We report here results from the first study that has evaluated the dietary intake of fish, PUFA and vitamin D in relation to the experience of positive psychotic-like symptoms in a large cohort of over 30 000 Swedish women. We found support for a protective effect regarding the risk of positive psychotic-like symptoms with high dietary intake of fish, omega-3 and omega-6 PUFA, as well as of vitamin D. The associations were J-shaped with the strongest reduced risk for an intermediate intake of fish or PUFA.
Our findings provide further support for the hypothesis that an aberration in lipid metabolism may be involved in the biochemical basis for psychiatric disorders [13
]. This suggestion has earlier been supported by ecological studies showing that variations in schizophrenia outcome between countries may be due to differences in the diet [6
]. Further evidence for the hypothesis that PUFA is involved in the etiology of psychiatric disease has been put forward through the study by Stokes et al. showing a negative correlation between dietary PUFA intake and the severity of psychotic symptoms [15
], as well as studies showing that patients with schizophrenia or depression have lower levels of PUFAs in brain tissue, red blood cells and skin fibroblasts and with a low intake of fish and PUFAs [16
]. Finally, supplementation of PUFA, especially EPA, has been reported to be of possible benefit for patients with schizophrenia [17
No earlier studies that we are aware of have evaluated the relationship between adult vitamin D levels and psychotic symptoms. Our findings indicated a protective effect of vitamin D for the risk of psychotic-like symptoms. Prenatal vitamin D deficiency has been proposed to be a risk factor for the development of schizophrenia [12
]. Results from the Northern Finland 1966 Birth Cohort showed that vitamin D supplementation during the first year of life was associated with a reduced risk of schizophrenia in males, but not in women [32
]. However, a small pilot study of maternal vitamin D levels in archived prenatal sera, showed no decrease in prenatal vitamin D in subjects who later developed schizophrenia [33
]. It has been hypothesised that vitamin D insufficiency could account for some of the increased risk of schizophrenia observed among dark-skinned immigrants moving to countries with less sun exposure [12
]. Vitamin D is to a large extent metabolized in the body through sun exposure, and people with dark skin need more sun exposure to maintain adequate blood levels. Our results of a protective effect of vitamin D intake must be considered as rough, since we only measure the dietary intake of vitamin D. A more complete picture of the vitamin D status could have been supplied through vitamin D levels in blood; however, no biological samples are available. The absorption of dietary vitamin D is generally high at all stages of life [34
], but serum levels also depend on the endogenous production of vitamin D due to sun exposure which is subject to seasonal variations. Thus, the correlation between vitamin D intake and serum levels may vary. However, Burgaz et al. recently reported that 2-3 weekly servings of fatty fish increased 25(OH)D by 45% in a population of Swedish women [35
]. Our results of the protective effect of fatty fish could in part be due to the content of vitamin D. While the epidemiological evidence linking low prenatal vitamin D and schizophrenia remains inconclusive, rodent models have provided compelling evidence about the role of vitamin D deficiency for brain development like larger lateral volumes [36
], subtle memory dysfunction and altered attention processing [37
] which have implications for neuropsychiatric disorders.
Unexpectedly, the intake of fatty fish (salmon, herring and mackerel) or shellfish more than twice a week increased the risk of being in the group with the highest level of psychotic-like symptoms. This puzzling finding may be due to unknown or known unhealthy constituents of fatty fish. For instance, environmental pollutants such as polychlorinated biphenyls (PCB) and dioxins are known to accumulate in fatty fish [38
]. Another possible explanation may be that the frequent intake of fish and PUFA may be advantageous in lower doses but disadvantageous in higher doses. Reports by Mischoulon et al. and Peet & Horrobin suggest that there might be such a therapeutic window for DHA and EPA regarding their protective role for schizophrenia or depression [39
]. Significant higher levels of DHA have been found in red cell membranes of un-medicated schizophrenic patients compared to healthy control subjects [42
]. Furthermore, our findings of a more pronounced protective effect of omega-6 fatty acids than for omega-3 fatty acid are in agreement with the results from an EPA supplementation study in schizophrenia presented by Horrobins et al. [43
]. This study surprisingly found that the effect of intermediate doses of EPA increased the levels of AA (an omega-6 PUFA) in the membrane of red cells, whereas higher doses of EPA did not, and the EPA-induced rise in AA was associated with a clinical improvement. The respective biological effects of omega-3 fatty acids and omega-6 fatty acids on the etiology of psychiatric symptoms could very well differ and the balance between the intakes of these fatty acids might be of importance. A high intake ratio of omega-3:omega-6 fatty acids favor omega-3 fatty acid metabolism. For example, high intake of omega-3 fatty acids partly replaces omega-6 fatty acids incorporation into membrane phospholipids and omega-3 fatty acids have a higher affinity than omega-6 fatty acids for several enzymes [44
]. It has been proposed that the ratio of omega-3:omega-6 fatty acids might be more important in inhibiting the development of several diseases, including cancer, inflammatory and heart diseases [45
]. We find no support for this in our results, since, the effect of omega-3:omega-6 fatty acids on psychotic-like symptoms were almost similar to those of omega-3 fatty acids. However, the mechanisms of action and protective abilities of PUFA could differ between different diseases.
The non-linearity in the association might seemingly argue against a possible causal relationship. However, the associations with dietary components and health are often non-linear with advantageous effects of a balanced nutrition [47
]. We have no baseline measure of symptom levels to further elucidate causality between different levels of intake of fatty fish or shellfish and the risk of positive psychotic-like symptoms. However, keeping the prevalence figures for psychotic disorders in mind, rather few participants in our population sample are likely to have a disorder, which may diminish the problem of reversed causality related to psychotic diagnosis or medication.
In our study, the definition of psychotic-like symptoms was based on self-reported frequency of psychosis-like experiences. The classification of women into three groups with different levels of symptoms (low, middle and high) was based on predefined, but arbitrary cut-offs from the self-reported answers to the CAPE questionnaire. We acknowledge the limitation that the scale have not been used earlier in Sweden and validated in the Swedish population. Among women in the group with the highest level of psychotic-like symptoms, overweight, obesity and smoking were more common, and this group also contained more women who had migrated to Sweden. These characteristics are often seen in patients with psychotic disorders [51
]. Based on the sum population prevalence of schizophrenia and other psychoses in middle-aged women [54
], we might expect that 2-3% of the study participants would cross the boundaries to clinically valid syndromes. Our high level symptoms group consisted of 840 persons or about 2.5% of the whole study group. The CAPE measures of psychosis are strongly correlated with measures of general psychopathology, including depression. The association between the positive and the depressive dimension in CAPE, which we unfortunately could not include for practical reasons, is fairly low when distress associated with positive symptoms is held constant (r = 0.25; Stefanis et al., 2002 [1
]). Thus, there are reasons to believe that the dimension of positive symptoms is an independent dimension.
In terms of generalisability of our results it is relevant to compare the levels of dietary intake of PUFAs and vitamin D in our cohort to other populations. The contribution of dietary intakes of PUFAs to total dietary fat is comparable to other Western cohorts [55
]. The quotient omega-6 PUFA/omega-3 PUFA was 4:1 in this study, and this value is comparable to other European cohorts [56
], but lower than that has been reported for US populations (Food and Nutrition Board. Dietary Reference Intakes for macronutrients Institute of Medicine, National Academic Press, Washington: 2005). The dietary intake of vitamin D in this cohort is comparable to several other European populations [58
]. It is relevant to note that in Sweden low-fat dairy products and margarines are fortified with vitamin D. Otherwise the dietary intake of vitamin D would be lower. Furthermore, men were not included in the study and there is gender difference in the prevalence of psychosis [59
]. However psychotic-like experiences in the general population might be more equally distributed among women and men [60
The strengths of our study include its thorough diet data design and large sample size. The ethnic homogeneity of our study population reduces the risk of confounding by unmeasured factors, both genetic and environmental. We were able to adjust for smoking, BMI, migration, education and alcohol that could confound our associations between diet and psychotic-like symptoms. We had no information about socioeconomic status, but the adjustment for education, which is strongly associated with socioeconomic status, did not change the estimates. Still, we cannot rule out that there are unknown confounders that we have not been adjusted for, for example drugs influencing levels of serum lipids or family history of psychiatric illness. If the proportion of non-urbanized participants was high in the study population this could have confounded our results, nevertheless this is not the case in the Swedish population. Misclassification of fish and PUFA or vitamin D intake due to measurement error associated with the food frequency questionnaire is unavoidable, but given the study design likely non-differential, and thus attenuating any true association. It is also relevant to note that Hibbeln et al. reported high correlations (r = 0.7) between the dietary intake of EPA and DHA and serum levels of EPA and DHA in subjects with schizophrenia [51
], indicating a good ability to report dietary intake accurately. An important limitation of this study is that we measured dietary intake only once, involving misclassification among those who changed their dietary pattern during follow-up. However, again some misclassification of dietary intake cannot be ruled out, such a misclassification is likely to be non-differential, and attenuating any true association. Unfortunately, we do not have information about specific vitamin D supplementation. But, adjustment for multivitamin supplement intake, containing vitamin D, did not change the estimates. None of the women in this study reported the use of dietary supplements containing fish oil or PUFA at baseline. Unfortunately, we do not have any information about the use of such supplements during follow-up. However, according to national figures from the National Food Administration, a low number of Swedish women took fish oil supplements (1%) at the time of the study.
We compared a number of characteristics for participants who completed the questionnaire, both in the parent study and the follow-up study, with those who only answered the questionnaire in the parent study. Age and overall dietary intake as well as the intake of fish and fatty acids did not differ significantly between those who participated in the follow-up study and those who did not (drop-outs). The drop-outs had a slightly higher BMI, lower education and were more often smokers and born outside of the Nordic countries. However, our main exposure (fish/fatty acids) did not differ between those who participated and those who did not participate in the study.