Cereal, Fruit, and Vegetable Fiber Intake and the Risk of Cardiovascular Disease in Elderly Individuals FREE

Currently 35 million persons aged 65 years or older live in the United States, accounting for nearly 13% of the population.¹ These older adults are the fastest-growing segment of the population, and by 2030, it is projected that 70 million persons, or 1 in every 5, will be 65 years old or older.¹ Cardiovascular disease (CVD) is the leading cause of death and disability among these older adults, who also account for a disproportionately large share of the $200 billion annual US health care expenditures for CVD.² Better understanding of CVD risks and outcomes in this population is clearly of merit; however, relatively few studies have focused on such relationships in elderly persons, especially with regard to diet and CVD risk.

Dietary fiber, comprising nondigestible polysaccharides, naturally occurring resistant starch and oligosaccharides, and lignins in plants,³ has been associated with reduced incidence of ischemic heart disease (IHD) and stroke in predominantly middle-aged populations.^4- 11 Potential cardiovascular benefits of dietary fiber include effects on serum lipid levels, postprandial glucose and triglyceride levels, insulin sensitivity, and blood pressure,^12- 14 which may prevent or delay development of atherosclerosis in young adulthood and middle age. However, such influences may be less effective among elderly persons, when atherosclerosis is more advanced, so that dietary fiber consumption late in life may not be associated with CVD risk. However, the impact of dietary fiber intake on CVD risk has not been specifically evaluated among older adults. We therefore prospectively evaluated the association of dietary fiber consumption with risk of incident CVD in the Cardiovascular Health Study, a population-based, longitudinal cohort study of determinants of coronary heart disease and stroke among persons aged 65 years or older (mean age, 72 years at baseline). Based on epidemiologic evidence in predominantly middle-aged adults,^7- 10 our primary hypothesis was that greater consumption of cereal fiber, but not fruit or vegetable fiber, would be associated with lower risk of incident CVD in this elderly population.

The design and recruitment experience of the Cardiovascular Health Study have been previously described.^15- 16 Briefly, 5201 men and women aged 65 years or older were randomly selected and enrolled from Medicare eligibility lists in 4 US communities in 1989-1990; an additional 687 black participants enrolled in 1992 were not included in this analysis because a food frequency questionnaire was not administered to these individuals at baseline. Each center's institutional review committee approved the study and all participants gave informed consent. We excluded 1289 participants with known coronary heart disease, heart failure, stroke, transient ischemic attack, or carotid endarterectomy at baseline¹⁷ and 324 participants with incomplete data on cereal, fruit, or vegetable fiber consumption, resulting in a total of 3588 participants included in this analysis. At baseline, participants completed standardized questionnaires on health status, medical history, and cardiovascular risk factors, and underwent a clinical examination, resting electrocardiography, carotid ultrasonography, and laboratory evaluation.^15- 19

Usual dietary intake was assessed at baseline using a 99-item, picture-sort version of the National Cancer Institute food frequency questionnaire with 5 response categories ranging from less than 5 times per year to 5 times or more per week.^20- 22 Intake of cereal fiber, fruit fiber, and vegetable fiber was estimated using the frequency of consumption of specific fiber-containing foods, weighted by their average fiber content according to US Department of Agriculture (USDA) sources.²³ Foods were included if an average serving of that food provided at least 5% of the recommended daily intake of total fiber, using USDA dietary guidelines for an average 2000 kcal diet and USDA estimates of fiber content of various foods.²³ Summary dietary indexes, such as total energy, were estimated from responses to the food frequency questionnaire using computer software,²⁴ considering portion sizes as medium, and entering the frequency as the midpoint of the relevant response category. For total energy, the frequency was entered as the mean for all participants when there were only 1 or 2 missing food items (n = 346). Food and nutrient intakes, including fiber intake, were adjusted for total energy using regression analyses²⁵; values are presented for the mean population energy intake of 1820 kcal/d.

We evaluated the validity of the food frequency questionnaire in a subset of 79 participants, comparing fiber intake assessed by the food frequency questionnaire with 6 detailed 24-hour dietary recall interviews spaced approximately 1 month apart. Accounting for within-person variation in 24-hour intake,²⁶ the adjusted Pearson correlation coefficient for total dietary fiber intake as assessed by the food frequency questionnaire and the dietary recalls was 0.56 (P<.001). If log-transformed values were used, the adjusted Pearson correlation coefficient was 0.68 (P<.001). These correlations are comparable to other validated, widely used food frequency questionnaires.⁹

Potential cardiovascular events were identified during annual examinations and interim 6-month telephone interviews.²⁷ Centralized cardiac and stroke adjudication committees reviewed and classified events using information from interviews, medical records, physician questionnaires, death certificates, medical examiner forms, Health Care Financing Administration hospitalizations, and available computed tomographic or magnetic resonance imaging scans (reviewed centrally for 70% of strokes, plus an additional 17% with reports only).^27- 28 Myocardial infarction (MI) was classified using an algorithm including chest pain, cardiac enzymes, and electrocardiogram finding. Fatal events consistent with IHD death but not meeting criteria for definite fatal MI were classified as coronary heart disease death if occurring within 72 hours of chest pain or with an antecedent history of IHD. Stroke was defined as a neurological deficit of rapid onset lasting longer than 24 hours unless death supervened, or as a subarachnoid hemorrhage. Strokes were classified as "ischemic" if there was evidence of focal brain deficit without evidence for primary hemorrhage; as "hemorrhagic" if there was evidence of blood in the subarachnoid space, ventricles, or parenchyma seen on cerebral imaging, at surgery, or at autopsy that did not appear consistent with hemorrhage into an infarction, or if there was bloody spinal fluid on lumbar puncture; and as "unknown type" if information was insufficient for classification. Adjudication decisions were usually unanimous in terms of both presence of stroke and stroke type.

For this analysis, the primary outcome was incident CVD, defined as combined incident stroke, fatal and nonfatal MI, and coronary heart disease death. We also planned to separately examine IHD death (fatal MI plus coronary heart disease death), nonfatal MI, total stroke, and stroke subtypes.

Fiber intake was evaluated in quintiles as indicator (dummy) variables and as a continuous variable. Cox proportional hazards models were used to estimate risk, with time calculated from study entry until the earliest of the first event under consideration, death, or June 30, 2000. To assess for confounding, multivariate Cox models were used. Prespecified covariates included age and sex. Other risk factors and dietary characteristics were added, both individually and in groups, based on clinical interest or if they appreciably altered (±3%) the hazard ratio (HR) associated with cereal, fruit, or vegetable fiber consumption. Tests for linear trend were calculated by assigning the medians of intake in quintiles treated as a continuous variable. Potential effect modification was assessed for age, sex, and diabetes using likelihood ratio tests and cross-products (interaction terms) of exposures and covariates. All P values are 2-tailed and <.05 was the level of significance. Analyses were performed using Stata 7.0 (Stata Corp, College Station, Tex), using the updated Cardiovascular Health Study database incorporating minor corrections through April 2002.

Average consumption of cereal fiber was 4.2 g/d; fruit fiber, 5.2 g/d; vegetable fiber, 6.9 g/d; and total fiber, 16.2 g/d. The main foods contributing to cereal fiber intake were dark breads and high fiber or bran cereals; to fruit fiber intake, apples, oranges, and bananas; and to vegetable fiber intake, beans, broccoli, peas, corn, and cauliflower. Fiber intake varied substantially, with average intake in the highest quintile being 7.9, 9.1, and 11.7 g/d for cereal, fruit, and vegetable fiber, compared with 0.8, 1.7, and 2.9 g/d in the lowest quintile, respectively. Greater fiber intake from each source was modestly correlated with fiber intake from the other 2 sources (between cereal and fruit fiber intake, Spearman correlation [r] = 0.16, P<.001; between cereal and vegetable fiber intake, r = 0.12, P<.001; and between fruit and vegetable fiber intake, r = 0.24, P<.001).

Table 1 presents selected baseline characteristics according to cereal, fruit, and vegetable fiber consumption. Cereal, fruit, and vegetable fiber intake were each positively associated with female sex, education, and exercise intensity, and inversely associated with current smoking, ever smoking, and pack-years of smoking. Each was inversely associated with intake of saturated fat, alcohol, and beef/pork, and positively associated with intake of fish, fruits, and vegetables. There were only slight differences in income, blood pressure, prevalence of diabetes or atrial fibrillation, internal carotid artery intimal medial thickness, body mass index, self-perceived health, medication use, and serum lipid and fasting glucose levels.

During 8.6 years mean follow-up, there were 811 incident CVD events (159 IHD deaths, 308 nonfatal MIs, and 344 strokes, including 250 ischemic, 72 hemorrhagic, and 22 unclassified strokes). Table 2 presents the unadjusted and adjusted risks of incident CVD according to cereal, fruit, and vegetable fiber and total fiber intake. After adjustment for age, sex, education, diabetes, ever smoking, pack-years of smoking, daily physical activity, exercise intensity, alcohol intake, fruit fiber intake, and vegetable fiber intake, cereal fiber consumption was inversely associated with incident CVD (P for trend = .02), with 21% lower risk (HR, 0.79; 95% confidence interval [CI], 0.62-0.99) in the highest quintile of intake compared with the lowest quintile. After similar adjustments, fruit fiber consumption (P for trend = .98), vegetable fiber consumption (P for trend = .95), and total fiber consumption (P for trend = .23) were not associated with risk of incident CVD (Table 2).

Because potential cardiovascular effects of dietary fiber may be mediated by effects on weight gain, blood pressure, glucose or serum lipid levels, as well as by dietary substitution effects (reducing intake of potentially detrimental foods), we evaluated the extent to which differences in some of these factors might mediate associations of cereal fiber intake with CVD risk. After further adjustment for body mass index, waist-hip ratio, systolic blood pressure, fasting glucose, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, C-reactive protein, saturated fat intake, meat intake, and fish intake, the association of cereal fiber consumption with CVD risk was only slightly mitigated, although no longer significant (P for trend = .07 across quintiles, with 19% lower risk in the highest quintile compared with the lowest quintile [HR, 0.81; 95% CI, 0.62-1.06]).

Because associations of cereal fiber consumption with CVD risk appeared to be graded and continuous, we also evaluated cereal, fruit, and vegetable fiber intake as linear (continuous) exposures. To describe a meaningful difference in dietary intake, we compared the 80th percentile with the 20th percentile of intake for each, approximately equivalent to the fiber content of 2 slices of whole grain bread per day (4.6 g/d) for cereal fiber, 1 large apple per day (4.7 g/d) for fruit fiber, and 1 medium stalk of cooked broccoli per day (5.0 g/d) for vegetable fiber. After adjustment for potential confounders (as in Model 2, Table 2), cereal fiber consumption was associated with a 14% lower risk of incident CVD (HR, 0.86; 95% CI, 0.75-0.99), comparing the 80th percentile with the 20th percentile of intake. In similar analyses, neither fruit fiber intake (HR, 1.00; 95% CI, 0.87-1.14) nor vegetable fiber intake (HR, 1.02; 95% CI, 0.91-1.15) were associated with incident CVD.

Further adjustment for a wide variety of other demographic, clinical, laboratory, and dietary characteristics had little effect on these results, including each of the characteristics listed in Table 1 as well as recruitment community; waist size; treated hypertension; diastolic blood pressure; estrogen use; claudication; family history of MI; fasting insulin, 2-hour oral glucose tolerance; fibrinogen; factor VII; factor VIII; lipoprotein(a); serum potassium and creatinine; and estimated intake of total fat, polyunsaturated fat, protein, carbohydrates, fried chicken or french fries, wine, vitamin A, carotene, vitamin C, thiamine, and niacin. There was also little evidence that these relationships varied according to age, sex, or diabetes status (Table 3), although the results are no longer significant in the subgroup analyses.

Because only a few food groups contributed to cereal fiber intake, we evaluated in post hoc analyses whether the observed lower CVD risk was related to fiber from any specific food group. Comparing the 80th percentile with the 20th percentile of intake and adjusting for potential confounders (as in Model 2, Table 2), the lower risk appeared predominantly related to fiber intake from dark breads (such as whole wheat, rye, or pumpernickel) (HR, 0.76; 95% CI, 0.64-0.90), rather than fiber from high fiber, bran, or granola cereals (HR, 0.99; 95% CI, 0.84-1.17), other cold cereals (HR, 0.98; 95% CI, 0.94-1.02), or cooked cereals (HR, 1.01; 95% CI, 0.92-1.11).

We also examined relationships of fiber consumption with risk of different CVD events. Because censoring occurred only for the type of event being examined, a total of 220 IHD deaths, 331 incident nonfatal MIs, 392 incident total strokes, 291 incident ischemic strokes, and 95 incident hemorrhagic strokes were included in these analyses. Comparing the 80th percentile with the 20th percentile of intake, higher cereal fiber intake was associated with lower risk of total stroke (HR, 0.78; 95% CI, 0.64-0.95) and ischemic stroke (HR, 0.76; 95% CI, 0.60-0.95), with CIs including unity for risk of IHD death (HR, 0.87; 95% CI, 0.67-1.13), nonfatal MI (HR, 0.94; 95% CI, 0.76-1.16), and hemorrhagic stroke (HR, 0.92; 95% CI, 0.62-1.37), after adjustment for potential confounders (as in Model 2, Table 2). In similar analyses, neither fruit fiber intake nor vegetable fiber intake was associated with any of the different CVD events (data not shown), except IHD death with which fruit fiber intake was associated with higher risk (HR, 1.32; 95% CI, 1.02-1.72). In post hoc analyses, this observed higher risk did not appear predominantly associated with fiber intake from any single fruit or group of fruits (data not shown). However, the estimate of risk was somewhat higher among persons with impaired glucose tolerance (HR, 1.39; 95% CI, 0.82-2.38) or treated diabetes (HR, 1.38; 95% CI, 0.90-2.10) compared with persons who had normal glucose tolerance (HR, 1.13; 95% CI, 0.74-1.73), although power was limited to confirm differences between these groups, and the risk was not statistically significant.

We observed an inverse association between consumption of fiber from cereal sources (including whole grains and bran) and risk of incident CVD in this large, population-based cohort of elderly men and women (average age, 72 years at baseline). While the observed difference in risk was not large, it was seen with a fairly modest difference in dietary intake, approximately equal to 2 slices of whole grain bread per day. Compared with medical or surgical interventions, nutritional changes are relatively low risk, low cost, and widely available. Therefore, the practical importance of even a small change in risk may be significant on a population or public health level.

There are several potential explanations for the observed relationship. Intake of cereal fiber may be a marker for some other behavior or clinical characteristic that reduces cardiovascular risk in this population. However, the lower risk was not greatly altered by adjustment for a wide variety of demographic, clinical, lifestyle, laboratory, and dietary characteristics. Additionally, the lower risk was specific for cereal fiber, not fruit or vegetable fiber, which would be difficult to explain by confounding alone since both fruit and vegetable fiber intake demonstrated similar associations with potential confounders—such as healthier lifestyles (greater exercise intensity, less smoking), healthier dietary behaviors (greater fish intake, less meat intake, less saturated fat intake), and lower risk demographic profiles (female sex, greater education)—but were not associated with lower risk.

There are biologically plausible mechanisms for beneficial effects of cereal fiber intake on CVD risk in elderly individuals. In experimental studies, fiber intake has been shown to affect serum lipid levels, postprandial absorption, blood pressure, and insulin sensitivity.^12- 14 Dietary fiber may also affect fibrinolysis and coagulation,^29- 30 which may be particularly important in the setting of established atherosclerotic plaques. Moreover, cereal fiber intake benefits endothelial cell function: daily intake of whole grain oat or wheat cereal for 1 month prevents impaired vascular reactivity in response to a high-fat meal.³¹ Cereal fiber consumption may also reduce CVD risk via a substitution effect, replacing intake of other foods having potentially detrimental effects. In our study, cereal fiber consumption was associated with slight differences in blood pressure, serum lipid and glucose measures, as well as lower meat and saturated fat intake; however, the inverse association of cereal fiber intake with CVD risk was only slightly mitigated by adjustment for these characteristics, as well as adjustment for body mass index, waist-hip ratio, diabetes status, and coagulation factors, suggesting that the lower risk was not entirely mediated through baseline differences in these characteristics.

The only prior study to evaluate associations of cereal fiber intake with total CVD did not observe lower risk¹¹; however, coronary revascularization procedures were included as CVD events, which may have produced bias toward the null if persons consuming cereal fiber were more likely to seek evaluation for symptoms, have a diagnostic work-up, or undergo revascularization. Coronary revascularization was not a prespecified major end point in the Cardiovascular Health Study due to potential health care utilization bias or selection bias, particularly among elderly persons, and therefore was not evaluated in our study. Detection of associations in the prior study may also have been limited by the relatively narrow range of cereal fiber intake among these female health professionals, with average intake in the highest quintile only approximately twice that of the lowest quintile; in our population-based cohort, in comparison, there was a 10-fold difference in average cereal fiber intake from the lowest quintile to the highest.

In a post hoc analysis, modest intake of cereal fiber from dark breads such as wheat, rye, or pumpernickel appeared to be associated with a lower risk of incident CVD. Prior studies have observed inverse associations of dark breads and whole-grain breakfast cereals with fatal and nonfatal MI.^32- 33 It is possible that differences in measurement error or bias between assessment of dark bread intake and other sources of cereal fiber may explain the observed relationships in our study. This preliminary finding requires confirmation in future studies.

We did not observe associations of fruit or vegetable fiber consumption with CVD risk. While modest associations cannot be excluded by our findings, our observations are consistent with prior studies in which consumption of cereal fiber, but not fruit or vegetable fiber, was independently associated with lower risk of coronary heart disease or stroke in predominantly middle-aged adults.^7- 10 These findings may explain in part why others did not detect associations of total dietary fiber intake with IHD risk^34- 36 and suggest that effects of dietary fiber may vary depending on the food source. However, biologic mechanisms for such observed differences are unclear. Our findings support the need to investigate potential mechanisms and alternative explanations to account for these observed differences.

When we examined risk of different CVD events separately, we observed inverse associations of cereal fiber consumption with risk of total stroke and ischemic stroke, and a trend toward inverse associations with risk of IHD death. Few prior studies have reported relationships of fiber consumption with risk of stroke or stroke subtypes. In the Health Professionals Follow-up Study, only cereal fiber, not fruit or vegetable fiber, was inversely associated with risk of total stroke.¹⁰ Cereal fiber intake has been associated with lower risk of IHD, especially fatal IHD, in prior studies of predominantly middle-aged adults^7- 9; our findings suggest the possibility of a similar association among older adults. In a randomized clinical trial, advice to increase cereal fiber intake did not affect recurrent MI or mortality³⁷; however, these were all middle-aged men recovering from recent MI, follow-up was limited to 2 years, and more than half of eligible participants were excluded because they were already consuming foods similar to the intervention diets. Randomized trials of cereal fiber intake for primary prevention of IHD or stroke have not been reported.

While we observed no association between fruit fiber consumption and total CVD risk in these elderly adults, there was an unexpected positive association between fruit fiber consumption and risk of IHD death, with subsequent analyses suggesting a stronger association among those with diabetes or impaired glucose tolerance, though with limited power to confirm this finding. Notably, prior studies in predominantly middle-aged adults did not observe increased IHD risk with fruit fiber intake.^7- 9,11 Given multiple comparisons of different CVD events and lack of consistency with prior studies, residual confounding or chance may account for this observed association, which should be viewed as an unexplained finding. Further investigation of this relationship, especially among older adults, may be warranted.

Our analysis has several strengths. The prospective design and exclusion of persons with known CVD at baseline reduce potential for bias from recall differences or dietary changes due to known disease. The population-based recruitment strategy enhances generalizability, and the broad range of fiber intake improves the ability to detect dose-response relationships. Standardized assessment of a wide variety of participant characteristics increases the capacity to adjust for confounding. Close follow-up, comprehensive review of potential events, and centralized adjudication reduce potential for missed or misclassified outcomes.

There are also potential limitations to our findings. Dietary fiber intake was assessed at baseline, and there may have been changes in fiber consumption over time. Also, intake of various foods was assessed categorically, and while dietary intake was probably ranked correctly at the extremes, misclassification may have occurred on conversion to continuous estimates, especially at higher levels of intake. In addition, only significant sources of dietary fiber intake, rather than all possible sources, were included in our measures. Without apparent bias, such resulting misclassification would diminish the ability to detect associations with CVD risk; our findings may therefore underestimate true differences in risk associated with fiber consumption. Dietary recall data were not available for separate validation of cereal, fruit, and vegetable fiber intake; it is possible that the observed null results for fruit and vegetable fiber represent less accurate assessments of their intake. We evaluated only cardiovascular events, and fiber intake may affect other outcomes, such as gastric cancer or diverticulosis.^38- 39 Also, our aim was to evaluate the fiber content of these foods, not the whole food or other constituents of these foods (such as vitamins and micronutrients), which may have different relationships with risk. We also did not have information on baseline use of multivitamin, nutritional, or fiber supplements; it is unknown how such supplement use might affect associations of dietary fiber intake with CVD risk. Finally, residual confounding by incompletely measured or unknown factors cannot be excluded.

Given the rapidly growing numbers of elderly adults in industrialized populations, it is increasingly important to examine relationships of diet with cardiovascular events among older adults. Our results suggest that dietary habits may affect cardiovascular risk beyond the earlier development and progression of disease in young adulthood and middle-age, supporting recommendations for increased consumption of dietary fiber from cereal and whole grain sources among older adults. Continued investigation is needed to further elucidate relationships and mechanisms of benefit and risk, with particular focus on fiber from cereal and whole grain sources.