Whole Grain Consumption and Risk of Ischemic Stroke in Women: A Prospective Study

Context Although increased intake of grain products has been recommended to prevent cardiovascular disease (CVD), prospective data examining the relation of whole grain intake to risk of ischemic stroke are sparse, especially among women.

Objective To examine the hypothesis that higher whole grain intake reduces the risk of ischemic stroke in women.

Design, Setting, and Participants A prospective cohort of 75,521 US women aged 38 to 63 years without previous diagnosis of diabetes mellitus, coronary heart disease, stroke, or other CVDs in 1984, who completed detailed food frequency questionnaires (FFQs) in 1984, 1986, 1990, and 1994, and were followed up for 12 years as part of the Nurses' Health Study.

Main Outcome Measure Incidence of ischemic stroke, confirmed by medical records, by quintile of whole grain intake according to FFQ responses.

Results During 861,900 person-years of follow-up, 352 confirmed incident cases of ischemic stroke occurred. We observed an inverse association between whole grain intake and ischemic stroke risk. The age-adjusted relative risks (RRs) from the lowest to highest quintiles of whole grain intake were 1.00 (referent), 0.68 (95% confidence interval [CI], 0.49-0.94), 0.69 (95% CI, 0.51-0.95), 0.49 (95% CI, 0.35-0.69), and 0.57 (95% CI, 0.42-0.78; P = .003 for trend). Adjustment for smoking modestly attenuated this association (RR comparing extreme quintiles, 0.64; 95% CI, 0.47-0.89). This inverse association remained essentially unchanged with further adjustment for known CVD risk factors, including saturated fat and transfatty acid intake (multivariate-adjusted RR comparing extreme quintiles, 0.69; 95% CI, 0.50-0.98). The inverse relation between whole grain intake and risk of ischemic stroke was also consistently observed among subgroups of women who never smoked, did not drink alcohol, did not exercise regularly, or who did not use postmenopausal hormones. No significant association was observed between total grain intake and risk of ischemic stroke.

Conclusions In this cohort, higher intake of whole grain foods was associated with a lower risk of ischemic stroke among women, independent of known CVD risk factors. These prospective data support the notion that higher intake of whole grains may reduce the risk of ischemic stroke.

Although stroke-related mortality has clearly declined since the mid-20th century,¹ incidence rates of stroke have been level since the mid-1980s, and stroke remains a leading cause of serious disability and death in women.² In the United States, approximately 600,000 cases of stroke accounted for 160,000 deaths in 1997.² At all ages, more women than men die of stroke. Many of the approximately 4.4 million stroke survivors have permanent disability, and the cost of stroke-related care amounts to $40 billion each year.³ Because known risk factors for the development of stroke, including hypercholesterolemia, hypertension, obesity, and diabetes mellitus, can be changed through dietary modification, primary prevention holds much promise.⁴ Few studies, however, have specifically examined the relationship of diet to stroke risk, especially among women.^5,6 Moreover, previous studies have often failed to distinguish stroke subtypes, although the pathologic mechanisms for ischemic and hemorrhagic strokes are known to differ.⁴

Several epidemiological studies have associated higher intakes of specific nutrients, such as dietary fiber, potassium, and vitamin E, with a lower risk of cardiovascular disease (CVD).^7-10 Few epidemiologic studies, however, have directly examined the relationship between foods rich in such nutrients, particularly whole grains, fruits, and vegetables, and risk of stroke, especially ischemic stroke. Such studies are important because whole foods contain numerous other beneficial phytochemicals—many of which are not documented in existing databases¹¹—and because the beneficial effects of specific nutrients may not be apparent when examined individually because of potentially important interactions between them. Recently, we and others reported that a higher intake of whole grains was associated with a lower risk of coronary morbidity¹² and mortality.¹³ For this report, we analyzed prospective data from the Nurses' Health Study from 1984-1996 to evaluate the hypothesis that greater whole grain intake reduces the risk of ischemic stroke. A direct assessment of the association between whole grains and ischemic stroke can provide not only important biological insights but also practical dietary guidance as well.

The Nurses' Health Study is a prospective cohort study of diet and lifestyle factors in relation to chronic diseases among 121,700 female registered nurses aged 30 to 55 years at enrollment. The cohort was initiated in 1976 when the participants returned a mailed questionnaire about various dietary and lifestyle risk factors for chronic diseases.¹⁴ In 1980, we assessed diet with a 61-item semiquantitative food frequency questionnaire (FFQ).¹⁴ In 1984, the FFQ was expanded to include 126 items. Because the expanded questionnaires contained numerous additional food items that are important for assessing details of carbohydrate intake, we considered 1984 as the baseline for the current analysis. We excluded respondents with previously diagnosed diabetes mellitus, angina, myocardial infarction, stroke, or other CVDs in 1984. The final eligible baseline population for our analyses consisted of 75,521 women aged 38 to 63 years.

Measurements of whole grain and refined grain foods along with other aspects of diet were repeated in 1986, 1990, and 1994 using FFQs similar to the one used in 1984. For each food, a commonly used unit or portion size (eg, 1 slice of bread) was specified, and the participant was asked how often on average during the previous year she had consumed that amount. Nine responses were possible, ranging from "never" to "six or more times per day." The type and brand of breakfast cereal were also assessed. The method used in classifying whole and refined grains has been described previously.^12,13 Specifically, whole grain foods included dark bread, whole grain breakfast cereal, popcorn, cooked oatmeal, wheat germ, brown rice, bran, and other grains (eg, bulgur, kasha, and couscous). Refined grain foods included sweet rolls and cakes/desserts, white bread, pasta, English muffins, muffins or biscuits, refined grain breakfast cereal, white rice, pancakes or waffles, and pizza. The list of breakfast cereals reported in the FFQ was evaluated for whole grain and bran content based on data provided by the package labels, and breakfast cereals with 25% or more whole grain or bran content by weight were classified as "whole grain." A full description of the FFQs and data on reproducibility and validity in this cohort have been previously reported.^14,15 The performance of the FFQ for assessing the individual grain products has been documented to be high.¹⁶ For example, comparing the FFQ with detailed diet records in a sample of the participants, correlation coefficients were 0.75 for cold breakfast cereal, 0.71 for white bread, and 0.77 for dark bread.

We confirmed incident cases of stroke by reviewing medical records. When nonfatal stroke was reported on a follow-up questionnaire, we obtained permission to review the pertinent medical records and then classified stroke according to the criteria recommended by the National Survey of Stroke.¹⁷Ischemic stroke was defined as a clinical syndrome, including a constellation of neurological findings of sudden or rapid onset persisting for 24 hours or more, and these findings were judged to be from thrombotic or embolic occlusion of a cerebral artery, resulting in infarction. "Silent" strokes were excluded, as were events secondary to trauma, infection, or cancer. Deaths were reported by next of kin or were obtained from postal authorities or the National Death Index.¹⁸ Because the pathologic mechanisms for ischemic and hemorrhagic strokes are known to differ,⁴ we focused on ischemic stroke in our analyses to reduce end-point misclassification. This is because the most likely plausible benefits of whole grains may be associated with the etiology of arterosclerosis.

Person-time for each participant was calculated from the date of return of the 1984 questionnaire to the date of confirmation of stroke, death, or June 1, 1996, whichever came first. We initially examined distributions of individual foods to create categories of consumption with adequate person-time at risk in each category. Incidence rates then were calculated by dividing the number of events by the person-time in each category. The relative risks (RRs) were estimated as the rate of ischemic strokes in a specific category of intake of whole grain or refined grain divided by the rate in the lowest category. We then conducted multivariate analyses adjusting for age (5-year categories), body mass index (BMI; calculated as weight in kilograms divided by the square of height in meters) (6 categories), level of physical activity (hours per week in 5 categories), smoking status (never, past, and current: 1-14, 15-24, and ≥25 cigarettes per day), alcohol intake (4 categories: never, 0.1-4.9 g/d, 5.0-14.9 g/d, and ≥15.0 g/d), parental history of myocardial infarction at age younger than 60 years (yes or no), self-reported (asked as "doctor diagnosed" in the questionnaire) hypertension (yes or no) and high blood cholesterol level (yes or no), menopausal status (premenopausal, postmenopausal without hormone replacement therapy, postmenopausal with past hormone replacement therapy, or postmenopausal with current hormone replacement therapy), use of multivitamins or vitamin E supplements (yes or no), saturated fat (quintiles), transfatty acids (quintiles), and total energy intake (quintiles). Because fruit and vegetable intake⁵ was inversely associated with risk of ischemic stroke, we conducted a secondary analysis further adjusting for intake of fruits and vegetables (in quintiles). To further minimize residual confounding effects of smoking, alcohol consumption, physical inactivity, and use of postmenopausal hormones, we examined the association between whole grain intake and risk of ischemic stroke in 4 subgroups of women: never smokers, nondrinkers, inactive lifestyle, and nonusers of postmenopausal hormones. Tests of linear trend across increasing categories of grain intake were conducted by assigning the medians of intakes in categories (servings per day) treated as a continuous variable. All P values were 2-sided.

To reduce within-person variation and to best represent long-term diet, we applied a cumulative average method that used repeated measures of diet during the 12-year follow-up. Details of this method have been reported elsewhere.¹⁹ Briefly, we used pooled logistic regression²⁰ to model the incident cases of ischemic stroke using the cumulative average diet from all cycles of the FFQs up to the beginning of each 2-year follow-up interval. For example, incidence of ischemic stroke from 1984-1986 was related to the intake of whole grain and other dietary variables assessed on the 1984 questionnaire, incidence from 1986-1990 was related to the average intake reported in 1984 and 1986, and incidence from 1990-1994 was related to the average intake reported in 1984, 1986, and 1990. Intermediate end points that occurred during the follow-up, including angina, hypercholesterolemia, diabetes mellitus, and hypertension, could have led to changes in diet and therefore may have confounded the associations between diet and disease. Hence, we stopped updating information on diet at the beginning of the interval during which those conditions were diagnosed in a participant.

At baseline in 1984, the mean daily intake of whole grain foods was 1.12 servings per day. The median intake of whole grain ranged from virtually no consumption (0.13 servings/d) in the lowest quintile to nearly 3 servings per day in the highest quintile (Table 1). Women with a high intake of whole grains smoked less, exercised more, and were more likely to use postmenopausal hormones or take multivitamins or vitamin E supplements. Greater whole grain intake also was associated with higher intakes of carbohydrates, fruits and vegetables, dietary fiber, and folate, but lower intakes of fats, cholesterol, and alcohol (Table 1). Body mass index and history of parental myocardial infarction at age younger than 60 years did not vary appreciably across quintiles of whole grain intake. In contrast, findings were generally reversed for the associations between these lifestyle factors and the intake of refined grains.

Intakes of both total grains and whole grains increased from 1984-1994, while intake of refined grain remained fairly stable over time (Table 2). The mean intake of whole grains increased from 1.12 servings per day in 1984 to 1.36 servings per day in 1994. The Spearman correlation coefficients between any 2 assessments of whole grain intakes reported from 1984-1994 ranged from 0.44-0.61, depending on the closeness in the time of assessments.

During 12 years of follow-up (861,900 person-years), 352 confirmed incident cases of ischemic stroke occurred. We observed a strong inverse association between whole grain intake and risk of ischemic stroke but no significant association between refined grain intake and ischemic stroke risk (Table 3). Comparing women in the highest with the lowest quintiles of intake, the age-adjusted RRs of ischemic stroke were 0.57 for whole grain intake (95% confidence interval [CI], 0.42-0.78; P = .003 for trend) and 0.90 for refined grain intake (95% CI, 0.64-1.26; P = .30 for trend). In multivariate models of whole grain intake and ischemic stroke risk, smoking was the strongest confounding factor. While adjustment for smoking attenuated the inverse relation between whole grain intake and risk of ischemic stroke, it remained strong and significant (RR, 0.64; 95% CI, 0.47-0.89, when comparing the 2 extreme quintiles; P = .04 for trend). In a multivariate model adjusting for known CVD risk factors, including saturated fat and transfatty acids, the inverse relation between whole grain intake and risk of ischemic stroke remained essentially unchanged (RR, 0.69; 95% CI, 0.50-0.98, when comparing the 2 extreme quintiles; P = .08 for trend) (Table 3). Additional adjustment for intake of fruits and vegetables did not materially change the inverse relation between whole grain intake and risk of ischemic stroke (RRs across increasing quintiles of whole grain intake were 1.00, 0.73, 0.78, 0.61, and 0.70; P = .09 for trend).

To further minimize residual confounding effects of smoking, alcohol consumption, physical inactivity, and use of postmenopausal hormones, we examined the stratified models of whole grain intake and risk of ischemic stroke in 4 subgroups of women: never smokers, nondrinkers, inactive lifestyle, and nonusers of postmenopausal hormones. The reduction in risk of ischemic stroke associated with a high intake of whole grain was even stronger among never smokers (RR, 0.50; 95% CI, 0.34-0.76; P = .006 for trend), while it remained essentially unchanged from the overall results among nondrinkers, women who did not engage in vigorous physical activity, or nonusers of postmenopausal hormones (Table 4). In addition, we found that the inverse relation of whole grain intake to risk of ischemic stroke was similar regardless of levels of intake of saturated fat or transfatty acids, BMI, or physical activity levels.

To determine whether the effects of whole grains exceed their potential known mediators, we examined whether the inverse relation of whole grain intake to ischemic stroke risk could be attributed to increased intakes of dietary fiber, folate, potassium, magnesium, and vitamin E. After these constituents of whole grains were simultaneously included in a multivariate model, the association of whole grain intake with risk of ischemic stroke was somewhat attenuated (RR, 0.76; 95% CI, 0.51-1.15, when comparing the highest vs the lowest quintiles; P = .38 for trend).

We also examined the relations of whole grain intake to risk of total stroke. An inverse association was found for total stroke (multivariate-adjusted RRs across increasing quintiles of whole grain intake were 1.00, 0.90, 0.92, 0.79, and 0.77; P = .06 for trend). Relatively few cases of hemorrhagic stroke (n = 76) occurred among the participants of the Nurses' Health Study between 1984 and 1996, so the power to detect an association between whole grain intake and risk of hemorrhagic stroke was limited. Whole grain intake did not appear to reduce the risk of hemorrhagic stroke, with an age- and smoking-adjusted RR for hemorrhagic stroke of 1.01 (95% CI, 0.44-2.32, when comparing the 2 extreme quintiles of whole grain intake; P = .98 for trend) and a multivariate-adjusted RR of 1.15 (95% CI, 0.51-2.58; P = .90 for trend). The number of incident fatal strokes was also small (n = 91), and the multivariate-adjusted RR was 0.91 (95% CI, 0.37-2.20; P = .33 for trend) when comparing the 2 extreme quintiles of whole grain intake.

In this large prospective study, the higher the consumption of whole grain foods the lower the risk of ischemic stroke, independent of a variety of CVD risk factors. Because whole grain intake was reported through a self-administered FFQ, measurement errors were inevitable. However, assessment of whole grain foods by the questionnaires used in this study had relatively high validity (r>0.70 for most whole grain foods when compared with a 7-day dietary record). Our FFQ was designed to minimize day-to-day variation by assessing an average long-term dietary intake. More importantly, it can clearly discriminate participants' dietary patterns and rank them accordingly.^21,22 For example, in a sample of nurses without diabetes mellitus or coronary heart disease who provided fasting blood samples, we observed a positive dose-response relation between fasting plasma levels of triglycerides and dietary carbohydrate that was consistent with findings from controlled metabolic experiments.^23,24 Furthermore, any measurement error in assessing whole grain intake should not be related to stroke end point because dietary assessments were completed before the end points occurred. This kind of nondifferential misclassification would tend to weaken the association between whole grain intake and ischemic stroke risk and would not explain the significant 40% lower risk rate for ischemic stroke associated with a high intake of whole grain.

Several potential alternative explanations need to be considered when interpreting our findings. First, the association between whole grain intake and ischemic stroke may be because of other potential confounding factors. As expected, a higher intake of whole grains was correlated with a generally healthy lifestyle. However, the apparent protective effect of whole grain consumption persisted in multivariate models that accounted for known CVD risk factors. Furthermore, the apparent inverse relation between whole grain intake and risk of ischemic stroke was remarkably consistent among subgroups of women who were never smokers, did not drink alcohol, did not report regular vigorous physical activity, or did not use postmenopausal hormones. This argues against the possibility of residual confounding from these known risk factors.

Because women with a high background risk of ischemic stroke, such as those with diabetes mellitus, hypertension, or hypercholesterolemia, may change their diet after being diagnosed with these conditions, the associations between whole grain intake and ischemic stroke may be confounded by these indications. In addition, intermediate end points that occurred during the follow-up, including angina, hypercholesterolemia, diabetes mellitus, and hypertension, also could have led to changes in diet and therefore may have confounded the associations between diet and disease. To account for this, we stopped updating information on diet at the beginning of the interval during which those conditions were diagnosed in a participant. Although we could not entirely exclude the possibility of confounding by indication, it seemed more likely that the protective effect of whole grain intake would be underestimated because women who perceived themselves to be at increased risk of ischemic stroke would tend to increase their intake of whole grain foods. Similar findings were observed when women with those conditions at baseline in 1984 were excluded from the main analysis. Also, these results are internally consistent with our previous report of an inverse association between whole grain consumption and risk of coronary heart disease in this cohort of female nurses.¹² Because of the differences in etiologic mechanisms between hemorrhagic and ischemic stroke, the finding that consumption of whole grains was not associated with risk of hemorrhagic stroke indirectly strengthens the interpretation that the inverse relation between whole grain and ischemic stroke that we observed may be mechanistic rather than because of chance or bias.

The protective effect of whole grains against ischemic stroke may involve multiple biological pathways,¹¹ and the active ingredients in whole grains that may impart this risk reduction are not completely understood. Whole grains contain an abundance of antioxidants, minerals, phytochemicals, and fibers in both the outer (bran) and inner (germ) layers, both of which are removed during processing. Of the myriad compounds found in whole grains, folate, vitamin E, magnesium, potassium, and fibers have received much attention. Folate can lower homocysteine blood levels by converting homocysteine to methionine, and a high intake of folate has been associated with a lower risk of coronary events.²⁵ High serum homocysteine and low serum folate levels also have been associated with carotid artery stenosis.²⁶ Greater intake of potassium, magnesium, and vitamin E also has been independently associated with reduced risks of coronary events or ischemic stroke.^7-10,27 Finally, intake of cereal fiber has been consistently associated with lower risk of coronary events in prospective cohort studies,^9,28 which also may explain the protective effects of whole grains. Because our primary objective was to evaluate the total impact of whole grains on risk of ischemic stroke, we did not control for the effects of these whole grain constituents in our primary analyses. In our secondary analyses, however, we examined whether the protective effects of whole grains are limited to these known constituents or go beyond them. Adjustment for these components of whole grain foods attenuated the inverse relation of whole grain intake with ischemic stroke, suggesting that the apparent protective effect of whole grains can be partially attributed to these protective constituents. However, the inverse relation remained (although it was not statistically significant), indicating that other constituents may contribute to additional protection. Before generalizing our findings to the other populations, it is important to consider that our study cohort was approximately 98% white. Because nonwhite women are underrepresented, when findings are expected to differ significantly according to underlying biological differences between ethnic groups, studies in other populations will be desirable to confirm our findings.

Survey data indicate that most of the grains consumed in the United States are processed and refined,²⁹ with the average consumption of whole grain products at approximately one-half serving per day.³⁰ In this cohort of female nurses, whole grains accounted for only one third of the servings of total grains, and median consumption of whole grain foods was 1 serving per day. Even the women in the top quintile of whole grain intake barely approached the 3 servings per day that are generally recommended.³¹ Given that the inverse relation we observed between whole grain intake and ischemic stroke risk in this cohort of women was continuous, although risk reduction seemed to level off at the fourth quintile (about a 30%-40% lower risk rate of ischemic stroke associated with 1.3 servings per day), replacing refined grains with whole grains by even 1 serving a day may have significant benefits in reducing the risk of ischemic stroke.

In conclusion, higher intakes of whole grain foods were associated with a lower risk of ischemic stroke, independent of known CVD risk factors, in this large population of women. These findings support the hypothesis that increasing intake of whole grains may help reduce the incidence of ischemic stroke.