Demographic and clinical risk factors of developing clinically-recognized varicose veins in older adults

Abstract

Introduction:

Varicose veins are common in older adults and are associated with adverse clinical outcomes such as deep venous thrombosis. Established risk factors for varicose veins include female sex, height, and obesity, but other risk factors are relatively uncharacterized.

Methods:

This was a prospective cohort analysis of 6241 participants aged 66–70 years from the Atherosclerosis Risk in Communities (ARIC) Study. Incident varicose veins were defined as two outpatient encounters (at least a week apart) or inpatient diagnoses through 2018 with ICD 9 code 454 or ICD 10 code I83. Participants with a history of clinically-recognized varicose veins at baseline were excluded. Cox regression was used to evaluate established (e.g., female, height, body mass index) and potential demographic and clinical risk factors.

Results:

During a median follow-up of 13 years, 349 (6%) of participants developed clinically-recognized varicose veins. Consistent with prior research, female sex, taller height, and higher body mass index were associated with varicose veins. After accounting for these, White race, prevalent heart failure, loop diuretic use, higher cardiac troponin T, and higher natriuretic peptide were independently associated with incident varicose veins.

Conclusions:

In this community-based cohort study of older adults, known and newly identified risk factors, including cardiac function and heart failure, were independently associated with incidence of clinically-recognized varicose veins. The potential usefulness of cardiac biomarkers for prevention and screening of varicose veins requires further investigations.

INTRODUCTION

Lower extremity chronic venous disease manifests in a broad clinical spectrum ranging from asymptomatic (e.g., no/palpable signs, telangiectasias veins) to symptomatic signs (e.g., ulcer). Varicose veins are part of the spectrum of lower extremity chronic venous diseases and affect approximately 33 million adults in the United States ¹. Recent studies demonstrate that varicose veins are associated with serious adverse health outcomes such as leg ulceration ² and deep vein thrombosis ^3–5. In addition, associations of varicose veins with incidence of peripheral artery disease and other vascular diseases have been reported ^{3, 6–10}.

A few risk factors for varicose veins, such as older age, female sex, taller height, obesity, pregnancy, and a family history of varicose veins have been reported in multiple studies ^11–21, but some other factors like physical activity ^{16, 20–24}, smoking ^{16, 17, 21–24}, and hypertension^{19, 21, 25} have yielded inconsistent results. However, many previous studies were cross-sectional ^{13–16, 18–20} or had short-term follow-up (a median follow-up <5 years) ^{11, 23}, and most investigated selected populations of White race ^{11, 13, 16–19, 21–24} or middle-aged individuals ^{11, 14, 16, 22–25}.

Therefore, we comprehensively investigated established, debated and potential risk factors of clinically-recognized varicose veins over 28 years of follow-up in Black and White participants of the community-based cohort study, the Atherosclerosis Risk in Communities (ARIC) study. This investigation may inform preventive approaches for varicose veins.

METHODS

Study population

The ARIC study includes 15,792 middle-aged (45–64 years of age), predominantly Black and White men and women who were enrolled through population sampling in 1987–1989 from four US communities: Forsyth County, North Carolina; Jackson, Mississippi; suburbs of Minneapolis, Minnesota; and Washington County, Maryland. Follow-up study visits were conducted in 1990–1992 (visit 2), 1993–1995(visit 3), 1996–1998 (visit 4), 2011–2013 (visit 5), 2016–2017 (visit 6), 2018–2019 (visit 7), 2020 (visit 8 [phone-based due to the COVID-19 pandemic]), 2021–2022 (visit 9), and 2023-ongoing (visit 10). In addition, the Carotid MRI sub-study was performed in a subset of participants between 2004 and 2006. ARIC participants were followed continuously for identifying hospitalizations and deaths by contacting participants or proxy annually (semiannually since 2012), and surveying discharge lists from local hospitals and death certificates from state vital statistics offices.

To capture outpatient visits for the present study, ARIC participant data was linked with Centers for Medicare and Medicaid Services (CMS) claims for the years 1991–2018. An identified a subsample of ARIC participants was eligible for Medicare services (age65+years) within the ARIC study. The earliest ARIC visit for every eligible participant was included to maximize both sample size and length of follow-up and a 5-year age interval was imposed to capture as many participants as possible. Specifically, all ARIC participants entered into this subsample cohort when they first reached the age of 66–70 years at any of four regular visits (visit 2 to visit 5 [1990–2013]) or the Carotid MRI sub-study visit (2004–2006). Age 66 was selected as the minimum age to allow for a one-year lookback period that would exclude prevalent varicose vein diagnoses. Visit 5 (2011–13) was used as the last visit to allow a relevant follow-up time between the exposures and incident varicose veins diagnosis by 2018. Of 6,325 participants in this subsample cohort, those with race other than Black or White (n=18) and those with prior history of clinically-identified varicose veins identified from linked CMS inpatient or outpatient claims or from ARIC inpatient data (n=66) were excluded (n=6,241 of final sample) (Figure 1). The ARIC study was approved by the Institutional Review Board at participating institutions (University of North Carolina, Wake Forest University, University of Mississippi, University of Minnesota, and Johns Hopkins University).

Figure 1.

Flow chart of ARIC subcohort for current analysis among those with 66–70 year-old

Measures

Variables that were routinely evaluated in ARIC clinic visits (measured at least across three study visits or time-fixed sociodemographic variables [i.e., age, sex, race and education]), including anthropometric measures (e.g., height, BMI, and blood pressure), lifestyle behaviors (i.e., smoking, alcohol intake and physical activity), medical history (i.e., hypertension, diabetes, history of cardiovascular disease and relevant medication uses), and lab data (i.e., lipid parameters and cardiac markers) were explored as potential risk factors. All factors of interest were collected at every study visit unless otherwise specified. Data was used from the visit closest to when the participant entered this nested cohort (e.g., if a participant reached 66–70 years for the first time at visit 3, we used data collected at visit 3). When this was not possible (either a participant had a missing value or ARIC did not measure a variable of interest at a relevant visit), data from the nearest previous visit was carried over.

Age, sex, and race were ascertained through interviews at visit 1. Information on educational attainment, obtained at visit 1, was categorized as basic (less than high school), intermediate (high school graduate or vocational school), and advanced (college, graduate school, or professional school). Smoking status was self-reported and categorized into current, former, or never. Physical activity was assessed at visits 1, 3, and 5 and in the carotid MRI sub-study using a modified Baecke Questionnaire consisting of sport, leisure, and work indices ranging from 1 (low) to 5 (high) ²⁶. Work index was not included in the current study since participants were ≥65 years old. Alcohol intake was categorized into current, former, and never-drinking. Use of antihypertensive (e.g., beta-blockers, angiotensin-converting-enzyme inhibitors, angiotensin receptor blocker, and loop diuretics), antidiabetic, and lipid-lowering medications within the past 2 weeks (the past 4 weeks for visit 5) were self-reported and confirmed by the inspection of medication containers.

Trained technicians measured standing height without shoes and with back and heels against a vertically mounted centimeter ruler to the nearest centimeter. Weight was measured with a balance scale. Body mass index was calculated as weight (in kilograms) divided by height (in meters) squared. Sitting blood pressures were measured three times by sphygmomanometer after a 5-min rest (only two measurements at visit 4), and the mean of the last two measurements was recorded. Blood samples were drawn by certified staff, and blood measurements were performed in study-specific central laboratories. Diabetes was defined as fasting serum glucose level ≥126 mg/dL, non-fasting glucose level ≥200 mg/dL, self-reported physician diagnosis of diabetes, or antidiabetic medication use. Total and high-density lipoprotein cholesterol concentrations were assessed by the enzymatic procedure ²⁷. Estimated glomerular filtration rate was calculated based on the serum creatinine-based CKD Epidemiology Collaboration equation ²⁸. Both high-sensitivity cardiac troponin T (hs-cTnT) and N-terminal-pro-B-type-natriuretic peptide (NT-proBNP) were measured using serum samples obtained at visits 2, 4, and 5. High-sensitivity C-reactive protein (hs-CRP) was measured by immunoturbidimetric assay on a BNII analyzer (Siemens Healthcare Diagnostics, Deerfield, Illinois) at visits 2, 4, and 5 ²⁹. History of coronary heart disease was defined as self-reported clinical history, evidence of prior myocardial infarction by electrocardiogram obtained at visit 1, or cases adjudicated by physician panel after visit 1. History of stroke was similarly defined by self-reported history at visit 1 and any adjudicated cases after visit 1. History of heart failure (HF) was defined as self-reported history at visit 1 and hospitalization with HF (International Classification of Diseases [ICD]-9 code 428) after visit 1.

The outcome of interest was a clinical diagnosis of varicose veins identified from ICD-9 codes 454 (until September 2015) or ICD-10 codes I83 (from September 2015 onward) at any inpatient or outpatient encounter found in ARIC hospitalization records and linked CMS Medicare claims. For outpatient cases, two encounters with a varicose veins diagnosis were used that were at least a week apart to ensure independent diagnoses rather than capturing a diagnosis carried forward from a single sequential clinical course ³⁰. All participants were followed up through the date of the varicose veins diagnosis, the date of death, or administrative censoring on December 31, 2018, whichever came first.

Statistical analysis

Baseline characteristics of participants were summarized by the status of incident varicose veins. For survival analysis, the cumulative incidence of clinically-recognized varicose veins was first estimated by demographic variables (i.e., sex and race) using the Kaplan-Meier method. Subsequently, each established and potential risk factor was included in demographically adjusted Cox proportional hazards models (Model 1: adjusted for age, sex, and race). Model 2 included, in addition to age, sex, and race, all risk factors identified as having a statistically significant association with incident varicose veins.

For sensitivity analysis, the associations of risk factors with inpatient and outpatient varicose veins were examined separately. Analysis was restricted to those who enrolled in Medicare fee-for-service (FFS) beneficiaries (i.e., Medicare Parts A and B) continuously for at least a year prior to baseline and stayed in this status during follow-up since participants who enrolled in Medicare Advantage plan (Medicare Parts C and D) could use healthcare providers outside of Medicare service. For this analysis, participants who terminated Medicare FFS were censored. In addition, analyses were repeated stratified by demographic subgroups (male vs. female and White vs. Black).

Statistical analyses were performed using Stata version 18 (StataCorp LP, College Station, TX), and two-sided p-values were reported.

RESULTS

Among 6,241 older adults without a history of varicose veins, the mean age at baseline was 67 (SD 1) years, 54% were female, and 78% were White. 62% of study population were enrolled in traditional FFS Medicare within a year prior to study baseline. Baseline characteristics are summarized in Table 1.

Table 1.

Baseline characteristics by presence and absence of incident varicose veins in a nested ARIC cohort

	Overall (N=6241)
Age, years	67.4 (1.3)
Female	53.6%
White	77.8%
Education
Basic	23.2%
Intermediate	39.9%
Advanced	36.9%
Height, m	1.7 (0.1)
BMI, kg/m2	28.4 (5.4)
SBP, mmHg	129.0 (19.6)
DBP, mmHg	69.6 (10.6)
Antihypertensive medication use	52.4%
Beta blocker	16.6%
ACEi	16.5%
ARB	2.5%
Calcium channel blocker	14.6%
Loop diuretic	5.8%
Thiazide diuretic	9.9%
Hypertension	55.8%
Diabetes	23.1%
Total cholesterol, mmol/L	5.2 (1.0)
HDL-cholesterol, mmol/L	1.3 (0.4)
Triglyceride, mmol/L	1.6 (0.9)
Lipid-lowering medication use	23.2%
eGFR, mL/min/1.73m2	82.5 (16.1)
Smoking status
Former	44.9%
Current	12.8%
Drinking status
Former	27.8%
Current	48.5%
Sport index	2.6 (0.8)
Leisure index	2.4 (0.6)
hs-cTnT, ng/L	6.7 [4.0–6.7]
NT-proBNP, pg/mL	79.5 [43.1–79.5]
hs-CRP, mg/L	4.5 (8.4)
Prevalent CHD	10.5%
Prevalent stroke	3.2%
Prevalent heart failure	6.4%
Medicare parts A and B (fee-for-service plan within 1 y prior to baseline)	62.0%

Abbreviation: ACEi, Angiotensin-converting enzyme inhibitors; ARB, Angiotensin-receptor blocker; BMI, body mass index; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HDL, high-density lipoprotein; hs-CRP, high-sensitivity c-reactive protein; CHD, coronary heart disease; hs-cTnT, high- sensitivity cardiac troponin T; NT-proBNP, N-terminal-pro-B-type-natriuretic peptide; SBP, systolic blood pressure

A total of 349 (5.6%) participants developed clinically-recognized varicose veins over a median follow-up of 13.1 [25^th-75^th percentile 6.9, 20.4] years (271 cases identified from outpatient visits only, 23 cases identified from ARIC hospitalization only and 8 cases only from CMS inpatient data). Female participants and White participants had a higher cumulative incidence of varicose veins than their counterparts (Figure 2). The independent associations of these demographic factors were confirmed (i.e., female and White participants) even after accounting for each other and age (adjusted HR 1.53 [95% CI 1.23, 1.91] for female vs. male and 2.13 [1.52, 2.98] for White vs. Black participants; Appendix Table 1).

Figure 2.

Kaplan-Meier survival curves for incident varicose veins diagnosis, by sex and race. The ARIC Study (1993–2018)

In addition, in the demographically adjusted model, established risk factors, such as height and BMI, were positively associated with incident varicose veins (Model 1: adjusted HR 1.44 [95%CI 1.21, 1.70] per 0.1 m increment in height and 1.35 [1.23, 1.48] per 5 kg/m² increment in BMI; Appendix Table 1). Furthermore, the use of loop diuretics, prevalent HF, and higher levels of hs-cTnT, NT-proBNP, and hs-CRP were associated with incidence of a clinically-recognized varicose veins. However, smoking and physical activity, were not associated with incidence of varicose veins. When a model was constructed including all variables significantly associated with incident varicose veins (Model 2; Table 2), the associations of most predictors remained significant, although loop diuretics and hs-CRP were no longer significantly associated with incident varicose veins. When inpatient and outpatient varicose veins were explored separately, results were generally similar, whereas female sex and height were significantly associated with only outpatient visits for varicose veins and prevalent HF in only inpatient varicose veins (Appendix Table 2).

Table 2.

Hazard ratios (95% CI) of incident varicose veins for demographic and clinical risk factors

	Total study population (N=5747)a
Selected risk factors	Model 1	Model 2
Age (per 1 year)	1.12 (1.01, 1.23)	1.08 (0.98, 1.20)
Female vs. Male	1.55 (1.23, 1.95)	2.52 (1.78, 3.56)
White vs. Black	2.18 (1.52, 3.12)	2.77 (1.91, 4.00)
Height (per 0.1 m increase)	1.41 (1.18, 1.68)	1.47 (1.23, 1.75)
BMI (per 5 kg/m2 increase)	1.38 (1.26, 1.52)	1.35 (1.22, 1.50)
Loop diuretic use vs. non-useb	2.03 (1.34, 3.06)	1.45 (0.95, 2.23)
Prevalent heart failure vs. nob	1.77 (1.17, 2.66)	1.54 (1.02, 2.33)
hs-cTnT (2-fold higher)b	1.18 (1.06, 1.32)	1.12 (1.01, 1.25)
NT-proBNP (2-fold higher)b	1.16 (1.07, 1.26)	1.16 (1.07, 1.26)
hs-CRP (2-fold higher)b	1.12 (1.04, 1.20)	1.05 (0.97, 1.13)

Abbreviation: BMI, body mass index; eGFR, estimated glomerular filtration rate; hs-CRP, high-sensitivity c-reactive protein; hs-cTnT, high-sensitivity cardiac troponin T; NT-proBNP, N-terminal-pro-B-type-natriuretic peptide

Model 1: adjusted for age, sex and race

Model 2: Model 1+risk factors shown significant association in Model 1

All risk factors showing significant association in Model 1 (except for age) are listed in the table. In addition to those variables, we originally included education, blood pressure, medication use (beta-blockers, angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers, thiazide diuretic, statin), hypertension, diabetes, lipid levels, smoking, alcohol intake, estimated glomerular filtration rate, urine albumin-to-creatinine ratio, physical activity, high-sensitivity c-reactive protein, and prevalent coronary heart disease and stroke

^aWe excluded those who had missing information on risk factors with statistically significant associations with incident varicose veins in demographically adjusted model (Web Table 1)

^bThese variables reflect the status of heart failure, and therefore were modeled separately. Hazard ratios of other variables were from model including loop diuretic.

Cardiac markers were examined after log-transforming.

When analyses were restricted to participants who enrolled in Medicare FFS continuously for at least a year prior to baseline and stayed in this status during follow-up, the results were generally consistent (Appendix Tables 3 and 4). Also, results were generally consistent across demographic subgroups (male vs. female and Whites vs. Blacks) (Appendix Tables 5 and 6). Although the interactions were not statistically significant, the association of loop diuretics with incident varicose veins tended to be stronger in male vs. female (HR 2.57 [95% CI 1.24, 5.31] vs. 1.18 [0.69, 2.00]) (Appendix Table 5).

DISCUSSION

In this community-based cohort study of older adults, female sex, greater height, and higher BMI were associated with increased risk of clinically-recognized varicose veins. In addition to these established risk factors, White race and several variables related to HF (i.e., prevalent HF, loop diuretic use, hs-cTnT, and NT-proBNP) were independently associated with subsequent risk of clinically-recognized varicose veins. Similar results were found when only participants who enrolled in the Medicare FFS continuously for at least 1 year before baseline were included. Also, results were generally consistent across demographic subgroups, but the association of loop diuretic use with incident varicose veins was stronger in male than female participants.

Racial differences in the prospective incidence of varicose veins may reflect limited access to care and thus lower chance of a varicose veins diagnosis in Black individuals, as is consistent in other medical conditions ^{31, 32}. Of interest, a previous study suggested that the identification of spider veins in the early stage of chronic venous insufficiency is easier in White race than in other races due to skin pigmentation ³³. However, importantly, two cross-sectional studies with dedicated venous evaluation not relying on clinical indications (one in >2000 US community-dwelling adults aged 40–79 years ¹⁵ and the other in >600 employees from a US hospital ³³) showed a higher prevalence of venous reflux on duplex ultrasound in persons of White race than in other racial or ethnic groups, including Black individuals.

In current study, several factors that reflect cardiac function and HF, were associated with incidence of clinically diagnosed varicose veins. Our observational study cannot prove causality, and this association may reflect the higher likelihood of physicians checking legs of patients with diagnosed HF compared to their counterparts. However, HF can result in elevated venous pressure and thus blood pooling in the leg venous system. The robust associations of two cardiac biomarkers (i.e., hs-cTnT and NT-proBNP) with incident varicose veins in our study further support the etiological link of cardiac overload to varicose veins. Importantly, those biomarkers remained associated with incident varicose veins even after adjusting for interim HF during follow-up (Table S7). Nonetheless, further studies are needed to assess whether these markers can be clinically useful to identify individuals at high risk of developing varicose veins.

The stronger association of loop diuretic use with incident varicose veins in male than in female participants deserves some discussion. Loop diuretics can reduce pressure on the veins in the legs, improving leg symptoms associated with venous insufficiency (e.g., swelling). One possible explanation for the interaction is that reflux in deep veins is more prevalent in men than women ^{15, 34}, although men are less likely to develop varicose veins than women, in general ³⁵. Thus, HF may need to be more severe to develop varicose veins in men than in women, and loop diuretic use might represent severe HF since loop diuretics are used to treat volume overload due to HF, whereas thiazide diuretics are often used for the treatment of hypertension.³⁶ Indeed, our data showed that loop diuretics were significantly associated with the risk of varicose veins, but not thiazide diuretics (Appendix Table 1). Also, among loop diuretic users in our study, NT-proBNP levels were higher in men than in women (median 211.7 [IQI 61.5, 935.3] pg/mL and 124.3 [58–294] pg/mL, respectively). However, our subgroup analysis was performed without a prespecified hypothesis and thus should be interpreted as hypothesis-generating.

There are a few clinical implications of these findings. Varicose veins are a common condition potentially resulting in serious adverse outcomes such as leg ulceration and deep vein thrombosis ^{2–4, 6}. Therefore, understanding risk factors for varicose veins is important for prevention and early diagnosis of varicose veins. Varicose veins are often considered a cosmetic problem despite their established association with adverse health outcomes ^3–10, and thus healthcare providers may not always pay attention to them. In this context, healthcare providers probably should consider examining the legs of individuals at high risk of varicose veins, such as female sex, tall height, and high BMI. Of these factors, BMI is modifiable and thus weight management has implications for preventing varicose veins. Indeed, previous literature suggested that physical exercise improves leg function in patients with varicose veins ³⁷, possibly lowering risk of varicose veins. Also, anti-obesity medications have recently attracted attention ³⁸. Although we do not expect these medications to be used for the purpose of preventing varicose veins, it is possible that they reduce varicose veins as a result of weight reduction. In addition, our results suggest that two major cardiac markers, cardiac troponin, and natriuretic peptide, may help identify individuals at high risk of varicose veins. Importantly, these cardiac markers are currently used for diagnosing cardiac diseases, and thus, data should be readily available in some patients. Nonetheless, the potential usefulness of these cardiac biomarkers to guide prevention and screening of varicose vein is a new concept and requires further investigation.

Limitations

This study has a few limitations. Since a duplex ultrasound was not available, ICD codes were used to define incident varicose veins, which may result in misclassification of diagnosis. Nonetheless, a few previous studies showed high positive predictive value of varicose veins based on ICD codes (93%–98%) with a duplex ultrasound examination or chart review as a gold standard ^{3, 39}. The latter study also reported a high negative predictive value of 98% ³⁹. In addition, confirming a few established risk factors of varicose veins (e.g., taller height) in our study provides face validity of our definition. The number of participants with clinically-recognized varicose veins was relatively low, limiting statistical power to identify modest risk factors. Potential risk factors, such as family history, were not available in our study. Since we only included older adults (66–70 years), we were not able to fully characterize some candidate variables, such as work-related activity. We were not able to examine the varicose veins severity (e.g., CEAP [Clinical, Etiological, Anatomical, and Pathophysiological] classification) due to small number of cases with varicose veins and concurrent conditions, including severe cases (e.g., cellulitis or ulcer). Also, our study population was older at a time when varicose veins often initially occur. It is possible that disease severity may not progress to the point requiring clinical attention until later in life, and many cases are likely never recognized clinically.

CONCLUSIONS

In this community-based cohort study of older adults, in addition to several already established predictors, White race, HF prevalence, and factors related to cardiac function were independently associated with the risk of clinically-recognized varicose veins. These results have implications for identifying older adults at high risk of varicose veins, although further studies are required to investigate the pathophysiological relationship between cardiac function and varicose veins.