hypothesis is that exponential regression may offer a more accurate representation of the non‐linear physiological relationship between DBP and SBP, thereby, leading to a more reliable AASI as an index of arterial stiffness.

2. Methods

2.1. Study Population

This is an observational and analytical study carried out in a population consisting of 1042 patients, all over 18 years of age, who have been entered into the database of the Research Group of the Chair of Clinical Medicine and Therapeutics A of the Luis Razetti School ‐ Central University of Venezuela, until June 1, 2023, who have undergone an Ambulatory Blood Pressure Monitoring study, whether hypertensive or not.

The sample was calculated taking into account a confidence level of 95% with a margin of error of up to 10%, obtaining a minimum necessary population of 89 individuals. The selection was carried out using a simple random sampling.

Exclusion criteria included patients with incomplete clinical data or ambulatory blood pressure recordings that did not meet analytical requirements—specifically, measurement intervals of every 15 min during daytime and every 20 min during sleep, with more than 70% successful readings.

As a study using a database, in compliance with the Declaration of Helsinki complemented by the Declaration of Taipei; the original study [11] was approved by the Bioethics Committee of the Military Hospital of Caracas. All participants gave their informed written consent.

2.2. Procedures

The database of the Research Group of the Chair of Clinical Medicine and Therapeutics A of the Luis Razetti School was obtained by collecting data from patients who agreed to be included. Data were obtained through different. Ambulatory blood pressure data were collected using three validated devices: the BR‐102 Plus (Schiller), which meets the European Society of Hypertension International Protocol (ESH‐IP 2002); the Welch Allyn ABPM 6100S, compliant with the AAMI SP10 ES1 standard; and the SunTech Bravo, which fulfills validation criteria established by the International Protocol of the European Society of Hypertension (ESH), the British Hypertension Society (BHS), and the ISO 81060‐2 standard.

Simple random sampling was performed until at least the calculated sample was obtained. Data were extracted using an instrument designed for this purpose, to obtain the necessary basic demographic and clinical data such as sex, age, average systolic and DBP in 24 h, diagnosis of hypertension, smoking habit, diagnosis of diabetes, pregnancy, type of antihypertensive treatment (monotherapy, combination therapy, and groups of medications).

Additionally, circadian blood pressure patterns were extracted from the ABMP data by calculating changes in SBP during awake hours compared to sleeping hours (Dipper between 10% and 20%, Over Dipper > 20%, Non‐Dipper < 10%, Riser < 0%). The average pulse pressure data for 24 h were also obtained.

To obtain AASI, we fitted an ordinary least squares linear regression of diastolic blood pressure (DBP, Y axis) on systolic blood pressure (SBP, X axis) for each participant and estimated the slope (m) with an unconstrained intercept; AASI was defined as 1‐m. We compared r2 from linear and exponential fits solely to describe goodness of fit.

2.3. Statistical Treatment

For data processing and analysis, the collected measurements were initially organized into a matrix using Microsoft Excel 365. Subsequently, statistical analyses were conducted using the XRealStatistics add‐in for Microsoft Excel, part of the Real Statistics Resource Pack. For nominal variables, they were expressed in frequencies and percentages. For discrete and continuous variables, estimates were made based on the central limit theorem to determine a distribution like the normal. The arithmetic means were calculated with the standard deviation. Normality of residuals was assessed with the Shapiro–Wilk test. ANOVA calculations were performed for different study groups to determine the difference in variance. Homogeneity of variances across circadian groups was evaluated using Levene's test. Homogeneity of regression slopes was tested by including interaction terms between each covariate and circadian pattern. The Student's T‐test was also used to compare groups with similar variances. Pearson's correlation calculation was used to estimate the linear relationship. The coefficient of determination was calculated to compare the linear correlation with the exponential correlation. A multivariate regression analysis was performed to adjust for confounding variables. A significance level of p < 0.05 was considered statistically significant for all tests conducted.

Box and Whisker graphs were made to represent the behavior of the distribution and variability of the data.

3. Results

3.1. Clinical and Epidemiological Characteristics

From a total of 1042 ABPM records in the database, 195 individuals were selected through simple random sampling. Of these, 28 were excluded due to incomplete clinical records and 61 due to ABPM recordings that did not meet the predefined quality criteria (measurement intervals and valid readings 70%). The final analytic sample consisted of 106 participants who fulfilled all inclusion criteria (Figure 1). The majority were female (63%), the mean age was 53 ± 17.32 years. The mean 24‐hour SBP was 122 ± 12.88 mmHg and the mean 24‐hour DBP was 73 ± 9.12 mmHg. Of this group of patients, at the time of ABPM, 67% of the individuals were known to have a diagnosis of hypertension. Of the total of 106 patients, 62% were under antihypertensive pharmacological treatment, and 5 patients (5%) did not follow the therapeutic regimen at the time of ABPM (Table 1).

FIGURE 1.

Study flowchart—screening, eligibility, exclusions, and final sample (n = 106).

TABLE 1.

Epidemiological and clinical data of patients evaluated through Ambulatory Blood Pressure Monitoring and Adjusted Associations with ASSI.

Characteristics	Value	Multivariate regression coefficient β	p (95% CI)
Total (n)	106
Gender
Male, n (%)	41 (37)	0.08 (0.017–0.14)	0.012
Female, n (%)	65 (63)
Age (years) (X ± SD)	53 ± 17.32	0.002 (0.0–0.04)	0.051
Systolic blood pressure (mmHg ± SD)	122 ± 12.88	0.006 (0.003–0.009)	<0.001
Diastolic blood pressure (mmHg ± SD)	73 ± 9.12	−0.006 (−0.01–0.02)	<0.01
Hypertension diagnosis
Hypertensive, n (%)	71 (67)	0.034 (−0.03–0.097)	0.278
Non‐hypertensive, n (%)	35 (33)
Smoking habit
Yes, n (%)	11 (10)	−0.02 (−0.2–0.07)	0.624
No, n (%)	95 (90)
Diabetes, n (%)	14 (15)	0.08 (−0.09–0.18)	0.077
Pregnant women, n (%)	13 (12)	0.47 (−0.06–0.15)	0.385
Without antihypertensive treatment, n (%)	40 (38)
Antihypertensive treatment, n (%)	66 (62)
Monotherapy, n (%)	37 (56)
Combination therapy, n (%)	29 (44)
ACEI or ARB, n (%)	49 (74)
Thiazides, n (%)	15 (23)
Calcium channel blockers, n (%)	23 (25)	—	—
β‐blockers, n (%)	20 (30)
α‐agonists, n (%)	4 (6)

Note: Multivariate linear regression model with AASI as dependent variable. All coefficients adjusted for age, sex, blood pressure, hypertension diagnosis, smoking habit, diabetes, and pregnant women.

Abbreviations: ACEI, angiotensin‐converting enzyme inhibitors; ARB, angiotensin receptor blockers.

The AASI value was calculated according to sex and it was observed that for the female group it is lower than for the male group. For the female group it was 0.3828 ± 0.1727 and for the males it was 0.4784 ± 0.1453 (p = 0.0043). In the case of the hypertensive group, higher AASI values were observed with 0.4381 ± 0.1631 compared to the non‐hypertensive group with 0.3826 ± 0.1750, however, it can be considered a result of chance due to the value of p = 0.1139 (Table 2).

TABLE 2.

Ambulatory Arterial Stiffness Index by gender and hypertension diagnosis.

Characteristics	AASI (X ± SD)	p
Gender
Male, n (%)	0.4784 ± 0.1453	0.0043
Female, n (%)	0.3828 ± 0.1727
Hypertension diagnosis
Hypertensive, n (%)	0.4381 ± 0.1631	0.1139
Non‐hypertensive, n (%)	0.3826 ± 0.1750

Abbreviation: AASI, Ambulatory Arterial Stiffness Index.

Level of significance p < 0.05.

3.2. Linear and Exponential Regression

Linear and exponential trend curves were drawn considering the dispersion graphs of each individual for all diastolic pressure measurements (Y axis) in all systolic pressure measurements (X axis). The determination index for each regression was calculated with the intention of comparing the difference in variance of each case and evaluating which trend best explains the dispersion of the data. Figure 2 shows that the linear relationship had an average determination coefficient of 0.53 ± 0.17, similar to the exponential relationship with an average determination coefficient of 0.52 ± 0.17 (p = 0.7032), which makes the behavior of both correlations similar with respect to variance.

FIGURE 2.

Comparison of the coefficient of determination (r ²) between the linear relationship vs. the exponential relationship of the Ambulatory Stiffness Index (AASI).

3.3. Pulse Pressure and AASI

When relating the AASI calculation to the average pulse pressure (PP) obtained in 24 h, the results were grouped by quartiles: first quartile < 45 mmHg, second quartile 46–52 mmHg, third quartile 53–60 mmHg, and fourth quartile > 61 mmHg. A progressive increase in the AASI value was evident, proportional to the increase in pulse pressure, for each recorded quartile.

Pearson's correlation coefficient was r = 0.9628. ANOVA results comparing variance between the first and second quartiles yielded p = 0.1305, and between the third and fourth quartiles p = 0.2569. T‐tests for equal variances resulted in p = 0.5045 and p = 0.4034, respectively (Table 3).

TABLE 3.

Ambulatory Arterial Stiffness Index according to the quartiles and two groups of the average pulse pressure in 24 h.

Pulse pressure (mmHg)	AASI (X ± SD)	ANOVA	T‐test	Pearson
		p	p	r
According to quartiles
<45	0.3734 ± 0.1503	0.1305	0.5045
46–52	0.3952 ± 0.1375			0.9628
53–60	0.5151 ± 0.1234	0.2569	0.4034
>61	0.5523 ± 0.1019
According to two groups
<52	0.3839 ± 0.1428	0.0836	<0.0001	—
>53	0.5330 ± 0.1108

Abbreviation: AASI, Ambulatory Arterial Stiffness Index.

Level of significance p < 0.05.

A sub‐analysis of the groups is performed according to pulse, dividing it into two samples: <52 mmHg and >53 mmHg. For the <52 mmHg group, an average AASI of 0.3839 ± 0.1428 is obtained, and for the >53 mmHg group, there is an average AASI of 0.5330 ± 0.1108. For this case, the variance calculation (ANOVA) obtained was p = 0.0836, but a T‐test analysis for equal variances shows p < 0.0001 (Table 3).

3.4. Circadian patterns and AASI

The different circadian patterns of SBP were determined according to the change in the relationship between blood pressure values during awake hours and the period of night‐time sleep. Three groups were used: Dipper (decrease in 10%–20%), non‐Dipper (decrease in < 10%) and Riser (nocturnal increase in SBP). There was only one patient with an over‐Dipper pattern (decrease > 20%), who was removed from this data.

When grouping the circadian pattern groups, an AASI of 0.3377 ± 0.1156 was obtained for the Dipper group; for the Non‐Dipper group, the AASI was 0.4324 ± 0.1611; and for the Riser subgroup, an AASI of 0.5013 ± 0.1077 was observed. One‐way ANOVA showed a significant difference between group means (unadjusted ANOVA p = 0.00015). After adjusting for sex and age in ANCOVA models, global p values were p = 0.361962 (adjusted for sex) and p = 0.00032 (adjusted for age) (Table 4).

TABLE 4.

Ambulatory Arterial Stiffness Index (AASI) by Circadian Pattern: ANOVA (unadjusted) and ANCOVA (adjusted).

Circadian patterns	n (%)	AASI (X ± SD)	Variance	p value	p value	p value	p value
				Unadjusted	Adjusted	Adjusted for sex	Adjusted for age
Dipper	30 (28)	0.3377 ± 0.1156	0.0138
Non‐Dipper	51 (48)	0.4324 ± 0.1611	0.0265	0.00015	<0.0001	0.361962	0.00032
Riser	25 (24)	0.5013 ± 0.1077	0.0120

Note: ANCOVA models: AASI was the dependent variable and circadian pattern (Dipper, Non‐Dipper, and Riser) the fixed factor. Age was included as a continuous covariate and sex was included as a binary covariate coded 0 = female, 1 = male. The “Variance” column reports the within‐group variance used in the one‐way ANOVA. Normality of residuals was assessed with the Shapiro–Wilk test. Homogeneity of regression slopes was tested by adding interaction terms between each covariate and circadian pattern; interaction p values were age × pattern = 0.9176 and sex × pattern = 0.7593. Levene's test for equality of variances yielded p = 0.07042. The p values for age and sex correspond to the effect of each covariate on AASI within the ANCOVA model; the global p value refers to the adjusted comparison between the groups of circadian pattern groups.

Abbreviation: AASI, Ambulatory Arterial Stiffness Index.

Level of significance p < 0.05.

In unadjusted comparisons (Table 2), men showed higher AASI values than women. However, when age and circadian pattern were included as covariates in ANCOVA, sex was no longer a significant predictor, and the sex × pattern interaction was not significant, indicating that the relationship between circadian pattern and AASI was consistent between men and women. Importantly, in the full ANCOVA model that simultaneously includes age and sex, the global effect of the circadian pattern on AASI remained statistically significant (p < 0.0001), confirming that the observed differences between the circadian groups are robust after multivariate adjustment (Table 4).

The results were compared intergroup, regrouping them as: Dipper/Non‐Dipper, Dipper/Riser, and Non‐Dipper/Riser, to assess the consistency of the AASI differences across circadian patterns. When comparing the Dipper and Riser subgroups, the variance comparison yielded a non‐significant result (p = 0.3717), while the T‐test indicated a significant difference (p < 0.0001) (Table 5).

TABLE 5.

Analysis of variance (ANOVA) for two samples of the Ambulatory Arterial Stiffness Index, comparing the circadian patterns of systolic blood pressure between sleep and awake hours.

Circadian patterns	ANOVA	T‐test
	p	p
Dipper/non‐Dipper	0.0316	—
Dipper/Riser	0.3717	<0.0001
Non‐Dipper/Riser	0.0198	—

Level of significance p < 0.05.

To evaluate these results, a box and whisker graph was created. The graph depicts the behavior of the Dipper and Riser groups, as well as the Non‐Dipper group, showing a greater dispersion of data in the latter (Figure 3).

FIGURE 3.

Ambulatory Arterial Stiffness Index according to the change in the circadian pattern of systolic blood pressure between sleep and awake hours.

4. Discussion

The AASI is an independent cardiovascular risk marker, capable of predicting cardiovascular mortality and the possibility of cerebrovascular events and even damage to target organs; with excellent reproducibility, which makes it an efficient and recommended calculation for all hypertensive patients [15].

In the search for data to improve AASI prediction as a useful tool in detecting cardiovascular risk, this work obtained results that could contribute to clarifying the circumstances that may influence the arterial stiffness values obtained by ABPM.

The predominance of female participants in ABPM use may reflect gender‐based differences in health‐seeking behavior, as observed in similar regional studies. Examples of this were the studies by Raimondo et al., Bahrainwala et al., and Mortazavi et al., where women participated in more than 50% of the monitoring [13, 14, 15]. However, this trend does not necessarily correlate with arterial stiffness, suggesting a need for stratified analysis in future research [16, 17, 18].

This is consistent with most studies on arterial stiffness, Said et al., Park et al., Lu et al., and even Boos et al. It is proposed that this situation is explained by the influence of estradiol and follicle‐stimulating hormone levels on arterial tree compliance, since this difference decreases until it equalizes in postmenopause [19, 20, 21, 22].

Contrary to expectations, our data did not reveal significant differences in AASI between hypertensive and normotensive individuals. Many trials that evaluate arterial stiffness using ABPM have not considered in their analysis a direct comparison of groups of hypertensive patients vs. non‐hypertensive patients. It has only been mentioned that the dispersion in the data in obtaining the AASI increases as the patient presents more cardiovascular risk factors (e.g., age, circadian pattern). In fact, the study by Raimondo et al. based on this phenomenon proposes that AASI is more reliable in demonstrating vascular health than in accurately determining the degree of vascular wall compromise [16]. However, Palencia Tejedor et al. demonstrated how AASI, as well as the level of brain natriuretic peptide (BNP), decrease with the use of hypertensive treatment [23]. In our case, 62% of the patients underwent some antihypertensive treatment (56% in monotherapy), which could influence the fact that there is no significant difference between AASI of hypertensive and normotensive patients.

Among the findings of the study by Valero et al., it was observed that the AASI decreased because there was less dispersion of the data in the regression line [10]. In other words, as the coefficient of determination (r2) increased, the AASI tended to be lower; and vice versa. The lower the dispersion of the data (higher r2), the lower the AASI tended to be; that is, an artery with better compliance or less rigidity would provide blood pressure values with less dispersion. This observation could be related to the hypothesis of Di Raimondo et al., when they suggested that AASI is better for determining good vascular health [16].

Taking these observations into account, linear and exponential relationships were compared in each of the 106 patients to check which type of regression best fits the dispersion of the data. It was shown that, for both cases, the coefficient of determination was quite similar. Thus, given that linear regression is mathematically simpler for calculating AASI, this should continue to be used. It should be noted that the average of these determination coefficients was not greater than 0.53. Although this demonstrates the dependence of diastolic pressure on systolic pressure, it is a moderate goodness of fit that could be associated with variance between groups of patients.

With the description of the AASI, it was assumed that there should be a direct correlation between the AASI and pulse pressure. Several studies have determined this association; for example, the study by Palencia‐Tejedor et al. demonstrated this association with a correlation coefficient of 0.49, which, although positive, is moderate [11]. Kollias et al. presented similar results after evaluating 51 trials with 29 186 patients; they obtained a correlation coefficient of 0.47 CI 95% (0.40–0.54), which can also be considered moderate, but the results among all included trials were heterogeneous (I2 93%, p < 0.001) [24].

Kollias et al., Di Raimondo et al., and Said et al. concluded that AASI and pulse pressure are independent variables as predictors of cardiovascular events [16, 19, 24]. In our case, pulse pressure groups were established in quartiles. In this way, the correlation coefficient between AASI and pulse pressure would show a strong linear relationship between the two parameters. According to pulse pressure, and it was obtained that at values greater than 53 mmHg the AASI rises above 0.50; which clarifies the direct relationship that may exist between these variables, as long as it is with pathological values; which further confirms the findings of Verdecchia et al., as an adequate cut‐off point for the abnormal pulse pressure value obtained by ABPM [25].

Regarding the circadian pattern, Jerrard‐Dunne et al. already warned, in an analysis adjusted for age and sex, that the smaller the drop in SBP during sleep, the greater the velocity of the pulse wave (or the higher arterial stiffness) [26]. That is, if we classify nocturnal dipping on a severity scale, recognizing the Dipper patient as normal, the non‐Dipper as abnormal (but moderately pathological) and the Riser patient as extremely pathological. It could be assumed that the greatest arterial stiffness would be obtained in a patient with a more severe directly proportional circadian pattern disorder. However, in the same trial, in a multivariate analysis, there was no difference between the Dipper and non‐Dipper groups, while the Riser group maintained high arterial stiffness.

Similarly, Di Raimondo et al. in their initial analysis present that the AASI value could be proportionally elevated to the severity of nocturnal dip alteration [16]. But the latter changes when performing a multivariate analysis (adjusted for age, sex, body mass index, and average systolic and DBP), showing that there was a significant relationship between the AASI value and Dipper and extreme Dipper patients. However, this relationship was lost in non‐Dipper and Riser cases. Finally, the study by Boos et al. also observed that the AASI value was elevated according to the alteration of the nocturnal dip (AASI in Dipper 0.39, AASI in non‐Dipper 0.48 and AASI in Riser 0.56) [22].

In our case, as in all the studies mentioned, an initial analysis shows that the AASI rises as the type of dip during sleep worsens, with significant results when performing analysis of variance. However, upon closer inspection, it was evident that the Dipper and Riser pattern groups have a similar data dispersion, with equal variances; unlike the non‐Dipper group where the variance changes significantly, and the dispersion is much wider, which makes it a group with different behavior from the others. This could well explain the findings in other different studies that have addressed the relationship between the circadian pattern and arterial stiffness, where there has been an emphasis on comparing groups with dissimilar nocturnal behavior, which merits an independent analysis, especially the non‐Dipper group [17, 27].

The crude difference in AASI by sex observed in unadjusted analyzes should be interpreted with caution. This apparent disparity does not persist once age and circadian pattern are taken into account, indicating that sex alone is not an independent determinant of AASI in this cohort when evaluated within the context of circadian variation. This reinforces the notion that the circadian pattern is the primary driver of differences in AASI. From a clinical perspective, this underscores the importance of interpreting arterial stiffness within the larger framework of patient characteristics, rather than attributing crude differences exclusively to sex.

Mortazavi et al. made a similar observation studying only the group of non‐Dipper patients. They found how the prevalence of this circadian phenomenon varied using different methods to determine the effective sleep period (schedules set by the ABPM team, sleep schedules reported by the patient, or by actigraphy) [18]. The non‐Dipper group should be evaluated independently, as it is a group of patients with different variance, and therefore merits multivariate analysis directed only at this group, rather than comparing it with the findings of Dipper, extreme Dipper, or Riser patients.

In closing, the AASI, although a valuable resource for determining cardiovascular risk, still presents findings that can generate discussion. These findings, as observed in our results, may be influenced by unaccounted hemodynamic phenomena that could affect the dispersion of data across different studies on the topic.

5. Conclusion

The Ambulatory Arterial Stiffness Index (AASI) is a reliable and reproducible marker to assess cardiovascular risk, and its routine use in ABPM is recommended. It has been observed that the AASI value tends to be lower in women, likely due to hormonal variations, while no significant differences were found between hypertensive and non‐hypertensive individuals, possibly due to the use of antihypertensive medications. Furthermore, when analyzing the relationship between diastolic and SBP, it is concluded that a linear relationship remains the most practical for calculating the AASI. In ABPM, a pulse pressure cutoff of 53 mmHg correlates with AASI values above 0.5.

Circadian timing also influences AASI, with low values in “Dipper” patients and high values in those with nocturnal blood pressure increases (“Risers”). Non‐Dippers show great variability, suggesting the need for more specific studies to understand their characteristics. The divergence of results in different studies on AASI indicates the importance of exploring variables not previously considered, including the influence of arterial strain pressure on measurements. Finally, the AASI could indirectly reflect arterial stiffness, being more closely linked to vascular elasticity and deformation capacity.

6. Future Work

Histologically, the elastin/collagen ratio of the arterial wall is known to determine the degree of stiffness that the vessel may present. Therefore, it may be surprising how antihypertensive treatment could modify this relationship in the short term, as there are trials in which the AASI value decreases proportionally with the use of these medications [11, 16]. This could raise further doubts as to whether the AASI only measures the degree of arterial stiffness [23, 28].

We usually assume that blood pressure measurement reflects the tension exerted by the arterial wall, through the pulse wave, either by the auscultatory method or the oscillometric method; and that the amplitude of the wave corresponds to the systole or diastole (in their equivalent value recorded in millimeters of mercury (mmHg)). However, there are phenomena of blood fluid dynamics and solid mechanics of blood vessels that, if considered, could influence the explanation of the findings of the present study and of previous work, as well as could explain some inconsistencies, which deserve to be analyzed and formulated new hypotheses.

It is quite clear that the sphygmomanometer cuff must exert pressure on the artery to deform it with the intention of decreasing the diameter in order to reduce and prevent the flow that would subsequently be gradually released and thus obtain the corresponding blood pressure records. It is clear that the sphygmomanometer cuff must apply pressure to the artery, reducing its diameter (or flow area) to temporarily stop blood flow. The pressure is then gradually released, allowing blood flow to resume, which enables blood pressure readings to be obtained. In order to achieve this temporary stop of the blood flow, it is necessary to overcome two forces or pressures antagonistic to the cuff pressure. One of them is the pressure exerted by the blood flow on the vessel walls, which is a variable influenced by its diameter (p‐d curves). The other is the pressure exerted by the vessel wall itself (this is affected by the rigidity or elasticity of the material). The cuff must oppose both pressures to achieve the necessary deformation to occlude the lumen of the artery.

All of the above implies that the pressure of the sphygmomanometer cuff (P_cuff ) must be at least equivalent to the sum of the blood flow pressure (P_{fluid / flow} ) plus the pressure to deform the artery (P_deformation ), and therefore this product, in mmHg, must be greater than each of its variables to achieve the phenomenon of arterial occlusion [28]. The second part of this study will look at the ABPM data under these lenses and how these pressures are related to the AASI.

Ethics Statement

As a study utilizing a database, in compliance with the Declaration of Helsinki, complemented by the Declaration of Taipei, the original study [11] was approved by the Bioethics Committee of the Military Hospital of Caracas. All participants gave their informed written consent. The present analysis was conducted on anonymized data derived from that study.

Consent

In the original study, all participants provided informed written consent. Given that this study is a retrospective analysis of the anonymized database, the requirement for additional patient consent was fulfilled in the original protocol.

Conflicts of Interest

The authors declare no conflicts of interest.

Permission to Reproduce Material From Other Sources

Not applicable, as no material from external sources was reproduced in this study.

Ayala‐Hernández J. R., López‐Sánchez C., Ayala‐Hernández O. M., and Palencia‐Tejedor C. E., “Ambulatory Arterial Stiffness Index: Regression Method Comparison and Its Association With Pulse Pressure and Circadian Patterns.” The Journal of Clinical Hypertension 27, no. 12 (2025): e70191. 10.1111/jch.70191

Data Availability Statement

The data supporting the findings of this study are openly available on DSpace at http://hdl.handle.net/10872/23236.