Cardiovascular disease is the leading cause of death and disability in older adults, the most rapidly growing segment of the population. While n-3 fatty acids in fish have several effects that may reduce stroke risk,1-4 results of previous studies of fish intake and stroke incidence have been inconsistent,5-10 and none have focused on elderly individuals, in whom disease burden may be high and dietary habits may have less influence. Health effects of fish intake may also vary with the type of fish meal consumed, with effects differing for intake of tuna or other broiled or baked fish compared with fried fish.11 We evaluated the relationship between fish consumption and risk of stroke in the Cardiovascular Health Study, a population-based, longitudinal cohort study of older adults.
Patients and methods
We studied 4,778 men and women aged 65 or older who were free of known cerebrovascular disease at baseline. Subjects were randomly selected and enrolled from Medicare eligibility lists in four US communities from 1989 to 1990. Medical history and cardiovascular risk factors were ascertained, and subjects underwent physical examination and laboratory testing. Usual dietary intake was assessed at baseline using a food frequency questionnaire, including questions on tuna fish, other fish (broiled or baked), and fried fish or fish sandwiches (fried fish burgers). In a subset of participants, intake of tuna/other fish correlated with plasma phospholipid concentrations of long-chain n-3 fatty acids (r = 0.59; P < .001), while intake of fried fish/fish sandwiches did not (r = 0.04; P = .78),11 consistent with the lean, white fish (eg, cod) typically used for these fried fish meals.
Incident strokes were identified during annual examinations, phone contacts, and reviews of hospital discharge records and centrally adjudicated using medical records, death certificates, medical examiner forms, Centers for Medicare and Medicaid Services hospitalizations, and available computed tomography or magnetic resonance imaging scans. Cox proportional-hazards models were used to evaluate the associations between fish intake and risk of stroke. Statistical tests were two-tailed (a = 0.05).
Fried fish/fish sandwich consumption was associated with male sex, nonwhite race, and lower education, while tuna/other fish consumption was associated with younger age, female sex, and higher education (Table 1). In general, tuna/other fish consumption was associated with a more favorable cardiovascular risk profile; however, body mass index and low-density lipoprotein (LDL) cholesterol levels were higher with greater intake. Associations with other dietary habits are also shown in Table 1.
During 12 years of follow-up, 626 incident strokes were documented. The incidence of stroke among subjects who consumed tuna/other fish five or more times a week was 12 per 1,000 person-years compared with 19 per 1,000 person-years among those who consumed this type of fish meal less than once a month (P = .002; Figure). Among those consuming fried fish/fish sandwiches one or more times a week, the incidence of stroke was 17 per 1,000 person-years versus 13 per 1,000 person-years for those who consumed fried fish meals less than once a month (P = .02; Figure).
After adjusting for potential confounding factors, tuna/other fish intake was inversely associated with risk of stroke, with a trend toward a 13% lower risk with consumption one to three times per month, a 25% lower risk with consumption one to four times a week, and a 26% lower risk with consumption five or more times a week, compared with less than once a month (P for trend = .04; Table 2). When stroke subtypes were separately analyzed, we found an inverse association between tuna/other fish intake and ischemic stroke, with a trend toward a 13% lower risk with consumption one to three times a month, a 27% lower risk with consumption one to four times a week, and a 30% lower risk with consumption five or more times per week, compared with less than once a month (P for trend = .02). Tuna/other fish intake was not significantly associated with risk of hemorrhagic stroke.
Conversely, fried fish/fish sandwich consumption was associated with a higher risk of stroke. In multivariate-adjusted analyses, consuming fried fish/fish sandwiches at least once a week was associated with 37% higher risk of total stroke and 44% higher risk of ischemic stroke compared with consumption less than once a month (P for trend = .006 for total stroke and .003 for ischemic stroke; Table 2). When evaluated continuously, each one serving per week predicted a 10% higher risk of total stroke (hazard ratio [HR], 1.10; 95% confidence interval [CI], 1.02—1.19) and a 13% higher risk of ischemic stroke (HR, 1.13; 95% CI, 1.04–1.22; adjustments as in Table 2).
Further adjustment for factors that might mediate associations between fish intake and stroke risk (systolic blood pressure, LDL cholesterol, high-density lipoprotein cholesterol, triglycerides, C-reactive protein) or exclusion of participants who used fish oil supplements (n = 178) or who had atrial fibrillation (n = 261) at baseline had little effect on these results (data not shown). Further, there was little evidence that findings varied with age, sex, education, diabetes, prevalent coronary heart disease, treated hypertension, systolic blood pressure, smoking, or aspirin use (P > .05 for each interaction).
Among these older adults, consumption of tuna or other broiled or baked fish was associated with a lower incidence of stroke. While these observational findings do not prove causality, this relationship was present after adjustment for a variety of demographic, clinical, and lifestyle factors. Furthermore, there are plausible biologic mechanisms that may explain the beneficial effects of tuna/other fish consumption on the incidence of stroke, particularly ischemic stroke. Fish oil has been found to influence blood pressure, serum lipid levels, inflammatory responses, endothelial cell function, and cerebral arteriolar reactivity in experimental studies.1-4 Each of these effects, alone or in combination, may reduce the risk of ischemic stroke.
Compared with no intake or infrequent intake (less than once a month), frequent intake of tuna/other fish (five or more times a week) was associated with a reduction in risk that was similar to that found with modest intake (one to four times times a week), with an approximately 25% lower risk of total stroke and a 30% lower risk of ischemic stroke. This suggests a potential threshold effect for the benefits of fatty fish intake at approximately one serving a week. Such a threshold effect, also suggested by a previous report,10 is compatible with the time course of incorporation of n-3 fatty acids from fish into cell membranes, whereby a single meal can produce changes lasting for days to weeks.12
Previous ecologic studies have suggested that high seafood intake is associated with a higher risk of hemorrhagic stroke, possibly because of antithrombotic effects.13 However, we did not observe a higher risk of hemorrhagic stroke with fish intake. Experimental studies suggest that fish oil only affects bleeding times at high doses (3—15 g daily),14 much higher than the typical dietary fish intake. Although the few numbers of hemorrhagic strokes in our study limit definitive conclusions, our results, together with those of prior cohort studies,9-10 suggest that modest tuna/other fish consumption does not significantly increase the risk of hemorrhagic stroke, particularly relative to the potential reduction in ischemic stroke risk.
Consumption of fried fish/fish sandwiches was associated with a higher risk of stroke, particularly ischemic stroke. These results are consistent with positive associations between fried fish/fish sandwich intake and risk of ischemic heart disease reported previously.11 Given the lack of correlation between fried fish intake and plasma phospholipid n-3 fatty acid levels, it is likely that most of these fish meals were made with lean, white fish, which is low in fish oil content. The observed higher stroke risk could, therefore, be due to an unfavorable risk—benefit ratio, with (1) relatively little benefit because of the low fish oil content and (2) potentially harmful effects from possible contaminants in the fish (eg, mercury) or from transfatty acids and lipid oxidation products in the fried fats and oils, particularly partially hydrogenated fats and oils used repeatedly for frying.15 Fried fish/fish sandwich consumption may also be a marker for some other factor that increases cardiovascular risk; however, the observed relationships were present after adjustment for a variety of demographic, clinical, and lifestyle characteristics. Further research is necessary to confirm these observations, to determine if they are specific for certain kinds of fish or frying oils, and to examine potential mechanisms.
Several characteristics of our study design enhance the validity of our findings. The prospective design as well as the exclusion of those with known cerebrovascular disease at baseline reduce potential biases from recall differences or dietary changes due to known disease. Standardized assessment of a wide variety of subject characteristics increased the ability to adjust for confounding factors. Close follow-up and centralized adjudication reduced the potential for missed or misclassified outcomes. The population-based recruitment strategy enhanced generalizability.
Potential limitations are also present. Usual fish intake was assessed at the beginning of the study and may have changed over time. Additionally, the food frequency questionnaire assessed categories of fish intake, rather than precise frequencies or quantities. These factors would produce error (misclassification) in the measurement of fish intake, which would bias results toward the null. For fried fish intake, we could not separately evaluate the type of fish versus the preparation method. Although we adjusted for a variety of participant characteristics, residual confounding by unmeasured or imprecisely measured factors cannot be excluded.
The proportion of older adults in the United States and most other nations is rising. Because cardiovascular disease risk increases dramatically with increasing age, determining which factors are related to healthy aging is of critical importance. We assessed the relationship between dietary habits and stroke risk late in life, with an average baseline age of 73 years and 12 years of follow-up. While the associations we observed may reflect dietary habits at earlier ages, our findings suggest that diet may influence stroke
risk later in life. Our results also indicate that the fish oil content or preparation method may be important when considering the effects of fish intake on stroke risk. Continued investigation of these relationships, including potential mechanisms of benefit and risk, is indicated.