The relation between estimated pulse wave velocity and peripheral arterial disease: A study based on NHANES

Abstract

Background

As a type of atherosclerotic lesion affecting the peripheral vascular system, peripheral arterial disease (PAD) has an elevated disability rate and seriously affects patients’ quality of life. Estimated pulse wave velocity (ePWV) is an alternative indicator of pulse wave velocity and is closely linked with atherosclerosis. However, the link between ePWV and PAD is unelucidated.

Objectives

This research was undertaken to dissect the linkage between ePWV and PAD.

Methods

This project enrolled 6250 participants in the NHANES between 1999 and 2004. The linkage between ePWV and PAD and its possible influencing factors were explored by constructing a weighted logistics regression model and subgroup analysis. In addition, the threshold effect analysis and restricted cubic spline (RCS) model were utilized to evaluate the non-linear link.

Results

The weighted logistic regression model demonstrated a great positive linkage between ePWV and PAD risk in the fully adjusted model (OR = 1.29, 95% CI: 1.18–1.42, p < .001). There was a significant nonlinear link between ePWV and PAD. When ePWV was higher than 9.297, the risk of PAD increased significantly (OR = 1.03, 95% CI: 1.02–1.04, p < .001). No significant linkage was detected when the value was below this threshold. Additionally, the subgroup analysis further demonstrated that ePWV had a stronger positive link with PAD in the elderly population (age ≥60 years) and in people with BMI ≤30.

Conclusion

ePWV is an effective predictor of PAD risk especially in the elderly and non-obese population.

Keywords

Peripheral arterial disease estimated pulse wave velocity national health and nutrition examination survey

Introduction

Peripheral artery disease (PAD) is an atherosclerotic lesion affecting the peripheral vascular system, which can lead to stenosis or occlusion of lower limb arteries. Due to insufficient blood supply to the legs, patients often have intermittent claudication, ischemic pain, and dysfunction.¹ Severe ischemia in the lower limb can lead to amputation and even death.² From 1990 to 2019, the total number of PAD patients worldwide has greatly climbed. As of 2019, the number of people aged 40 and above with PAD globally is 113 million.³ Due to the high disability and the rising incidence of PAD, PAD has imposed a heavy burden of disease on patients and society. Therefore, early identification and management of PAD is the key to reducing its adverse prognosis.

Pulse wave velocity (PWV) is the velocity at which a pulse wave is generated by a heartbeat in an artery.⁴ PWV is tightly linked to the mechanical properties of the arterial wall, and an elevation in PWV usually indicates that the arterial wall becomes stiff and less elastic, thus being considered a marker of arterial stiffness.⁵ Despite the huge potential of PWV in preventing and controlling hypertension in cardiovascular disease (CVD), its application in clinical practice is limited,⁶ thus restricting its use in large-scale observational studies. The estimated pulse wave velocity (ePWV) is measured based on blood pressure and age, which is easy to obtain and can be a substitute index for PWV.⁷ ePWV has great bearings on vascular aging,⁸ heart failure,⁹ and risk of death.¹⁰ However, its predictive role in PAD is not illuminated.

Therefore, this research utilized a large sample of data from the National Health and Nutrition Examination Survey (NHANES) to dig out the potential value of ePWV in predicting the risk of PAD, aiming to proffer a theoretical foundation for future clinical applications.

Methods

Study population

Conducted by the Centers for Disease Control and Prevention and the National Center for Health Statistics in the US, NHANES is a cross-health sectional, population-based survey for assessment of the nutritional and health status of adults and children in the non-institutionalized population. We investigated the link between ePWV and PAD using survey data from three cycles from 1999 to 2004 (n = 31,126). After excluding missing and invalid PAD data (n = 23,576), missing and invalid ePWV data (n = 209), and covariate data with missing covariate data (n = 1091), a total of 6250 people were enrolled in our analysis (Figure 1).

Figure 1.

Flowchart of screening NHANES participants.

Independent variable

ePWV was measured based on mean blood pressure (MBP) and age: ePWV = 9.587-0.402 × age +4.560 × 0.001 × age²-2.621 × 0.00001 × age² × MBP+ 3.176 × 0.001 × age × MBP −1.832 × 0.01 × MBP. MBP = diastolic blood pressure (DBP) + 0.4 × (systolic blood pressure (SBP) - DBP).⁷ With the use of a mercury sphygmomanometer, trained NHANES staff took blood pressure (SBP and DBP) measurements 3 or even 4 times in the mobile examination center or home examination.

Dependent variable

The dependent variable was defined as PAD using the ankle-brachial index (ABI). For participants without bilateral amputations and weighing ≤ 400 pounds, SBP measurements were taken using blood pressure cuffs on the right brachial artery and 2 posterior tibial arteries. When participants right arm blood pressure could not be measured because of specific physical conditions, the left arm was measured. Two measurements were performed and averaged for participants aged 40–59 years. Only one measurement was performed for participants aged ≥60 years. The calculation of the ABIs on the right and left sides was the ratio of the average systolic pressure in the posterior tibial artery to the average systolic pressure in the arm. The smaller ABI was utilized. Participants with ABI ≥1.5 were likely to develop severe arterial stiffness and were therefore excluded. If patients had an ABI <0.9, PAD was defined.^11,12

Variables

The covariates included body mass index (BMI), gender, smoking, age, education level, hypertension, Poverty Income Ratio (PIR), race, diabetes, alcohol consumption, and CVD. Objects fell into three distinct PIR categories: high-income (PIR >3.5), moderate-income (1.3 < PIR ≤3.5), and low-income (PIR ≤1.3).¹³ Individuals were sorted into the following three categories: underweight/healthy weight (<25 kg/m²), overweight (25–30 kg/m²), and obesity (>30 kg/m²).¹⁴ Smoking status had three categories based on participants’ smoking history and current smoking behavior: current smoking (total number of cigarettes smoked 100 or more, and currently smoking), ever smoking (total number of cigarettes smoked 100 or more, and currently no smoking), and never smoking (total number of cigarettes smoked less than 100).¹⁵ Alcohol consumption was defined as drinking at least 12 drinks per year.¹⁶ Participants with diabetes were defined as having any of the following conditions: (1) level of fasting blood glucose ≥126 mg/dL; (2) concentration of glycated hemoglobin (HbA1C) > 6.5%; (3) currently taking antidiabetic medication to lower blood sugar; (4) self-reported diabetes.^17,18 If any of the following were met, patients were determined to have hypertension: (1) diastolic blood pressure ≥90 mmHg; (2) self-reported hypertension; (3) systolic blood pressure ≥140 mmHg; (4) being on antihypertensive medication.¹⁹ CVD was determined as self-reported having had at least one of the diseases among heart attack, angina pectoris, coronary heart disease, stroke, and congestive heart failure.²⁰

Statistical analysis

Statistical analysis was processed using R software (V4.4.1). The baseline table was drawn by using the “tableone” package (sample size and percentage (n (%)) represented categorical variables; mean and standard deviation (mean (sd)) represented continuous variables). A weighted logistics regression model was established by using the “survey” package to probe into the relation between ePWV and PAD. Two models were created to adjust for covariates. Model 1 had adjustments for demographic information (age, race, gender), and Model 2 had adjustments for BMI, age, race, gender, smoking, alcohol consumption, PIR, hypertension, education level, diabetes, and CVD. The “rms” package was applied to construct the RCS to analyze the nonlinear association. Subgroup analysis was further undertaken to figure out the relation between ePWV and PAD in different subgroups. The difference between subgroups was evaluated by the likelihood ratio test of interaction terms (p < .05: significant difference).

Results

Baseline characteristics

This analysis enrolled a total of 6250 participants, among whom there were 462 patients with PAD (4.7%). The baseline characteristics of all participants are shown in Table 1. Most of the PAD patients were over 60 years old, with smoking habits, low education level (Senior high school education and below), and the middle and below poverty index. Compared to the non-PAD population, patients with PAD had considerably higher rates of diabetes, CVD, and hypertension. In addition, the ePWV of PAD patients was higher than that of the non-PAD group (10.90 vs 8.90, p < .001).

Table 1.

Baseline characteristics of included subjects.

Characters	Total	Non-PAD	PAD	p
Overall	6250	5788 (95.3)	462 (4.7)
Age				<0.001
<60	3087 (66.4)	3016 (68.5)	71 (25.1)
≥60	3163 (33.6)	2772 (31.5)	391 (74.9)
Gender				0.340
Male	3217 (49.1)	2980 (49.2)	237 (46.1)
Female	3033 (50.9)	2808 (50.8)	225 (53.9)
Race				<0.001
Mexican American	1305 (4.5)	1228 (4.6)	77 (3.9)
Other Hispanic	233 (4.4)	221 (4.5)	12 (2.5)
Non-Hispanic White	3443 (78.6)	3191 (78.6)	252 (76.9)
Non-Hispanic Black	1098 (8.8)	984 (8.5)	114 (15.2)
Other race	171 (3.7)	164 (3.8)	7 (1.5)
BMI (kg/m²)				0.171
<25	1743 (30.0)	1609 (30.2)	134 (26.5)
25–30	2484 (38.4)	2300 (38.5)	184 (37.1)
>30	2023 (31.6)	1879 (31.3)	144 (36.4)
Alcohol				0.006
No	2018 (29.4)	1849 (29.0)	169 (36.4)
Yes	4232 (70.6)	3939 (71.0)	293 (63.6)
Smoke				<0.001
Never	2893 (45.9)	2747 (46.7)	146 (29.4)
Past	2122 (33.1)	1925 (32.6)	197 (43.0)
Now	1235 (21.0)	1116 (20.7)	119 (27.6)
Education				<0.001
Less than high school	1096 (6.9)	979 (6.6)	117 (13.9)
High school or equivalent	2398 (37.7)	2201 (37.3)	197 (45.2)
College or above	2756 (55.4)	2608 (56.1)	148 (40.8)
PIR				<0.001
Low	1565 (16.4)	1412 (15.8)	153 (27.0)
Medium	2367 (33.4)	2157 (32.9)	210 (44.3)
High	2318 (50.2)	2219 (51.2)	99 (28.7)
Diabetes				<0.001
No	5437 (91.0)	5088 (91.6)	349 (80.3)
Yes	813 (9.0)	700 (8.4)	113 (19.7)
Hypertension				<0.001
No	1910 (35.9)	1846 (37.0)	64 (15.5)
Yes	4340 (64.1)	3942 (63.0)	398 (84.5)
CVD events				<0.001
No	5347 (88.1)	5038 (89.2)	309 (66.5)
Yes	903 (11.9)	750 (10.8)	153 (33.5)
ePWV	9.00 (2.04)	8.90 (1.97)	10.90 (2.40)	<0.001

n (%) represented the categorical variable and mean (sd) represented the continuous variable. n was unweighted. n (%), mean, and sd were weighted.

Association of ePWV with PAD risk

The weighted logistics regression model was utilized to analyze the relation between ePWV and PAD. In the fully adjusted model, ePWV was significantly positively linked with PAD risk (OR = 1.29, 95% CI: 1.18–1.42, p <.05). Next, the ePWV was stratified by tertiles. In the model with all covariates adjusted, the PAD risk of the highest tertile group was 2.22 times that of the lowest tertile group (95% CI: 1.05-4.70, P<0.05). The trend test manifested that in all models, with the growth of ePWV, the risk of PAD elevated significantly (P for trend<0.05) (Table 2).

Table 2.

Associations between ePWV and PAD from the NHANES 1999–2004 cohort.

Characteristic	OR (95% CI)
Characteristic	Crude model	Model 1	Model 2
CVD mortality
ePWV (continuous)	1.47 (1.41–1.54)^***	1.26 (1.15–1.37)^***	1.29 (1.18–1.42)^***
ePWV (categorical)
T1 (≤7.80)	Ref.	Ref.	Ref.
T2 (7.80-9.46)	2.54 (1.50–4.30)^***	1.85 (1.05–3.25)^*	1.56 (0.80–3.04)
T3 (>9.46)	8.79 (5.56–13.90)^***	2.65 (1.39–5.06)^**	2.22 (1.05–4.70)^*
P_trend	<0.001	0.005	0.022

The crude model had no adjustments. Model 1 adjusted for demographic information (age, race, gender), and Model 2 adjusted for age, gender, race, BMI, smoking, alcohol consumption, PIR, education level, hypertension, diabetes, and CVD. *p-value <.05, **p-value <.01, ***p-value <.001.

Non-linear link between ePWV and PAD

We further created the RCS model to dissect the nonlinear relation between ePWV and PAD. A great non-linear link between ePWV and PAD was detected (P-non-linear = 0.0229) (Figure 2). When the cutoff value of ePWV was 9.297, the right side of the cutoff value (ePWV ≥9.297) was greatly positively linked with PAD (OR = 1.03, 95% CI: 1.02–1.04, p < .001), while the left side (ePWV< 9.297) was not greatly linked with the risk of PAD. The Wald test further confirmed the statistical difference between the two groups (p < .001) (Table 3).

Figure 2.

Associations between ePWV and PAD by the RCS model. The RCS model was adjusted for age, race, gender, BMI, PIR, education, smoking, alcohol, hypertension, diabetes, and CVD. The solid line and blue area, respectively, represented the estimated values and their corresponding 95% CI.

Table 3.

Threshold effect analysis of ePWV on PAD risk.

Outcomes	OR (95% CI), p-value
ePWV
cutoff
<9.297	1.00 (0.99–1.01), 0.693
≥9.297	1.03 (1.02–1.04), <0.001
p for Wald test	<0.001

Adjustments were made to age, gender, race, BMI, smoking, alcohol consumption, PIR, education level, hypertension, diabetes, and CVD.

Subgroup analysis

Next, we undertook subgroup analyses with stratifications made on age, gender, and BMI. After adjusting for all confounding factors, ePWV had a great positive link with the risk of PAD in the elderly group (age ≥60) (OR = 1.33, 95% CI: 1.21–1.47, p < .001). No association was discovered in those under 60 years of age. In addition, BMI had a significant interactive effect on the linkage between ePWV and PAD (P for interaction = 0.004). A significant positive link between ePWV and PAD was shown in both the healthy group (BMI <25) and overweight group (25 ≤ BMI ≤30), but not in the obese group (BMI >30) (Table 4).

Table 4.

Subgroup analysis for the relationship between ePWV and PAD.

Characteristic	Model 2		P for interaction
Characteristic	OR (95% CI)	p	P for interaction
Age			0.499
<60	1.11 (0.78–1.59)	0.551
≥60	1.33 (1.21–1.47)	<0.001
Gender			0.645
Male	1.29 (1.14–1.46)	<0.001
Female	1.30 (1.14–1.49)	<0.001
BMI (kg/m²)			0.004
<25	1.43 (1.20–1.70)	<0.001
25-30	1.33 (1.12–1.57)	0.002
>30	1.14 (0.96–1.36)	0.142

Model 2 adjusted for age, gender, race, BMI, smoking, alcohol consumption, PIR, education level, hypertension, diabetes, and CVD.

Discussion

Atherosclerosis is the main pathological mechanism of PAD,²¹ which facilitates thrombus formation by triggering prothrombotic responses in blood vessels, platelet activation, and aggregation, and vasoconstriction, leading to vascular sclerosis and eventually symptomatic luminal limitation or total occlusion.²² In this nationally representative observational study, we discovered a significant positive link between ePWV and PAD risk, especially when ePWV was higher than 9.297. This threshold may reflect the marginal influence of arterial stiffness on PAD. Arterial elasticity was likely to be relatively preserved when ePWV was below this value, and therefore a significant link between ePWV and PAD was not exhibited.

The sub-group analysis demonstrated that ePWV was more remarkably linked with the risk of PAD in the elderly population (age ≥60), which may be related to vascular aging. ePWV has a bearing on vascular aging and may be an effective tool for studying vascular aging.⁸ During the aging process, the loss of vascular elasticity gradient and increased low-grade inflammation elevated the mechanical load on the vascular wall, leading to smooth muscle cell dedifferentiation and loss of plasticity, as well as deposition of elastic fibers and collagen fibers, thus resulting in age-related arterial stiffness increase.²³ This may explain the persistent uplift in the incidence of PAD with age, from around 5% in high-income countries aged 45–49 to 18% in those aged 85–89.²⁴ In addition, arteriosclerosis induced by vascular aging is closely connected with the elevated risk of hypertension.²⁵ Hypertension is one of the main risk factors for PAD.²⁶ These factors can together explain the great positive link between ePWV and PAD risk in the elderly population, while the younger population has better arterial elasticity,⁵ and the predictive effect of ePWV may be relatively weak. The elderly population is a high-risk population for PAD, and the risk assessment based on ePWV can be an early prediction tool.

We also unraveled a great interaction effect of BMI on the relation between ePWV and PAD risk. In non-obese individuals (BMI ≤30), ePWV had a great positive link with the risk of PAD. This linkage was not observed in obese individuals (BMI >30). A protective effect of higher BMI against PAD has been discovered in a former observational study,²⁷ which can be explained by the obesity paradox. Obesity is generally considered to be a risk factor for various diseases such as heart failure and type 2 diabetes, while the obesity paradox means that in certain chronic diseases, the prognosis of obese patients may be better than that of normal-weight or lean patients.²⁸ Obese patients receive more aggressive medical treatment and therefore possess a lower risk of PAD, while normal-weight patients may receive less treatment due to lower overall risk factors and thus possess a higher risk of PAD.^29–31 Another possible explanation is that obese patients have higher energy reserves, anti-inflammatory immune function, inflammatory preconditioning, and cardioprotective metabolic effects,³² which may partially offset the effect of increased arterial stiffness on PAD risk. Moreover, smoking is a pivotal risk factor for PAD,³³ while smokers tend to have a lower BMI,³⁴ which may partly explain why a stronger positive linkage exists between ePWV and PAD risk in non-obese people.

This research offers strong evidence to support a positive linkage between ePWV and PAD risk. However, certain limitations persist. Firstly, given the observational nature of the study, we were unable to definitively establish a causal relationship between ePWV and PAD. Moreover, devices that directly measure PWV can better reflect the true arterial stiffness, while ePWV, although as an alternative indicator of PWV, may be limited in accuracy as an indirect measurement tool. Third, further in-depth analysis is not conducted on the differences observed in the age and BMI subgroup analysis results. Therefore, we suggested that future studies should further dig out the linkage between ePWV and PAD risk in different age groups and BMI groups to more fully interpret the underlying mechanisms behind these differences.

Conclusion

Our findings indicated that ePWV is a practical indicator for predicting the risk of PAD, especially in the elderly population and non-obese individuals. Therefore, it is recommended to include ePWV testing in clinical practice for a more comprehensive assessment of the risk of PAD in individuals, especially for early intervention and management of high-risk groups.

Footnotes

Author contributions

Meiling Ning contributed to the study design. Xuehe Jiang conducted the literature search and acquired the data. Shuang Jia and Na Cui performed data analysis. Limei Yu and Meiling Ning wrote the article. Xuehe Jiang revised the article and gave the final approval of the version to be submitted. All authors read and approved the final manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Limei Yu

References

Criqui

Aboyans

. Epidemiology of peripheral artery disease. Circ Res 2015; 116: 1509–1526.

Barnes

Eid

Creager

, et al. Epidemiology and risk of amputation in patients with diabetes mellitus and peripheral artery disease. Arterioscler Thromb Vasc Biol 2020; 40: 1808–1817.

GBD 2019 Peripheral Artery Disease Collaborators . Global burden of peripheral artery disease and its risk factors, 1990–2019: a systematic analysis for the global burden of disease study 2019. Lancet Global Health 2023; 11: e1553–e1565.

Gregoire

Laloy-Borgna

Aichele

, et al. Flexural pulse wave velocity in blood vessels. J Acoust Soc Am 2024; 155: 2948–2958.

Fiori

Fuiano

Scorza

, et al. Non-invasive methods for PWV measurement in blood vessel stiffness assessment. IEEE Rev Biomed Eng 2022; 15: 169–183.

Podrug

Šunjić

Koren

, et al. What is the smallest change in pulse wave velocity measurements that can Be attributed to clinical changes in arterial stiffness with certainty: a randomized cross-over study. J Cardiovasc Dev Dis 2023; 10: 44.

Greve

Blicher

Kruger

, et al. Estimated carotid-femoral pulse wave velocity has similar predictive value as measured carotid-femoral pulse wave velocity. J Hypertens 2016; 34: 1279–1289.

Heffernan

Stoner

London

, et al. Estimated pulse wave velocity as a measure of vascular aging. PLoS One 2023; 18: e0280896.

Wang

Huang

, et al. Estimated pulse wave velocity and risk of new-onset heart failure. ESC Heart Fail 2024; 11: 2120–2128.

10.

Heffernan

Wilmoth

London

. Estimated pulse wave velocity and all-cause mortality: findings from the health and retirement study. Innov Aging 2022; 6: igac056.

11.

Melamed

Muntner

Michos

and et al. Serum 25-hydroxyvitamin D levels and the prevalence of peripheral arterial disease: results from NHANES 2001 to 2004. Arterioscler Thromb Vasc Biol 2008; 28: 1179–1185.

12.

Menke

Muntner

Wildman

, et al. Relation of borderline peripheral arterial disease to cardiovascular disease risk. Am J Cardiol 2006; 98: 1226–1230.

13.

Zhou

Huang

, et al. Relationship between dietary live microbe intake and the prevalence of COPD in adults: a cross-sectional study of NHANES 2013–2018. BMC Pulm Med 2024; 24: 225.

14.

Yao

Zhang

. Associations between copper and zinc and risk of hypertension in US adults. Biol Trace Elem Res 2018; 186: 346–353.

15.

Zhang

Liu

Zhang

, et al. Association between blood lead levels and hyperlipidemiais: results from the NHANES (1999–2018). Front Public Health 2022; 10: 981749.

16.

Dong

Sun

, et al. Association of coffee, decaffeinated coffee and caffeine intake from coffee with cognitive performance in older adults: national health and nutrition examination survey (NHANES) 2011–2014. Nutrients 2020; 12: 840.

17.

Zhao

. Association of serum vitamin C with liver fibrosis in adults with nonalcoholic fatty liver disease. Scand J Gastroenterol 2022; 57: 872–877.

18.

Zhang

Wei

Wang

, et al. The neutrophil-to-lymphocyte ratio is associated with all-cause and cardiovascular mortality among individuals with hypertension. Cardiovasc Diabetol 2024; 23: 117.

19.

Zhou

Xie

Liu

, et al. The dietary inflammatory index and its association with the prevalence of hypertension: a cross-sectional study. Front Immunol 2022; 13: 1097228.

20.

Liao

Zhang

Shi

, et al. Association of aldehyde exposure with cardiovascular disease. Ecotoxicol Environ Saf 2020; 206: 111385.

21.

Campia

Gerhard-Herman

Piazza

, et al. Peripheral artery disease: past, present, and future. Am J Med 2019; 132: 1133–1141.

22.

Habib

Petrucci

Rocca

. Pathophysiology of thrombosis in peripheral artery disease. Curr Vasc Pharmacol 2020; 18: 204–214.

23.

Lacolley

Regnault

Avolio

. Smooth muscle cell and arterial aging: basic and clinical aspects. Cardiovasc Res 2018; 114: 513–528.

24.

Fowkes

FGR

Aboyans

Fowkes

FJI

, et al. Peripheral artery disease: epidemiology and global perspectives. Nat Rev Cardiol 2017; 14: 156–170.

25.

Jin

, et al. Aging, arterial stiffness, and blood pressure association in Chinese adults. Hypertension 2019; 73: 893–899.

26.

Song

Rudan

Zhu

, et al. Global, regional, and national prevalence and risk factors for peripheral artery disease in 2015: an updated systematic review and analysis. Lancet Global Health 2019; 7: e1020–e1030.

27.

Heffron

Dwivedi

Rockman

, et al. Body mass index and peripheral artery disease. Atherosclerosis 2020; 292: 31–36.

28.

Simati

Kokkinos

Dalamaga

, et al.

Obesity paradox: fact or fiction?

Curr Obes Rep 2023; 12: 75–85.

29.

Hira

Kataruka

Akeroyd

, et al. Association of body mass index with risk factor optimization and guideline-directed medical therapy in US veterans with cardiovascular disease. Circ Cardiovasc Qual Outcomes 2019; 12: e004817.

30.

Lempesis

Liu

Dalamaga

. The catcher in the gut: tirzepatide, a dual incretin analog for the treatment of type 2 diabetes mellitus and obesity. Metabol Open 2022; 16: 100220.

31.

Vallianou

Tsilingiris

Kounatidis

, et al.

Sodium-glucose cotransporter-2 inhibitors in obesity and associated cardiometabolic disorders: where do we stand?

Pol Arch Intern Med 2022; 132: 16342.

32.

Karampela

Chrysanthopoulou

Christodoulatos

, et al. Is there an obesity paradox in critical illness? Epidemiologic and metabolic considerations. Curr Obes Rep 2020; 9: 231–244.

33.

Mackay

Pell

. Meta-analysis of the association between cigarette smoking and peripheral arterial disease. Heart 2014; 100: 414–423.

34.

Plurphanswat

Rodu

. The association of smoking and demographic characteristics on body mass index and obesity among adults in the U.S., 1999–2012. BMC Obes 2014; 1: 18.