It has been suggested that the consumption of plant foods such as fruits and berries, vegetables and cereals provides protection against cerebrovascular diseases. These foods are rich sources of fibre, K and antioxidant compounds, which through different mechanisms may have protective effects against stroke(Reference Khaw and Barrett-Connor1). Although a recent meta-analysis suggested a reduced risk of stroke for those with an ample daily consumption of fruits or vegetables(Reference He, Nowson and MacGregor2), the original cohort studies examining the prediction of fruit and vegetable consumption on the risk of stroke have given inconsistent results(Reference Ness and Powles3–Reference Yochum, Kushi and Meyer13). Studies linking wholegrain or bran consumption with the risk of cerebrovascular diseases are also inconclusive, with some suggesting a reduced risk(Reference Liu, Manson and Stampfer14, Reference Mink, Scrafford and Barraj15) and others no association(Reference Ascherio, Rimm and Hernan4, Reference Key, Thorogood and Appleby5, Reference Steffen, Jacobs and Stevens16).
Since some of the risk factors of ischaemic stroke and haemorrhagic stroke differ, the possible effects of fruits and vegetables on the development of these diseases may also be different. Most cohort studies to date have considered either total stroke or ischaemic stroke as the sole outcome variable, and few have considered haemorrhagic strokes. The results from these studies are inconsistent, suggesting either an inverse association between fruit or vegetable intake and haemorrhagic stroke incidence(Reference Sauvaget, Nagano and Allen6, Reference Hirvonen, Virtamo and Korhonen7) or no significant association(Reference Gillman, Cupples and Gagnon8, Reference Yokoyama, Date and Kokubo9). It has been suggested that the intake of cereal fibre and total stroke or haemorrhagic stroke are inversely associated(Reference Ascherio, Rimm and Hernan4, Reference Oh, Hu and Cho17), but no studies on the consumption of different types of cereals and haemorrhagic stroke exist.
The inconsistency of the published results suggests that the possible protection may be either present only for some plant foods (or certain nutrients in them or their combination), or at specific levels of the plant food intake, or only in some subpopulations (for example, sex, age or lifestyle). For this reason we initiated the present study, which focuses on the prediction of plant food intake, including fruits and berries, vegetables and cereals, on the incidence of cerebrovascular diseases and its subgroups, ischaemic stroke and intracerebral haemorrhage, in subpopulations of a Finnish cohort with a low intake of fruits and vegetables and high intake of wholegrain cereals.
Methods
Subjects
The present study was based on the Finnish Mobile Clinic Health Examination Survey, carried out during 1968–72. The survey included a dietary history interview of 10 054 men and women, aged 15 years and over. After excluding those under 40 or over 74 years of age and those suffering from cardiovascular diseases before baseline examination, the final cohort comprised 3932 men and women.
The present study was conducted according to the guidelines laid down in the Declaration of Helsinki. Data on population samples held at the National Institute for Health and Welfare can be used on the basis of the law on Personal Data and the legislation applying to the National Institute for Health and Welfare.
Baseline examination
All the participants completed a questionnaire that was sent to their home, which was then checked at the baseline examination. The questionnaire yielded information on the respondent's occupation, previous and current diseases, use of medicines, lifestyle, i.e. leisure-time physical activity (high physical activity v. low physical activity) and smoking. Subjects were classified according to smoking status as never smokers, ex-smokers, smokers of cigars or pipe only, smokers of fewer than fifteen cigarettes per d, and smokers of fifteen cigarettes or more per d. Those who had never smoked and ex-smokers were combined to form a category of non-smokers. A health examination included measurement of height and weight, from which the BMI was calculated. A BMI of under 28 kg/m2 was considered to represent normal weight and a BMI over 28 kg/m2 overweight. Blood pressure was registered using the auscultatory method after a 5 min rest(Reference Aromaa18). Four hypertension categories were formed on the basis of systolic blood pressure (SBP), diastolic blood pressure (DBP) and antihypertensive medication(Reference Knekt19). Individuals with SBP ≥ 170 mmHg and DPB ≥ 100 mmHg and individuals using antihypertensive medication were considered definitely hypertensive. Individuals with SBP ≥ 160 mmHg and DBP ≥ 95 mmHg but not defined as definitely hypertensive were considered to have mild hypertension, and those with SBP < 140 mmHg and DBP < 90 mmHg were considered normotensive. All individuals with intermediate values were considered to have borderline hypertension. The serum cholesterol level was determined with an autoanalyser modification of the Liebermann–Burchard reaction(Reference Huang, Chen and Wefler20). Known cases of diabetes were identified from information given by the participants and a modified glucose tolerance test was carried out to diagnose new diabetes cases at baseline(Reference Montonen21, Reference Reunanen, Aromaa and Pyörälä22).
Dietary assessment
At a baseline dietary history interview, the subjects were asked about their usual diet during the previous year(Reference Järvinen, Seppänen and Knekt23). The interview was carried out by trained staff, which consisted of personnel with an education in home economics or nutrition. A preformed questionnaire accompanied by food models was utilised to help in estimating of the amounts of food consumed. The questions were mostly open-ended, so subjects could also describe their diet freely and in detail. Subjects reported their food consumption on a daily, weekly or monthly basis. The ingredients of mixed foods were broken down into their components using a recipe file and the consumption of fruits and berries, vegetables, and cereals as well as subgroups of them, and separate food items were calculated per d. Energy intake was calculated based on the intake of protein, fat and available carbohydrate according to Finnish food composition tables(Reference Rastas, Seppänen and Knuts24).
Repeatability of the dietary history method was estimated after 4–8 months (short term) and 4–7 years (long term) by repeating the interview(Reference Järvinen, Seppänen and Knekt23). The intra-class correlation coefficients for short-term repeatability were 0·63 for total vegetables without potatoes (0·50 for cruciferous vegetables, 0·44 for legumes and 0·59 for root vegetables), 0·62 for cereals, 0·64 for fruits and 0·36 for berries. The corresponding coefficients for long-term repeatability were 0·47 for vegetables without potatoes (0·35 for cruciferous vegetables, 0·26 for legumes and 0·39 for root vegetables), 0·39 for cereals, 0·41 for fruits and 0·10 for berries.
Morbidity and mortality
Disease events leading to hospitalisation were identified by linking data from the Finnish Hospital Discharge Register from 1970–94 to the dietary data(Reference Heliövaara, Reunanen and Aromaa25). Information on mortality was based on death certificates obtained for all the deceased from Statistics Finland(Reference Reunanen, Aromaa and Pyörälä22). The primary diagnoses of hospital records and the principal causes of death were classified as cerebrovascular diseases using the three digit codes 344 or 430–438 of the International Classification of Diseases, eighth revision (ICD-8)(26). The diagnoses were further classified as total strokes (ICD-8 codes 430–438, 344) including all acute strokes, subarachnoidal haemorrhages and other, undefined strokes; ischaemic stroke (ICD-8 codes 432–434) and intracerebral haemorrhage (ICD-8 code 431). During the 24-year follow-up 625 cases (334 in men and 291 in women) of cerebrovascular disease occurred, of which fifty-eight were intracerebral haemorrhages (ICD-8 code 431) and 335 were thrombotic or embolic occlusions in precerebral or cerebral arteries (ICD-8 codes 432–434), defined in the present study as ischaemic strokes. The rest of the 232 cases were other or undefined cerebrovascular diseases.
Statistical methods
The relative risks of cerebrovascular diseases between quartiles of different plant food intake (fruits and berries, vegetables, and cereals) were estimated using Cox's model(Reference Cox27). The follow-up time was defined as the time from the baseline examination to the date of hospitalisation for cerebrovascular disease, death or the end of follow-up, whichever came first. Adjustments for variables that fulfilled the criteria for confounding factors(Reference Breslow and Day28) were made by including them in the models. Three main models were defined: (1) a model adjusted for age (continuous) and sex; (2) model 1 further adjusted for the non-dietary risk factors of BMI (continuous), smoking (non-smokers, current smokers), physical activity (high, low), serum cholesterol (continuous), hypertension (normotensive, borderline hypertensive, mildly hypertensive, definitely hypertensive) and total energy intake (continuous); (3) model 2 further adjusted by including all three plant food groups (fruits and berries, vegetables, and cereals) simultaneously. A test for trend was carried out based on the likelihood ratio test by including the quartile of plant food intake as a continuous variable in the model. Potential effect-modifying factors (i.e. sex and hypertension) were studied by including interaction terms between them and the different plant foods in the model. Finally, exclusion of the first 4 years of follow-up was carried out to control for the potential effect of undiagnosed cerebrovascular diseases at baseline on results. The analyses were performed by using SAS 9 (SAS Institute Inc., Cary, NC, USA).
Results
Individuals who experienced a stroke during the follow-up tended to be older and to have a higher BMI, higher serum cholesterol levels and hypertension (Table 1). Older age, smoking and low physical activity were associated with a lower consumption of fruit and berries (Table 2). Older age and low physical activity were also related to a lower consumption of vegetables. Furthermore, there was a higher prevalence of diabetes among those with the lowest vegetable intake. Those who consumed the fewest cereals were older and more often smokers, and had lower physical activity, higher serum cholesterol level, and a higher prevalence of hypertension and diabetes. Compared with women, men tended to eat fewer fruits and berries, and vegetables, but more cereals (Table 3).
Table 1 Age- and sex-adjusted levels of selected variables for individuals who did and did not experience some type of stroke during follow-up
(Mean values or percentages)
* Significance of the differences between the groups.
† Adjusted for sex.
‡ Adjusted for age.
§ Including potatoes.
∥ Including fresh, dried and canned fruits, and juices.
¶ Including only fresh berries.
Table 2 Age- and sex-adjusted levels of selected variables in the lowest and highest quartiles of intake of fruits and berries, vegetables or cereals
(Mean values or percentages)
* P value for trend.
† Adjusted for sex.
‡ Adjusted for age.
§ Including potatoes.
∥ Including fresh, dried and canned fruits, and juices.
¶ Including only fresh berries.
Table 3 Intakes (g/d) of selected plant foods in men and women
(Mean values and ranges)
* Including fresh, dried and canned fruits, fresh berries, and juices, jams and marmalades.
† Including fresh, dried and canned fruits, and juices.
‡ Including only fresh berries.
§ Including juices.
∥ Excluding potatoes.
Fruits and berries
An inverse association was found between total consumption of fruits and berries and subsequent incidence of cerebrovascular diseases (Table 4). This association was especially strong for fruits. The relative risk between the highest and lowest quartiles of fruit consumption was 0·75 (95 % CI 0·59, 0·94; P for trend = 0·006). The association was similar both for ischaemic stroke and intracerebral haemorrhage, the corresponding relative risks being 0·73 (95 % CI 0·54, 1·00; P = 0·03) and 0·47 (95 % CI 0·24, 0·92; P = 0·02), respectively. These associations were mainly due to the intake of citrus fruits (including juices) and no significant association was found for other fruits or berries. All results persisted after exclusion of cases occurring during the first 4 years of follow-up and after inclusion of cases occurring during the first 15 and first 20 years of follow-up (data not shown).
Table 4 Multivariate relative risks (RR) of cerebrovascular diseases, ischaemic stroke and intracerebral haemorrhage with 95 % confidence intervals by quartiles of fruit and berry intake
* Multivariate model: adjusted for age, sex, BMI, smoking, physical activity, serum cholesterol level, blood pressure and energy intake.
† Quartiles for total fruit and berry intake (including juices, jams and marmalades) (g/d): men, 0–47, 48–101, 102–174, 175–1094; women, 0–81, 82–151, 152–238, 239–1325.
‡ Quartiles for fruit intake (including fresh, dried and canned fruits, and juices) (g/d): men, 0–12, 13–52, 53–118, 119–1007; women, 0–36, 37–94, 95–168, 169–1082.
§ Tertiles for citrus fruit intake (including juices) (g/d): men, 0, 1–36, 37–740; women, 0–8, 11–67, 69–1040.
∥ Quartiles for other fruit intake (excluding citrus fruits) (g/d): men, 0–3, 4–21, 22–57, 58–522; women, 0–10, 11–42, 43–94, 95–876.
¶ Quartiles for (fresh) berry intake (g/d): men, 0–2, 3–9, 10–18, 19–308; women, 0–5, 6–12, 13–23, 24–246.
A study of the interactions between fruit and berry consumption and sex revealed a statistically significant interaction for cerebrovascular diseases (P = 0·02) and ischaemic stroke incidence (P = 0·02) but not for intracerebral haemorrhage (P = 0·78). The relative risks were 0·58 (95 % CI 0·42, 0·80) for all cerebrovascular diseases combined and 0·55 (95 % CI 0·36, 0·85) for ischaemic stroke in men and 1·14 (95 % CI 0·83, 1·57) and 1·30 (95 % CI 0·84, 1·99) in women. No significant interaction for hypertension was found (P>0·47).
Vegetables
The total consumption of vegetables, with or without potatoes, was not associated with the incidence of cerebrovascular diseases, ischaemic stroke or intracerebral haemorrhage (Table 5). Of the vegetable subgroups, the consumption of cruciferous vegetables was, however, inversely associated with the risk of all three types of cerebrovascular diseases. The relative risks for cerebrovascular diseases, ischaemic stroke and intracerebral haemorrhage between the highest and lowest quartiles of these vegetables were 0·79 (95 % CI 0·63, 0·99; P = 0·04), 0·67 (95 % CI 0·49, 0·92; P = 0·009) and 0·49 (95 % CI 0·25, 0·98; P = 0·04), respectively. In the case of ischaemic stroke, also the consumption of legumes and the consumption of root vegetables were associated with a reduced risk of the disease. The relative risk between the extreme quartiles was 0·72 (95 % CI 0·54, 0·96; P = 0·02) for legumes and 0·72 (95 % CI 0·53, 0·98, P = 0·03) for root vegetables. No significant associations were found for the intake of potatoes. The associations that were found remained after exclusion of the cases occurring during the first 4 years of follow-up and after inclusion of cases occurring during the first 15 and first 20 years of follow-up (data not shown). No significant interaction between the consumption of all vegetables combined and the potential effect-modifying factors sex and hypertension was detected (data not shown).
Table 5 Multivariate relative risks (RR) of cerebrovascular diseases, ischaemic stroke and intracerebral haemorrhage with 95 % confidence intervals by quartiles of vegetable intake
* Multivariate model: adjusted for age, sex, BMI, smoking, physical activity, serum cholesterol level, blood pressure and energy intake.
† Quartiles for total vegetable intake including potatoes (g/d): men, 9–252, 253–337, 338–448, 449–1354; women, 10–203, 204–273, 274–353, 354–1026.
‡ Quartiles for total vegetable intake excluding potatoes (g/d): men, 0–44, 45–84, 85–137, 138–535; women, 1–56, 57–95, 96–150, 151–800.
§ Quartiles for potato intake (g/d): men, 2–169, 170–239, 240–326, 327–1072; women, 0–108, 109–156, 157–223, 224–896.
∥ Quartiles for cruciferous vegetable intake (g/d): men, 0–1, 2–6, 7–13, 14–269; women, 0–2, 3–7, 8–15, 16–188.
¶ Quartiles for legume intake (g/d): men, 0–2, 3–5, 6–9, 10–101; women, 0–1, 2–3, 4–6, 7–43.
** Quartiles for root vegetable intake (g/d): men, 0–5, 6–17, 18–40, 41–356; women, 0–11, 12–29, 30–56, 57–579.
Cereals
Neither the consumption of whole grain nor refined grain was associated with the incidence of all cerebrovascular diseases combined, ischaemic stroke or intracerebral haemorrhage (Table 6). Exclusion of the cases that occurred during the first 4 years of follow-up and after inclusion of cases occurring during the first 15 and first 20 years of follow-up (data not shown) did not alter the results (data not shown). No significant sex or hypertension interaction was found (data not shown). To study whether the associations observed for fruits and berries, vegetables and cereals were independent of each other, we entered all three food groups simultaneously in the same model. All the results remained unchanged. For example, the relative risks of cerebrovascular diseases between the highest and lowest quartiles of the consumption of fruits and berries, vegetables and cereals after inclusion of the variables in the same model were 0·83 (95 % CI 0·65, 1·04), 0·95 (95 % CI 0·74, 1·21) and 1·10 (95 % CI 0·82, 1·47), respectively.
Table 6 Multivariate relative risks (RR) of cerebrovascular diseases, ischaemic stroke and intracerebral haemorrhage with 95 % confidence intervals by quartiles of total cereal, wholegrain and refined grain intake
* Multivariate model: adjusted for age, sex, BMI, smoking, physical activity, serum cholesterol level, blood pressure and energy.
† Quartiles for cereal intake (g/d): men, 10–223, 224–295, 296–390, 391–1535; women, 20–156, 157–210, 211–285, 286–1092.
‡ Quartiles for wholegrain intake (g/d): men, 0–139, 140–201, 202–279, 280–1321; women, 0–89, 90–134, 135–194, 195–963.
§ Quartiles for refined grain intake (g/d): men, 0–50, 51–82, 83–124, 125–567; women, 0–43, 44–68, 69–99, 100–457.
Discussion
Fruits and berries
The present cohort study showed an inverse association between fruit consumption and risk of cerebrovascular diseases, ischaemic stroke and intracerebral haemorrhage in a population of Finnish men and women combined with a relatively low fruit intake (about 70 % of the median value of other countries)(29). The findings concerning cerebrovascular diseases and ischaemic stroke are in accordance with the majority of previous studies(Reference Key, Thorogood and Appleby5, Reference Johnsen, Overvad and Stripp10, Reference Joshipura, Ascherio and Manson11), but not all(Reference Yokoyama, Date and Kokubo9). At variance with the present study, however, the only study on haemorrhagic strokes in men and women combined did not find any association between fruit consumption and risk of haemorrhagic stroke(Reference Yokoyama, Date and Kokubo9). In line with previous studies on ischaemic strokes(Reference Johnsen, Overvad and Stripp10, Reference Joshipura, Ascherio and Manson11), the strongest association was found with citrus fruits, which are known to be a major source of flavonoids(Reference Manach, Scalbert and Morand30).
The present study supports the hypothesis whereby fruit consumption has a protective effect on stroke incidence(Reference Khaw and Barrett-Connor1). Fruits have been found to reduce blood pressure(Reference Appel, Moore and Obarzanek31, Reference John, Ziebland and Yudkin32), the major risk factor in stroke(33). Nutrients that potentially contribute to this reduction include K, vitamin C and dietary fibre(Reference He and MacGregor34–Reference Kurl, Tuomainen and Laukkanen36). In addition to its antihypertensive effect, it has been suggested that in animal models K prevents stroke through other mechanisms, such as inhibiting free radical formation, proliferation of arterial smooth muscles and obstruction of blood vessels(Reference Tobian, Lange and Johnson37). Fruits may also prevent atherosclerosis through vitamins C and E, and other biologically active compounds such as flavonoids. Beside other possible effects, they all work as antioxidants inhibiting blood-clot formation(Reference Van Duyn and Pivonka38) by protecting fatty acids in cell membranes from oxidation. Fruits might also lower serum cholesterol or the homocysteine level. Potential nutrients contributing to these effects are dietary fibre for serum cholesterol(Reference Brown, Rosner and Willett39) and folic acid for serum homocysteine, which is a possible risk factor for stroke(Reference Coull, Malinow and Beamer40).
Unlike in the majority of previous studies(Reference Ascherio, Rimm and Hernan4, Reference Sauvaget, Nagano and Allen6, Reference Gillman, Cupples and Gagnon8, Reference Yokoyama, Date and Kokubo9, Reference Keli, Hertog and Feskens12, Reference Yochum, Kushi and Meyer13, Reference Mink, Scrafford and Barraj15), we found an association between fruit intake and cerebrovascular diseases in men but not in women. In the case of ischaemic stroke, we found an inverse association between fruit intake and disease risk in men, whereas previous studies found a reduced risk either in women(Reference Joshipura, Ascherio and Manson11), both in men and in women(Reference Sauvaget, Nagano and Allen6) or in neither(Reference Hirvonen, Virtamo and Korhonen7, Reference Yokoyama, Date and Kokubo9). Studies on fruit intake and haemorrhagic stroke risk have yielded similar results, which suggests an inverse association in men(Reference Hirvonen, Virtamo and Korhonen7) congruently to the present study, or both in men and in women(Reference Sauvaget, Nagano and Allen6) or no association at all(Reference Yokoyama, Date and Kokubo9). The inverse association observed in the present study was mainly due to an elevated risk in the lowest quartile of fruit intake in men. The intake in that category was much lower for men (0–12 g/d) than for women (0–36 g/d).
Vegetables
The total consumption of vegetables, with or without potatoes, was not associated with the risk of cerebrovascular diseases in the present study. Some of the previous studies, however, found an inverse association between total or some individual vegetable consumption and cerebrovascular disease incidence(Reference Sauvaget, Nagano and Allen6, Reference Gillman, Cupples and Gagnon8, Reference Yokoyama, Date and Kokubo9), whereas others reported no significant association(Reference Ascherio, Rimm and Hernan4, Reference Key, Thorogood and Appleby5, Reference Keli, Hertog and Feskens12, Reference Yochum, Kushi and Meyer13). The results were also inconsistent for ischaemic and haemorrhagic stroke. Only a few studies found an inverse association for ischaemic(Reference Sauvaget, Nagano and Allen6, Reference Hirvonen, Virtamo and Korhonen7) or haemorrhagic stroke(Reference Hirvonen, Virtamo and Korhonen7), whereas the majority found no significant association between vegetable consumption and either type of stroke(Reference Sauvaget, Nagano and Allen6, Reference Yokoyama, Date and Kokubo9, Reference Johnsen, Overvad and Stripp10, Reference Joshipura, Ascherio and Manson11).
Study of vegetable subgroups showed that a higher cruciferous vegetable intake was associated with a lower risk of both ischaemic stroke and intracerebral haemorrhage. Previous studies on the relationship between cruciferous vegetables (or a single cruciferous vegetable, such as broccoli) and total stroke(Reference Yochum, Kushi and Meyer13) or ischaemic stroke(Reference Johnsen, Overvad and Stripp10, Reference Joshipura, Ascherio and Manson11) have given inconsistent results. In the case of ischaemic stroke, Joshipura et al. (Reference Joshipura, Ascherio and Manson11) found a reduced stroke risk for a higher intake of cruciferous vegetables, whereas Johnsen et al. (Reference Johnsen, Overvad and Stripp10) did not.
The present study gives limited support to the general hypothesis that the consumption of vegetables gives protection against stroke. One reason for the lack of association in the present study might be the comparatively low consumption of vegetables in Finland in general. At the time of the baseline examination of the present study, the intake of vegetables was only about 20 % that of the median intake of vegetables in previous study populations in other countries(29). Furthermore, fresh vegetables, which contain more beneficial nutrients, became more available only after the baseline examination(Reference Pietinen, Lahti-Koski and Vartiainen41). Nevertheless, our findings suggest that the consumption of cruciferous vegetables may be of benefit against cerebrovascular diseases. Cruciferous vegetables are an important source of flavonoids, as are citrus fruits(Reference Manach, Scalbert and Morand30), the consumption of which was also found to be inversely associated with the risk of cerebrovascular diseases. These results are in agreement with the findings from previous studies on flavonoids and risk of cerebrovascular diseases in this same study population(Reference Knekt, Isotupa and Rissanen42, Reference Knekt, Kumpulainen and Järvinen43).
Cereals
It has been suggested that the consumption of cereals, especially wholegrain, reduces the risk of chronic diseases(Reference Mellen, Walsh and Herrington44, Reference Trowell45). Possible mechanisms could be a reduction in serum cholesterol, blood pressure and weight, mediated through cereal fibre, PUFA and different vitamins in grains(Reference He and Whelton35, Reference Flight and Clifton46). However, we did not find any association between the consumption of cereals, whether wholegrain or refined, and the incidence of cerebrovascular diseases. In previous studies the results have been inconsistent, suggesting that some factors may modify the possible beneficial effects of cereals. A reduced risk of stroke was only found for women when bran was added to their food(Reference Mink, Scrafford and Barraj15). In the case of ischaemic stroke, one study(Reference Liu, Manson and Stampfer14) found an inverse relationship for wholegrain consumption in women. Other studies(Reference Ascherio, Rimm and Hernan4, Reference Key, Thorogood and Appleby5, Reference Steffen, Jacobs and Stevens16) found no significant association between the consumption of cereals, wholegrain bread or bran cereals and cerebrovascular disease or ischaemic stroke.
Methodological issues
There were several advantages in the present study: the prospective study design with a long follow-up giving enough outcomes of a disease with a long latency period; a comprehensive follow-up through nationwide, population-based registries; the possibility for separate consideration of ischaemic strokes and intracerebral haemorrhages; a study population consisting of both men and women; the use of the dietary history interview method for measuring food intake. An additional advantage of the present study was the possibility to explore the existence of potential effect-modifying factors, enabled by a study population of sufficient size and a sufficiently long follow-up period.
There remain, however, some methodological issues that may have affected the results. First, it cannot be excluded that the associations observed are due to the presence of some other healthy dietary factors or the absence of some unhealthy factors related to the consumption of plant foods. Although plant foods are generally considered to be healthy, the beneficial health effect may become masked by unhealthier accompaniments (for example, adding full-fat dairy products to wholegrain products or preserving vegetables in salty liquids and fruit and berries in excess sugar). The associations may also be related to lifestyle, which might be healthier among high consumers of fruits and vegetables, or due to residual confounding, for example, alcohol consumption, which was left out of the analysis because of insufficient data. Second, during the 24-year follow-up, considerable changes in the diet of the Finnish population occurred(Reference Pietinen, Lahti-Koski and Vartiainen41). The consumption of fresh fruits doubled and the consumption of vegetables tripled as a result of improved availability, better storage facilities, and, presumably, an improved level of health education. These changes may have caused conservative estimates of the strength of associations between plant food intake and the incidence of cerebrovascular diseases. Finally, methodological issues related to the dietary history interview may have caused conservative estimates of the strength of associations. The short- and long-term reliability of the variables studied were, however, reasonably high and thus allowed the relevant associations to be revealed(Reference Järvinen, Seppänen and Knekt23).
Conclusions
In conclusion, the results of the present study suggest that fruit consumption, especially citrus fruit, may provide protection against all cerebrovascular diseases, as well as against ischaemic stroke and intracerebral haemorrhage in men but not in women. Furthermore they suggest that the consumption of cruciferous vegetables, but not general consumption of vegetables, may reduce the risk of the same disease. Consumption of cereals did not, however, predict a reduced risk of cerebrovascular diseases. These results thus support the hypothesis that the dietary intake of plant foods reduces the risk of cerebrovascular diseases. The present results, however, indicate that the reduction may be due to selected plant foods. Because of the inconsistency of the findings between men and women and type of plant food in the present study and in studies carried out in different settings, further studies on plant foods and the risk of cerebrovascular diseases and the mechanisms behind their association are needed.
Acknowledgements
The present study received funds from the Social Insurance Institution, Finland.
A. M. participated in the interpretation of the data and in the drafting of the manuscript, and conducted the statistical analyses. P. K. participated in the analysis and interpretation of the data, in the acquisition, and in the drafting of the manuscript, and supervised the conducting of the study. J. M. participated in the analysis and the interpretation of the data and in the critical revision of the manuscript. M. A. L. provided statistical expertise and participated in the analysis and the interpretation of the data, and in the writing of the manuscript. M. H. participated in the study concept and design, the acquisition of data, and the critical revision of the manuscript. R. J. participated in the study concept and design, the acquisition and interpretation of data, and the critical revision of the manuscript.
None of the authors had any personal or financial conflict of interest.
