. 2024 Jan 18;22(1):e8488. doi: 10.2903/j.efsa.2024.8488

TABLE 25.

Key epidemiological studies on arsenic and cardiovascular disease: Stroke.

Reference study population design	Outcome definition	Population size (n) case/control	Arsenic exposure	Results	Additional information/confounders
Zierold et al. (2004) Private well households in Wisconsin, USA Cross‐sectional	Self‐reported stroke	Population 1185 Number of cases not presented	w‐As 0–2389 μg/L, median 2 μg/L,16% > 10 μg/L. < 2 2–10 > 10 μg/L	OR 1.0 0.94 (0.40–2.14) 1.53 (0.60–4.07)	Adjusted for age, sex, smoking, BMI. Risk of information bias.
Wade et al. (2009) Retrospective cohort study in Ba Men, Inner Mongolia, China	Stroke mortality	78,251 person‐years during follow‐up 1997–2004. 447 deaths in those exposed since before 1995 (results in this table), 118 of these in stroke.	w‐As (μg/L) 0–5 5.1–20 20.1–100 100.1–300 > 300	Adj. IRR (cases) 1.0 (53) 0.47 (0.3–0.8) (16) 0.51 (0.3–0.8) (41) 0.52 (0.3–1.1) (7) 1.02 (0.2–6.7) (1)	Adjsted for age, sex, smoking, education, alcohol, farm work. Cause of death (blinded to w‐As) from proxy interviews and medical records.
Medrano et al. (2010) Ecological study of mortality in 651 municipalities in Spain	Stroke mortality	Population 14.4 million Number of cases: 81,368 18,327 3895	w‐As (μg/L) (median) < 1 (0.7) 1–10 (3.9) > 10 (23.3)	RR 1 1.00 (0.96–1.05) 1.02 (0.95, 1.09)	RR derived from reported % increase. p value trend 0.091 Adjusted for sex, age and covariates at municipal or provincial level (income, hospital beds, prevalence of smoking, hypertension, high serum cholesterol, diabetes, overweight/obesity and low physical activity, dietary factors, water hardness, magnesium, pH and temperature)
Chen, Wu, Liu, et al. (2013) Case‐cohort study in Bangladesh (HEALS)	Stroke incidence	148 cases of stroke Subcohort 1109	w‐As (μg/L) (n cases) 0.1–25 (50) 25.1–107 (46) 108–864 (52) Mean total u‐As in subcohort 277 μg/g creatinine	aHR 1.0 0.86 (0.49–1.51) 1.38 (0.84–2.27)	Adjusted for sex, age, smoking, BMI, education, hypertension and diabetes. Associations with u‐iAs and stroke not reported. %MMA (median 13%) was not associated with risk of stroke. Same cohort as in Chen et al. (2011a), but more cases
Farzan et al. (2015) Cohort study in New Hampshire, USA	Stroke mortality	Population 3939 Stroke deaths 43	As (μg/g) in toenails range (median) 0.01–3.26 (0.09) w‐As (μg/L) range (median) 0–158 (0.29)	HR per 1 unit ln‐transformed toenail‐As: 0.90 (0.61–1.33)	Cohort based on previous case–control study on skin cancer. Adjusted for smoking, education, skin cancer (and presumably age and sex, though not mentioned)
Wu et al. (2015) Case cohort study in Bangladesh (HEALS)	Incidence of stroke	165 cases of stroke Subcohort 1375	w‐As (μg/L, mean; number of cases) 0.1–16 (4.3; 44) 17–85 (47; 50) 86–864 (191; 71)	Adjusted HR 1.0 1.14 (0.65–1.98) 1.87 (1.06–3.29)	Same cohort as Chen, Wu, Liu, et al. (2013), but more cases. Mean u‐As 119 μg/L (259 μg/g)
Tsinovoi et al. (2018) Case cohort study in US (REGARDS)	Incidence of ischemic stroke	With u‐iAs: Population 199, Cases: 41. Total data set 671 cases and subcohort 2486	u‐iAs (μg/g creatinine) 0.1–3.52 3.53–66.8	Adjusted HR 1.0 2.18 (0.64–7.43)	Adjusted for analyses performed versus u‐tAs based on the whole data set showed no association
Ersbøll et al. (2018) Cohort study in Denmark (Danish prospective cohort Diet, Cancer and Health (DCH), two cities, Copenhagen and Aarhus.	Stroke incidence	Cohort 53,941 Incident stroke cases 2195 (1801 ischemic and 381 haemorrhage, 13 other), 678 in the Aarhus cohort	w‐As (μg/L 20 years mean). Total cohort median 0.7 0.05–0.57 (0.44) 0.57–0.76 (0.58) 0.76–1.93 (1.18) 1.93–25.3 (2.11) Aarhus cohort median 2.1 0.08–1.83 (1.30) 1.83–2.11 (2.09) 2.11–2.11 (2.11) 2.21–25.3 (2.11)	Adjusted IRRs 1.0 1.21 (1.07–1.36) 1.05 (0.92–1.19) 1.17 (1.04–1.32) 1.0 0.81 (0.62–1.04) 1.09 (0.88–1.34) 1.79 (1.41–2.26)	Adjusted for age, sex, smoking, BMI, waist, alcohol, physical activity, education, fruit intake, vegetable intake and calendar year. Note low w‐As and low contrast. Results for ischemic and haemorrhagic stroke relatively similar

Reference study population design

Outcome definition

Population size (n) case/control

Arsenic exposure

Results

Additional information/confounders

Zierold et al. (2004)

Private well households in Wisconsin, USA

Cross‐sectional

Self‐reported stroke

Population 1185

Number of cases not presented

w‐As 0–2389 μg/L, median 2 μg/L,16% > 10 μg/L.

< 2

2–10

> 10 μg/L

1.0

0.94 (0.40–2.14)

1.53 (0.60–4.07)

Adjusted for age, sex, smoking, BMI. Risk of information bias.

Wade et al. (2009)

Retrospective cohort study in Ba Men, Inner Mongolia, China

Stroke mortality

78,251 person‐years during follow‐up 1997–2004. 447 deaths in those exposed since before 1995 (results in this table), 118 of these in stroke.

w‐As (μg/L)

0–5

5.1–20

20.1–100

100.1–300

> 300

Adj. IRR (cases)

1.0 (53)

0.47 (0.3–0.8) (16)

0.51 (0.3–0.8) (41)

0.52 (0.3–1.1) (7)

1.02 (0.2–6.7) (1)

Adjsted for age, sex, smoking, education, alcohol, farm work.

Cause of death (blinded to w‐As) from proxy interviews and medical records.

Medrano et al. (2010)

Ecological study of mortality in 651 municipalities in Spain

Stroke mortality

Population 14.4 million

Number of cases:

81,368

18,327

3895

w‐As (μg/L) (median)

< 1 (0.7)

1–10 (3.9)

> 10 (23.3)

1.00 (0.96–1.05)

1.02 (0.95, 1.09)

RR derived from reported % increase. p value trend 0.091

Adjusted for sex, age and covariates at municipal or provincial level (income, hospital beds, prevalence of smoking, hypertension, high serum cholesterol, diabetes, overweight/obesity and low physical activity, dietary factors, water hardness, magnesium, pH and temperature)

Chen, Wu, Liu, et al. (2013)

Case‐cohort study in Bangladesh (HEALS)

Stroke incidence

148 cases of stroke

Subcohort 1109

w‐As (μg/L) (n cases)

0.1–25 (50)

25.1–107 (46)

108–864 (52)

Mean total u‐As in subcohort 277 μg/g creatinine

aHR

1.0

0.86 (0.49–1.51)

1.38 (0.84–2.27)

Adjusted for sex, age, smoking, BMI, education, hypertension and diabetes.

Associations with u‐iAs and stroke not reported. %MMA (median 13%) was not associated with risk of stroke. Same cohort as in Chen et al. (2011a), but more cases

Farzan et al. (2015)

Cohort study in New Hampshire, USA

Stroke mortality

Population 3939

Stroke deaths 43

As (μg/g) in toenails range (median)

0.01–3.26 (0.09)

w‐As (μg/L) range (median) 0–158 (0.29)

HR per 1 unit ln‐transformed toenail‐As: 0.90 (0.61–1.33)

Cohort based on previous case–control study on skin cancer. Adjusted for smoking, education, skin cancer (and presumably age and sex, though not mentioned)

Wu et al. (2015)

Case cohort study in Bangladesh (HEALS)

Incidence of stroke

165 cases of stroke

Subcohort 1375

w‐As (μg/L, mean; number of cases)

0.1–16 (4.3; 44)

17–85 (47; 50)

86–864 (191; 71)

Adjusted HR

1.0

1.14 (0.65–1.98)

1.87 (1.06–3.29)

Same cohort as Chen, Wu, Liu, et al. (2013), but more cases.

Mean u‐As 119 μg/L (259 μg/g)

Tsinovoi et al. (2018)

Case cohort study in US (REGARDS)

Incidence of ischemic stroke

With u‐iAs: Population 199, Cases: 41. Total data set 671 cases and subcohort 2486

u‐iAs (μg/g creatinine)

0.1–3.52

3.53–66.8

Adjusted HR

1.0

2.18 (0.64–7.43)

Adjusted for analyses performed versus u‐tAs based on the whole data set showed no association

Ersbøll et al. (2018)

Cohort study in Denmark (Danish prospective cohort Diet, Cancer and Health (DCH), two cities, Copenhagen and Aarhus.

Stroke incidence

Cohort 53,941

Incident stroke cases 2195 (1801 ischemic and 381 haemorrhage, 13 other), 678 in the Aarhus cohort

w‐As (μg/L 20 years mean). Total cohort median 0.7

0.05–0.57 (0.44)

0.57–0.76 (0.58)

0.76–1.93 (1.18)

1.93–25.3 (2.11)

Aarhus cohort median 2.1

0.08–1.83 (1.30)

1.83–2.11 (2.09)

2.11–2.11 (2.11)

2.21–25.3 (2.11)

Adjusted IRRs

1.0

1.21 (1.07–1.36)

1.05 (0.92–1.19)

1.17 (1.04–1.32)

1.0

0.81 (0.62–1.04)

1.09 (0.88–1.34)

1.79 (1.41–2.26)

Adjusted for age, sex, smoking, BMI, waist, alcohol, physical activity, education, fruit intake, vegetable intake and calendar year.

Note low w‐As and low contrast.

Results for ischemic and haemorrhagic stroke relatively similar

Abbreviations: adj, adjusted; aHR, adjusted hazard ratio; BMI, body mass index; DCH, ‘Diet, Cancer and Health’ study; HEALS, Health Effects of Arsenic Longitudinal Study; HR, hazard ratio; IRR, incidence risk ratio; MMA, sum of monomethylarsonous acid and monomethylarsonic acid; n, number; OR, odds ratio; REGARDS, Reasons for Geographic and Racial Differences in Stroke; RR, risk ratio; u‐As, urinary arsenic; u‐iAs, urinary inorganic arsenic; USA, United States of America; u‐tAs, urinary total arsenic; w‐As, water‐arsenic.