Body mass index and risk of stroke in women

Obesity and excess weight have been proven to increase the risk of various diseases,¹ including some cancers, osteoarthritis, type 2 diabetes mellitus, hypertension, and coronary heart disease.^2,3 Obesity has also been shown to increase the risk of stroke.^4-9 Several studies have found a positive correlation between body mass index (BMI) and stroke in men, but the association between BMI and stroke and its subtypes is not well understood in women. In this study, we assessed the relationship between BMI and total, ischemic, and hemorrhagic stroke among apparently healthy female health care professionals.

Patients and methods

We prospectively studied 39,053 women participating in the Women’s Health Study, a randomized trial evaluating the use of low-dose aspirin and vitamin E for the primary prevention of cardiovascular disease and cancer. Participants were aged 45 years or older and were apparently healthy at baseline in 1993.^10,11

Weight and height were reported on a questionnaire given at the start of the study, as well as on several follow-up questionnaires. The weight in kilograms was divided by the height in meters squared to achieve the BMI. BMI was then divided into 7 a priori categories, ranging from below 20.0 kg/m² to 35.0 kg/m² or above. BMI was also assessed as a continuous term, as well as divided into groups of 25 kg/m² or less (normal weight), 25 to 29.9 kg/m² (overweight), and 30.0 kg/m² or above (obese), based on the World Health Organization (WHO) classification. The incidence of stroke was self-reported and confirmed by a review of medical records. The relationship between BMI and stroke was assessed using the Cox proportional hazards model.

Results

Four strokes of unknown origin, 81 hemorrhagic strokes, and 347 ischemic strokes occurred (total = 432) after 10 years of follow-up. Participants had a mean (± SD) BMI of 26.0 (± 5.1) kg/m². Eighteen percent of participants had a BMI of 30 kg/m² or higher, 31% had a BMI between 25.0 and 29.9 kg/m², and 50.8% had a BMI of less than 25.0 kg/m². Participants with a BMI of 30 kg/m² or higher were more likely to have diabetes, hypertension, and increased cholesterol levels. This group also used less postmenopausal hormones, exercised less, drank less alcohol, and smoked fewer cigarettes.

As shown in the Table, among the 7 categories of BMI, there was a statistically significant trend toward increased risk of total and ischemic stroke (P < .01), after adjustment for use of postmenopausal hormones, ex&shy;ercise, alcohol intake, smoking, and age. For participants with a BMI of 27 kg/m² or above, the risk was substantially higher. The hazard ratio (HR) for participants with a BMI of 35 kg/m² or above was 2.81 (95% confidence interval [CI] = 1.45-5.43) for ischemic stroke and 2.05 (95% CI = 1.18-3.55) for total stroke, compared with participants with BMIs of 20 kg/m² or less.

The association between BMI and total and ischemic stroke was strongly attenuated when information on cholesterol level, hypertension, and diabetes was included in the multivariable models (HR for total stroke = 1.19, 95% CI = 0.68-2.10; HR for ischemic stroke = 1.54, 95% CI = 0.79-3.02).

BMI (kg/m²) categories

< 20.0

20.0-22.9

23.0-24.9

25.0-26.9

27.0-29.9

30.0-34.9

>= 35.0

n = 2024

n = 9882

n = 7935

n = 6038

n = 6043

n = 4770

n = 2361

value

P

Total

No. of cases

21

93

82

62

81

57

36

(n = 432)

Age-adjusted

1.00

0.96 (0.59-1.53)

1.00 (0.62-1.61)

0.96 (0.58-1.57)

1.32 (0.81-2.13)

1.25 (0.75-2.06)

1.84 (1.07-3.17)

< .01

Model 1*

1.00

1.03 (0.64-1.65)

1.07 (0.66-1.74)

1.06 (0.65-1.75)

1.44 (0.89-2.34)

1.37 (0.83-2.28)

2.05 (1.18-3.55)

< .01

Model 2†

1.00

0.99 (0.62-1.59)

0.98 (0.61-1.59)

0.93 (0.56-1.53)

1.18 (0.73-1.93)

1.05 (0.63-1.74)

1.38 (0.79-2.41)

.16

Model 3‡

1.00

0.99 (0.62-1.60)

0.98 (0.61-1.59)

0.92 (0.56-1.51)

1.14 (0.70-1.87)

0.97 (0.58-1.61)

1.19 (0.68-2.10)

.49

Ischemic

No. of cases

13

76

59

49

69

50

31

(n = 347)

Age-adjusted

1.00

1.26 (0.70-2.27)

1.16 (0.63-2.11)

1.22 (0.66-2.24)

1.81 (1.00-3.27)

1.77 (0.96-3.25)

2.58 (1.35-4.95)

< .01

Model 1*

1.00

1.35 (0.75-2.43)

1.22 (0.67-2.23)

1.33 (0.72-2.46)

1.94 (1.07-3.52)

1.91 (1.03-3.54)

2.81 (1.45-5.43)

< .01

Model 2†

1.00

1.29 (0.72-2.33)

1.10 (0.60-2.02)

1.15 (0.62-2.12)

1.56 (0.85-2.84)

1.41 (0.76-2.63)

1.81 (0.93-3.54)

.04

Model 3‡

1.00

1.30 (0.72-2.34)

1.10 (0.60-2.02)

1.12 (0.61-2.10)

1.49 (0.82-2.73)

1.29 (0.69-2.41)

1.54 (0.79-3.02)

.21

Hemorrhagic

No. of cases

(n = 81)

8

16

22

13

10

7

5

Age-adjusted

1.00

0.43 (0.18-1.00)

0.70 (0.31-1.57)

0.53 (0.22-1.28)

0.42 (0.17-1.07)

0.39 (0.14-1.09)

0.64 (0.21-1.95)

.37

Model 1*

1.00

0.48 (0.20-1.12)

0.81 (0.36-1.84)

0.63 (0.26-1.52)

0.50 (0.20-1.29)

0.46 (0.17-1.30)

0.76 (0.24-2.38)

.55

Model 2†

1.00

0.47 (0.20-1.10)

0.77 (0.34-1.74)

0.58 (0.24-1.41)

0.45 (0.17-1.15)

0.39 (0.14-1.11)

0.59 (0.18-1.89)

.26

Model 3‡

1.00

0.47 (0.20-1.10)

0.78 (0.34-1.75)

0.58 (0.24-1.41)

0.45 (0.17-1.16)

0.38 (0.13-1.09)

0.56 (0.17-1.82)

.24

values are for trend across categories. Only P values of < .05 were considered statistically significant and are in boldface.

*Adjusted for age, smoking status, exercise, alcohol consumption, and postmenopausal hormone use.

†Adjusted for all variables in Model 1 plus history of hypertension.

‡Adjusted for all variables in Model 2 plus history of diabetes and increased cholesterol level. (Reprinted, with permission, from Circulation. 2005;111[15]:1992-1998.)

P

BMI was not a strong risk factor for hemorrhagic stroke, contrasting with the positive association between BMI and total as well as ischemic stroke. Although not statistically significant, there was an indication that the risk of hemorrhagic stroke in&shy;creased among women with a BMI below 20 kg/m², compared with participants with BMIs between 20.0 and 22.9 kg/m².

Using the BMI classification from the WHO, compared with participants who had a BMI of less than 25 kg/m², participants with a BMI of 30 kg/m² or above (obese) had HRs of 1.50 (95% CI = 1.16-1.94) for total stroke, 1.72 (95% CI = 1.30-2.28) for ischemic stroke, and 0.82 (95% CI = 0.43-1.58) for hemorrhagic stroke. The risks for total and ischemic stroke were reduced to 1.15 (95% CI = 0.85-1.55) and 1.04 (95% CI = 0.80-1.37), respectively, after controlling for in&shy;creased cholesterol level, hypertension, and diabetes.

For each increase in 1 unit of BMI, there was a 4% increase in the risk of total stroke and a 5% increase in the risk of ischemic stroke, which was statistically significant; there was no relationship between continued BMI and hemorrhagic stroke. As before, these associations were attenuated to nonsignificant findings after controlling for cholesterol level, hypertension, and diabetes. Throughout the follow-up period, the trend shown for the relationship between BMI and the risk of stroke at baseline was no different when the BMI data were updated over the follow-up period.

The association between BMI and stroke as well as stroke subtypes was not statistically significantly modified by age, smoking, or exercise.

Discussion

Excess weight has not been recognized as an established risk factor for stroke¹² and generally has not been included in the overall estimation of stroke risk.^13,14 Although the association between stroke and obesity has been shown in men,^4,5,15,16 an association in women has been shown in some,^7,17-19 but not all studies.^20-22 A significant relationship was shown be&shy;tween BMI and total and ischemic stroke in this study, but not between BMI and hemorrhagic stroke. In addition, our study indicated that exercise, smoking status, or age did not modify this relationship.

Our findings support data from the Nurses’ Health Study, which showed results comparable to ours; in that study, women with a BMI of 32 kg/m² and above had a 1.6-fold increased risk of total stroke and more than twice the risk of ischemic stroke, versus those who had a BMI below 21 kg/m².⁷ These risk estimates were markedly reduced when in&shy;creased cholesterol level, hypertension, and diabetes were included, as they were in our study. In addition, there was a nonsignificant inverse relationship between hemorrhagic stroke and BMI, similar to our re&shy;sults. The reason other studies may not have shown a relationship be&shy;tween excess weight and stroke may have been because they were controlled for the possible biologic mediator hypertension,^20,22 or because the sample size was small.²¹

In some studies, a statistically significant excess risk of stroke re&shy;mained, even after controlling for increased cholesterol level, hypertension, and diabetes. BMI was shown to be a statistically significant and independent predictor for stroke in a recent population-based study from Sweden.²³ Al&shy;though nearly all studies assessing the association between excess weight and ischemic stroke indicated that in&shy;creased cholesterol level, hypertension, and diabetes are potential biologic mediators, the sig&shy;nificance of excess weight as a major risk factor for stroke should not be discounted.^4,5,7,9,24 Be&shy;cause ex&shy;cess weight has such an important influence on the occurrence of diabetes and hy&shy;pertension, it may be the largest single predictor of stroke risk.

Conclusions

BMI was shown to be a strong risk factor for total and ischemic stroke but not for hemorrhagic stroke in this large study of apparently healthy women. The presence of increased cholesterol level, hypertension, and diabetes attenuated this relationship. The high rate and length of follow-up, the large number of participants and strokes, and the prospective method of data collection were strengths of this study. The results of our study show that the number of total and ischemic strokes may be decreased if obesity is prevented.