Intake of Fish and Omega-3 Fatty Acids and Risk of Stroke in Women FREE

An inverse relationship between fish intake and risk of stroke has been reported in several,^1- 2 but not all,^3- 4 prospective studies. Mechanisms for protection against stroke by fish intake may include inhibition of platelet aggregation^5- 6; lowered blood viscosity⁶; suppressed formation of leukotrienes (lipid mediators for neutrophil and macrophage aggregation)⁷; and reduction of plasma fibrinogen⁸ blood pressure levels⁹ and insulin resistance.¹⁰ The possibility that fish consumption may increase the risk of hemorrhagic stroke was suggested by ecologic studies in which Greenland Eskimos with very high intake of omega-3 polyunsaturated fatty acids (omega-3 fatty acids) had an excess risk of mortality from hemorrhagic stroke compared with Danish whites.^11- 12 However, the average food supply per capita for omega-3 fatty acids (eicosapentaenoic acid and docosahexaenoic acid) among persons living in the United States is about 0.1 to 0.2 g/d,¹³ much lower than that of Danish whites (0.8 g/d) and Greenland Eskimos (10.5 g/d).¹⁴ Thus, the potential for any increased risk of hemorrhagic stroke may be minimal at the average intake level of US residents. To our knowledge, no prospective study has examined previously the relationship between intake of fish and omega-3 fatty acids and risk of stroke by stroke subtype. We investigated this relationship prospectively during 14 years of follow-up. Our a priori hypothesis was that intake of fish and omega-3 fatty acids would be associated with reduced risk of ischemic stroke and would not be associated with increased risk of hemorrhagic stroke among middle-aged US women.

The Nurses' Health Study began in 1976, when 121 700 female registered nurses (98% white) living in 11 states who were then aged 30 to 55 years completed questionnaires about their lifestyle and medical history, including previous cardiovascular disease, cancer, diabetes, hypertension, and high serum cholesterol concentration.¹⁵ Every 2 years, follow-up questionnaires have been sent to update information and identify new major illnesses. A total of 98 759 women returned the 1980 dietary questionnaire. For these analyses, we excluded women who left 10 or more items blank (n = 4056), those with reported total food intakes judged to be implausible (n = 2235), and those who had a history of cancer (except nonmelanoma skin cancer), angina, myocardial infarction, coronary revascularization, stroke, or other cardiovascular diseases prior to 1980 (n = 6739). Women who reported a history of physician-diagnosed diabetes or high serum cholesterol levels (n = 5890) were also excluded because these disorders might have caused a change in diet, particularly with regard to intake of fish and fatty acids. After these exclusions, 79 839 women remained for these analyses.

The semiquantitative food frequency questionnaire used in 1980 included a survey of 61 foods, including a single question assessing fish intake.¹⁶ A common unit or portion size for each food (eg, 6-8 oz [168-224 g] for fish) was specified, and each woman was asked how often on average during the previous year she had consumed that amount. Nine responses were possible for each food item, ranging from "almost never" to "six or more times per day." In 1984, 1986, and 1990, the dietary questionnaire was expanded to include 4 fish and seafood items: (1) dark-meat fish such as mackerel, salmon, sardines, bluefish, or swordfish (3-5 oz [84-140 g]);(2) canned tuna (3-4 oz [84-112 g]); (3) other fish (3-5 oz [84-140 g]); and (4) shrimp, lobster, or scallops as main dish (3.5 oz [98 g]). The average daily intake of nutrients was calculated by multiplying the frequency of consumption of each item by its nutrient content per serving and totaling the nutrient intake for all food items.

For calculating intake of long chain omega-3 fatty acids (eicosapentaenoic acid and docosahexaenoic acid), we assigned grams per serving as follows: 1.51 g for dark-meat fish, 0.42 g for canned tuna fish, 0.48 g for other fish, and 0.32 g for shrimp, lobster, or scallops. These omega-3 fatty acids values were derived by weighting the mean values of omega-3 fatty acids¹⁷ for the most common types of fish based on US landing data in 1984 (US Department of Commerce). For dark-meat fish with landings of 10 million pounds or more, the percentage of the total landing weight was 70% for salmon, 18% for Atlantic herring, 7% for canned sardines, 3% for mixed mackerel, and 1% for bluefish and swordfish. For canned tuna, these values were 75% for light tuna, and 25% for white tuna. For other fish, these values were 43% for flounder, 19% for Atlantic cod, 16% for whiting (Hake), 9% for halibut, 8% for Atlantic pollock, and 5% for haddock. For shrimp, lobster, and scallop, the value of omega-3 fatty acids in moist shrimp was used. The reason for the relatively high omega-3 fatty acids grams per serving for the "other fish" category (0.48 g) was due to the inclusion of flounder and whiting, which contain 0.57 to 0.59 g of omega-3 fatty acids per serving (4 oz [112 g]) even though they are regarded as light-meat fish.

Nutrient intakes were adjusted for total energy intake by the residual approach.¹⁶ To make the intake of marine omega-3 fatty acids from the 1980 questionnaire be as comparable as possible with the later, more detailed questionnaires, we assigned 1.16 g of long chain omega-3 fatty acids per portion (6-8 oz [168-224 g]) on the 1980 questionnaire. This number was calculated as a weighted average of long chain omega-3 fatty acids composition from dark-meat fish, canned tuna, and other fish, using the relative consumption of these types of fish on the 1984 dietary questionnaire. Intake of long chain omega-3 fatty acids was primarily from fish (87% of the total intake) and secondarily from chicken (7%) and liver (2%), which is similar to that in the US food supply data.¹³

The reproducibility and validity of the 1984 dietary questionnaire were assessed in a random sample of 127 men aged 45 through 70 years living in the Boston, Mass, area by comparing the data from the questionnaire with the data from two 1-week dietary records, collected approximately 6 to 8 months apart and with the fatty acid composition of adipose tissue. Spearman correlation coefficients for the fish items between 2 questionnaires administered 1 year apart were 0.63 for dark-meat fish; 0.54 for canned tuna; 0.48 for other fish; and 0.67 for shrimp, lobster, or scallops as a main dish.¹⁸ The mean total fish intake was 3.7 servings per week according to the questionnaire and 3.6 servings per week according to two 1-week dietary records (Spearman correlation coefficient, 0.61; P<.001).¹⁸ The energy-adjusted intake of eicosopentaenoic acid from fish also was correlated with percentage of eicosopentaenoic acid in adipose tissue (Spearman correlation coefficient, 0.49; P<.001).¹⁹ Information on fish oil supplement was not requested until 1990 in the Nurses' Health Study. At that point the prevalence of consumption of this supplement was only 1.6%.

Strokes were included in these analyses if they occurred after the date of return of the 1980 questionnaire and before June 1, 1994. Women who reported a nonfatal stroke on a follow-up questionnaire were asked for permission to review their medical records. Medical records were available for review for 81% of stroke cases and were confirmed by physicians blinded to the data on diet and other risk factors. Nonfatal strokes for which confirmatory information was obtained by telephone or letter but for which no medical records were available were regarded as probable (19% of nonfatal strokes). Deaths were ascertained by reports of relatives or postal authorities and a search of the National Death Index. Mortality follow-up was more than 98% complete.²⁰ For fatal strokes, confirmatory information was obtained by telephone interview, letter, medical records, or death certificate. When no medical records were available (death certificate information only), fatal strokes were regarded as probable (21.8% of fatal strokes). For the analyses of total stroke, both confirmed and probable strokes were used.

Strokes were confirmed by medical records according to the criteria of the National Survey of Stroke,²¹ which requires a constellation of neurologic deficits of sudden or rapid onset lasting at least 24 hours or until death. Events were classified as subarachnoid hemorrhages, intraparenchymal hemorrhages, ischemic strokes (thrombotic or embolic), or stroke of undetermined type. Subarachnoid hemorrhage was defined as hemorrhage in the subarachnoid space, usually caused by the rupture of a saccular aneurysm of the cerebral arteries, less commonly by arteriovenous malformations or other causes. Intraparenchymal hemorrhage was defined as hemorrhage in intraparenchymal regions of the brain not due to an aneurysm or arteriovenous malformation. Ischemic stroke includes cerebral infarction caused by thrombi (thrombotic stroke) or by emboli from extracranial sources (embolic stroke). For each subtype of stroke, a definite diagnosis was made when computed tomographic (CT) scan, magnetic resonance imaging (MRI), angiography, surgery, or autopsy confirmed the lesion. If such confirmation was lacking, a probable diagnosis was made.

All confirmed thrombotic strokes were further classified as large-artery occlusive infarction, lacunar infarction, or unclassified thrombotic infarction based on the results of CT scan, MRI, or autopsy according to the criteria of Perth Community Stroke Study.²² Large-artery occlusive infarction was defined as infarction involving the cerebral artery regions in the cerebrum and cerebellum, presumably caused by thrombosis of large or medium-sized cerebral arteries. A definite diagnosis was made when CT scan, MRI, or autopsy showed confirmatory findings. If imaging studies lacked positive findings, but the patient had cerebral signs, a diagnosis of probable large-artery occlusive infarction was made. Lacunar infarction, which is caused by occlusion of small penetrating arteries, was defined as infarction of a focal, small, and deep area such as internal capsule, corona radiata, basal ganglia, or brainstem without involvement of cortex. A definite diagnosis was made when CT scan, MRI, or autopsy showed confirmatory findings. If imaging studies were negative, but the patient had a lacunar syndrome (pure motor stroke, pure sensory stroke, ataxic hemiparesis, dysarthria-clumsy hand syndrome, or sensorimotor stroke), a diagnosis of probable lacunar infarction was made. Other thrombotic strokes were regarded as unclassified thrombotic infarction. Additional details and validation of the stroke subclassification system have been published elsewhere.²³ When we combined the data for definite and probable cases of large-artery occlusive infarction and lacunar infarction, results were similar (70 and 20 for definite and probable large-artery occlusive infarction; and 63 for definite and probable lacunar infarction, respectively).

Analyses were based on incidence rates of stroke during 14 years of follow-up (1980-1994). For each woman, person-months of follow-up were calculated from the date of return of the 1980 questionnaire to the first end point, death, or June 1, 1994, whichever came first. Women who reported having cardiovascular disease or cancer on previous questionnaires were excluded from subsequent follow-up.

Because of the long follow-up period, dietary exposures and variables were updated to better represent long-term dietary patterns, using the 1980, 1984, 1986, and 1990 dietary questionnaires. We calculated intakes of fish and omega-3 fatty acids as a cumulative average of intake from all available dietary questionnaires up to the start of each 2-year follow-up interval in which events were reported. For example, the incidence of stroke from 1980 through 1984 was related to fish and omega-3 fatty acids intake reported on the 1980 questionnaire, and the incidence from 1984 through 1986 was related to the average intake reported on the 1980 and 1984 questionnaires. Because changes in diet after the development of intermediate end points such as angina, hypercholesterolemia, and diabetes may confound information on diet and disease, we stopped updating information on diet at the beginning of the interval during which these intermediate end points developed in an individual subject. The other nutrient variables (saturated fat, trans-unsaturated fat, animal protein, linoleic acid, and calcium) and intake of fruits and vegetables were also calculated as a cumulative average of intake.

Height was ascertained in 1976. Information on regular exercise was ascertained on the 1980 questionnaire. Usual aspirin use was updated in 1982, 1984, and 1988, and usual alcohol intake was updated in 1984, 1986, and 1990. All other exposure variables (ie, smoking, body mass index, menopausal status, postmenopausal hormone use, and use of multivitamins) were updated on each follow-up questionnaire.

The relative risk (RR) of stroke was defined as the incidence rate of stroke among women in various categories of fish intake and in quintiles of omega-3 fatty acids intake divided by the corresponding rate among women in the lowest category of intake. Relative risks with 95% confidence intervals (CIs) were adjusted for age in 5-year categories, and Mantel-Haenszel tests for trend across the dietary categories were conducted by assigning median values for each category.

We also conducted analyses stratified by aspirin use to assess possible effect modification by this variable. Because aspirin reduces platelet aggregation by inhibiting synthesis of thromboxane A₂ in platelets, intakes of fish and omega-3 fatty acids might be expected to have a smaller effect in the prevention of stroke among regular aspirin users.²⁴ To adjust simultaneously for other cardiovascular risk factors and selected nutrient variables associated with risk of all stroke or stroke subtypes, we used pooled logistic regression with seven 2-year follow-up intervals.²⁵

Among 79 839 women followed up for 14 years, 574 incident cases of stroke occurred during 1086 261 person-years of follow-up. These strokes included 119 subarachnoid hemorrhages, 62 intraparenchymal hemorrhages, 303 ischemic strokes (264 thrombotic and 39 embolic), and 90 strokes of undetermined type. Among the thrombotic infarctions, there were 90 large-artery occlusive infarctions and 142 lacunar infarctions.

Table 1 shows selected cardiovascular risk factors, as well as intakes of selected nutrients and foods, by categories of fish and omega-3 fatty acids intake. Compared with women who ate fish less than once per month, women who ate fish 2 or more times per week were slightly older, had a lower prevalence of current smoking, and had a higher prevalence of overweight, hypertension, vigorous activity, and regular aspirin use and mutivitamin use. Fish intake was positively associated with total energy intake and intakes of chicken, egg, fruits and vegetables, and dairy foods, and it was inversely associated with intake of red meat. Moreover, fish intake was positively associated with dietary intakes of animal protein, calcium, and was inversely associated with intake of saturated fat, trans-unsaturated fat, and linoleic acid. Similar but weaker associations were observed between omega-3 fatty acids quintiles because the 2 highest categories of fish intake (ie, 2-4 and ≥5 times per week) mostly corresponded to the highest quintile of omega-3 fatty acids intake.

Table 2 presents RRs of total stroke and stroke subtypes according to fish intake. There were significant inverse associations between fish intake and age- and smoking-adjusted risk of total stroke, ischemic stroke, and thrombotic infarction, specifically lacunar infarction. After further adjustment for other cardiovascular and selected dietary variables, the inverse remained significant for thrombotic infarction and lacunar infarction, with a reduced risk of these stroke subtypes among women who ate fish 2 or more times per week. The multivariate RRs were 0.49 (95% CI, 0.26-0.93; P = .03), and 0.28 (95% CI, 0.12-0.67; P = .004), respectively. We found no excess risk of hemorrhagic stroke, either intraparenchymal or subarachnoid hemorrhage, among women who ate fish frequently.

Intake of omega-3 fatty acids was inversely associated with age- and smoking-adjusted risks of total stroke, hemorrhagic stroke, and subarachnoid hemorrhage (Table 3). Women in the highest quintile of omega-3 fatty acids intake had reduced risk of ischemic stroke, thrombotic infarction, and lacunar infarction although the trends did not reach statistical significance. There was no association between omega-3 fatty acids intake and risk of intraparenchymal hemorrhage. After further adjustment for other cardiovascular risk factors and selected dietary variables, women in the highest quintile of omega-3 fatty acids intake had significantly reduced risks of total stroke and lacunar infarction and a borderline reduction in risk of thrombotic infarction. The multivariate RRs in the highest vs the lowest quintiles of omega-3 fatty acids intake were 0.72 (95% CI, 0.53-0.99; P for trend = .12) for total stroke, 0.37 (95% CI, 0.19-0.73; P for trend = .004) for lacunar infarction, and 0.67 (95% CI, 0.42-1.07; P for trend = .19) for thrombotic infarction. The inverse associations of omega-3 fatty acids intake with risks of ischemic stroke and subarachnoid hemorrhage were no longer of statistical significance. We also conducted analyses using omega-3 fatty acids intake categories that corresponded to the same percentages as the fish intake categories (9%, <1 time per month; 34%, 1-3 times per month; 39%, 1 time per week; 15%, 2-4 times per week; and 3%, ≥5 times per week); using these categories the multivariate RRs of thrombotic infarction were 0.84 (95% CI, 0.48-1.47) for the second category, 0.63 (95% CI, 0.36-1.10) for the third category, 0.57 (95% CI, 0.29-1.13) for the fourth category, and 0.23 (95% CI, 0.05-1.08) for the highest category, P for trend = .04.

When we stratified our results by aspirin consumption, the inverse association between fish intake and risk of thrombotic infarction was more evident among women without regular aspirin use (Table 4). Women who did not use aspirin and were in the highest quintile of omega-3 fatty acids intake had a significantly reduced risk of thrombotic infarction whereas the trend among aspirin users was nonsignificant. Further classification of thrombotic infarction subtypes (large-artery occlusive infarction and lacunar infarction) was not possible due to the small number of cases in these strata. When we conducted the analysis according to omega-3 fatty acids intake categories using the same percentages as fish categories, the interaction between aspirin, omega-3 fatty acids, and risk of thrombotic infarction became more evident. Among women who did not use aspirin, the multivariate RRs of thrombotic infarction were 0.77 (95% CI, 0.39-1.50) for the second category, 0.56 (95% CI, 0.29-1.10) for the third category, and 0.34 (95% CI, 0.14-1.83) for the fourth category with no cases in the highest category (P for trend = .01). The respective RRs among regular aspirin users were 0.98 (95% CI, 0.34-2.85), 0.77 (95% CI, 0.27-2.21), 0.96 (95% CI, 0.29-3.17), and 0.75 (95% CI, 0.12-4.63) (P for trend = .94).

We observed a significant inverse association between fish intake and risk of stroke, primarily thrombotic stroke, after adjustment for cardiovascular risk factors and selected dietary variables. Risk of thrombotic infarction was significantly reduced by 48% among women who ate fish 2 to 4 times per week. The observed reduction in stroke risk associated with a relatively low frequency of fish intake is consistent with several previous prospective studies.^1- 2 In a Dutch study,¹ men who consumed more than 0.7 oz (20 g) of fish per day were at half the risk of total stroke as men who consumed less fish. The First National Health and Nutrition Examination Survey (NHANES I) Epidemiologic Follow-up Study² indicated that women who ate fish more than once a week were at about half the risk of total stroke as women who never ate fish. Women who ate fish once a week or less had a 22% to 23% lower risk than those who never ate fish; however, the trend did not reach statistical significance.² Two prospective studies among white men showed no significant relationship between fish intake and risk of total stroke.^3- 4 The Physicians' Health Study³ found an RR of 0.6 (95% CI, 0.3-1.6) for total stroke among men who ate fish 5 or more times per week compared with men who ate fish less than once a week. Of note, half of the men in the Physicians' Health Study were also taking aspirin, so the effect of fish intake would have been attenuated. The Chicago Western Electric study⁴ found an RR of 1.3 (95% CI, 0.7-2.2) among men who ate 1.2 oz (35 g) or more fish per day compared with men who ate no fish. This study relied only on death certificates and information from the Health Care Financing Administration for stroke ascertainment. In our study, the hypothesized inverse association with fish intake was seen primarily for ischemic stroke, specifically thrombotic infarction. Furthermore, we found a reduced risk associated with fish intake for lacunar infarction but not for large-artery occlusive infarction. We found no excess risk of hemorrhagic stroke, either subarachnoid hemorrhage or intraparenchymal hemorrhage, with fish intake.

We also examined the association between omega-3 fatty acids intake and risk of stroke and observed a reduced risk of total stroke among women in the highest quintile. Among stroke subtypes, the risk reduction was of borderline statistical significance for thrombotic infarction and of statistical significance for lacunar infarction. When stratified by aspirin use, women in the highest omega-3 fatty acids intake quintile who did not use aspirin had a significant 49% reduction in the risk of thrombotic stroke. These results support the hypothesis that omega-3 fatty acids are the protective component in fish that reduce the risk of thrombotic infarction. However, we cannot exclude the possibility that some other ingredients in fish may contribute to a reduction in risk or that some residual confounding by other risk factors remains.

Several mechanisms may be involved in the lower stroke risk associated with omega-3 fatty acids. A high-dose supplementation of these fatty acids (eg, 15 g/d of eicoapentanenoic acid) reduces the formation of thromboxane A₂ in platelets but does not substantially reduce the synthesis of prostaglandin I₂ in vascular endothelium cells, leading to reduced platelet aggregation without an adverse effect on vasodilation.^5- 6,26 Furthermore, eicosapentaenoic acid (in part converted from docosahexaenoic acid) is transformed into a nonaggregatory agent, thromboxane A₃, which increases the synthesis of a vasodilator, prostaglandin I₃, leading to further reductions in platelet aggregation and increased vasodilation.²⁶ These alterations of prostanoid metabolism are induced 3 to 4 days after the intake of fish oil supplement²⁷ and persist for 8 to 10 weeks after cessation of intake of the supplement.^28- 29 High-dose supplementation of omega-3 fatty acids (eg, 15 g/d) also lowers blood pressure levels in hypertensive persons,⁹ and reduces plasma fibrinogen concentrations in healthy volunteers.⁸ These effects may contribute to the prevention of atherosclerotic development and the thrombotic process.^9,30 In in vitro studies, omega-3 fatty acids decrease product generation by the 5-lipoxygenase pathway (such as leukotriene B₄) of neutrophil and monocytes and attenuate the leukotriene-mediated chemotaxis and endothelial-cell adherence of neutrophils.⁷ These fatty acids also reduce production of platelet-derived growth factorlike protein from vascular endothelial cells,³¹ which may attenuate proliferation of endothelial cells in the process of atherosclerosis.

In addition to these potential benefits leading to reduced risk of ischemic stroke, a decrease in whole blood viscosity^6,32 and an increase in capillary blood flow³³ associated with omega-3 fatty acids may play a role in the prevention of lacunar infarction, which involves small cerebral arteries in its pathogenesis.³⁴ Furthermore, dietary omega-3 fatty acids may reduce insulin resistance and glucose intolerance,^10,35 and this may reduce the risk of lacunar infarction because glucose intolerance and diabetes are strongly associated with risk of this event.^36- 37

In our study, the inverse association between estimated intake of omega-3 fatty acids and risk of thrombotic infarction was observed primarily among women who did not take aspirin regularly. Compared with omega-3 fatty acids, aspirin has a stronger inhibitory effect on synthesis of thromboxane A₂ in platelets; a single dose of aspirin (325 mg) reduces platelet aggregation for at least 3 days.³⁸ Reduced platelet aggregation may be one of the major factors in the prevention of thrombotic infarction. Thus, the strong effect of aspirin on the risk of thrombotic infarction is likely to obscure a moderate association between omega-3 fatty acids and the risk among women who take aspirin regularly.

We found no excess risk of either subarachnoid or intraparenchymal hemorrhage with intake of fish or omega-3 fatty acids in this population, which was not surprising. While bleeding time is prolonged when the intake of omega-3 fatty acids exceeds 3 g/d,^28,39- 40 a level of intake that corresponds approximately to the ingestion of fish 3 times or more per day. In this study of US women, fewer than 0.1% ate fish 3 times or more per day.

In conclusion, consumption of fish and omega-3 fatty acids was associated with a reduced risk of total stroke and thrombotic infarction primarily among women who did not take aspirin regularly. Consumption of fish and omega-3 fatty acids was not related to risk of hemorrhagic stroke. These results suggest that regular intake of fish may be beneficial for the prevention of thrombotic infarction in middle-aged US women.