Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

No Association between Fish Intake and Depression in over 15,000 Older Adults from Seven Low and Middle Income Countries–The 10/66 Study



Evidence on the association between fish consumption and depression is inconsistent and virtually non-existent from low- and middle-income countries. Using a standard protocol, we aim to assess the association of fish consumption and late-life depression in seven low- and middle-income countries.


We used cross-sectional data from the 10/66 cohort study and applied two diagnostic criteria for late-life depression to assess the association between categories of weekly fish consumption and depression according to ICD-10 and the EURO-D depression symptoms scale scores, adjusting for relevant confounders. All-catchment area surveys were carried out in Cuba, Dominican Republic, Venezuela, Peru, Mexico, China, and India, and over 15,000 community-dwelling older adults (65+) were sampled. Using Poisson models the adjusted association between categories of fish consumption and ICD-10 depression was positive in India (p for trend = 0.001), inverse in Peru (p = 0.025), and not significant in all other countries. We found a linear inverse association between fish consumption categories and EURO-D scores only in Cuba (p for trend  = 0.039) and China (p<0.001); associations were not significant in all other countries. Between-country heterogeneity was marked for both ICD-10 (I2>61%) and EURO-D criteria (I2>66%).


The associations of fish consumption with depression in large samples of older adults varied markedly across countries and by depression diagnosis and were explained by socio-demographic and lifestyle variables. Experimental studies in these settings are needed to confirm our findings.


Depression is projected to become the leading cause of the global burden of disease by 2030, [1] with the steepest increases in prevalence in ageing societies [2] and in countries with low and middle incomes. [3], [4] Increased prioritisation of mental health interventions in public health policies is urgently required particularly in these settings [5], [6] where the mental–physical treatment gap is greatest. [7] The prevention and treatment of late-life depression represent a major current research and policy effort, [8] and some protective factors may also have treatment promises.

Lifestyle risk factors including nutrition, vascular pathology and inflammation [9] may influence the heritability of late life depression. [10] The n-3 Long Chain Polyunsaturated Fatty Acids (n-3 LC PUFAs), namely eicosapentaenoic acid (EPA) and docosahexanoic acid (DHA) most commonly found in oily fish play important physiologic roles in humans including their actions on cell signaling and transduction, receptors density regulation and metabolite production.[11][13] Plasma n-3 LC PUFA concentrations appear to correlate with serotonin and dopamine status in the central nervous system [14] which may also be important in the pathophysiology of depression. [15], [16].

Ecological studies suggest an inverse association of fish consumption with prevalent depression, [14], [17] post-partum depression [18] and bipolar disorders. [19] Population-based epidemiological studies employing food frequency questionnaires to ascertain diet similarly report inverse associations between fish consumption and prevalent depression,[20][22] depressed mood [23] or mental health, [24] although these findings are not consistent, [25], [26] particularly when diet history questionnaires are used. [27], [28] Depression diagnosis in these studies was self-reported using a variety of questionnaires, which may further contribute to the inconsistency in findings. The results of intervention studies which have largely focused on the potential treatment effect of EPA and/or DHA supplementation on depression, have been inconclusive and between study heterogeneity is marked. [29].

Numerous social and demographic factors including gender, marital status, functional impairment and illness influence risk for depression in older people.[30][34] The relationships between risk factors are complex and confounding may play an important role in modifying any association of dietary fish consumption with depression. [35] There are currently no studies assessing the association between fish consumption and depression in low and middle income countries and the degree of consistency of the evidence across cultures is virtually unknown.

In the absence of any evidence on the importance of fish consumption as a protective factor against depression in low and middle income countries, we set out to test the strength of the relationship of fish consumption with prevalent depression in large representative samples of community-dwelling older people, aged 65 years and over, in China, India, Cuba, Dominican Republic, Mexico, Peru and Venezuela. We further assess the robustness and consistency of any such associations taking a wide range of potential confounders into account and comparing results obtained using two different diagnostic criteria.

Materials and Methods

Ethics Statement

The 10/66 research programme was approved by the Research Ethics Committee at King’s College London and by local ethical committees in each country.

Study Design

This study is part of the 10/66 research programme on ageing, mental health and non-communicable diseases in countries with low and middle incomes. The study designs and procedures have been extensively described, [36] and validated. [37] Further details are available on the study website Here we describe in brief the methods and measures directly pertinent to the current report.

Between January 2003 and November 2007, we conducted all catchment area one-phase surveys amongst community dwelling older people (65+ years) in Cuba, Dominican Republic, Peru, Venezuela, Mexico, China and India. We recruited all residents aged 65+ years with no exclusion criteria. Power calculations suggested a target sample size of 2,000 participants per country. Written consents were obtained from all participants or from a close relative or caregiver in case of incapacity or illiteracy.

The defined 10/66 study protocol includes data on household, participants and informants’ socio-demographic characteristics and lifestyle risk factors, the Geriatric Mental State (GMS) examination, [38] physical and neurological examinations (NEUROEX), [39] a comprehensive cognitive battery [40] and a structured informant interview. The study protocol, including the dietary assessment, has been translated, with cultural adaptations, into Spanish, Chinese and Tamil by local physicians fluent in English. A comprehensive study manual (along with videos) covers all procedural and content aspects of the research programme. Local principal investigators received standardized one-week trainings and scrutinized the work of the local teams with the continuous assistance of the London coordinating centre.

Depression Diagnosis

Depression diagnosis criteria applied in the 10/66 population-based studies have been validated [41] and extensive details on GMS-based criteria and algorithms have been described elsewhere. [42] In this study, with respect to the month preceding the interview, we use the ICD-10 depressive episode criteria, [43] derived using a validated clinical-based computerized algorithm [44] applied to the GMS and disregarding severity. Additionally we use a score derived from the EURO-D [45] scale for late-life depression symptoms, based on 12 GMS domains (depressed mood, pessimism, suicidality, guilt, sleep, interest, irritability, appetite, fatigue, concentration, enjoyment and tearfulness), which ranges from 0 (no symptoms) to 12. Limited to the descriptive analyses we also consider a cut-off point ≥4 to identify cases as in the SHARE studies in Europe. [46] ICD-10 criterion strictly identifies clinically significant cases, while EURO-D diagnosis proved considerably more sensitive in identifying subsyndromal cases whose depressive symptomatology is significantly associated to severe disability. [42] Past history of depression was determined on self-reported clinical diagnosis.

Dietary Assessments

We asked standardized questions (“how often do you eat fish/meat in a week?”) in face-to-face interviews, and recorded the average weekly consumption (“never”, “some days”, “most days”, “every day”) along with the average number of daily portions of vegetables, fruits and units of alcohol consumed per week. Interviewers gathered confirmations from informants (generally a close relative) when dietary habits appeared implausible (n = 58), or for participants with moderate or severe dementia according to Clinical Dementia Rating scale (CDR) [47] (n = 367). We have previously reported the concurrent validity of our dietary assessment across the 10/66 study sites and tested internal consistency amongst dietary measures (i.e. fish, meat, fruits and vegetables and alcohol) and assessed Kendall’s τ correlations with socio-demographic characteristics. Reported fish consumption followed expected patterns of associations with higher educational levels and better socio-economic circumstances. The dietary intake assessment also identified inverse associations of reported fish consumption with prevalent dementia. [48].

Other Relevant Measures

We recorded participants’ age (confirmed by documentations and the informants or with respect to historical events) and gender; educational level (in five grades from illiteracy to completed tertiary school); marital status (never married, married/cohabitant, widowed and divorced/separated); number of household assets (car, television, refrigerator, telephone, plumbed toilet, water and electricity utilities) and physical activity level (not at all, fairly active, active and very physically active). We registered self-reported physical impairments and clinically diagnosed illnesses (including stroke, coronary heart disease and diabetes mellitus) using a standard questionnaire. [49] Dementia diagnosis was established applying the cultural and education-fair 10/66 validated algorithm [37], [50] and an overall cognitive score (COGSCORE) was calculated based on our neuropsychological battery. [40].

Statistical Analysis

Participants’ characteristics.

We describe the socio-demographic and health characteristics of participants by country and by ICD-10 depression. On inspection we combined participants who reported to eat fish “most days” and “every day”. We then describe weekly fish intakes (“none”, “some days”, “most days”) across countries and by depression status (yes/no) according to ICD-10 criteria and EURO-D caseness. The association of dietary fish intake with participants’ socio-demographic and physical health and dementia characteristics have been reported elsewhere. [48].

Association between dietary fish and depression.

We used Poisson regressions to calculate prevalence ratios (PRs) with robust 95% confidence intervals, adjusting for clustering of characteristics within households, to determine the risk of prevalent ICD-10 depression associated with level of fish consumption. With the diagnosis of ICD-10 depression as dichotomous outcome we calculated PRs by entering weekly fish intakes in the model as a categorical variable with three levels (‘never eat fish’, ‘eat fish some days’ and ‘eat fish most days’). The middle category of ‘eat fish some days’ was the reference group (PR = 1) in order to aid linearity checks across estimates (i.e. a dose-response-like pattern configures when PR<1 and PR>1 amongst least and most days fish consumers respectively). [51] We then entered fish consumption in the above models as a continuous variable and interpreted results as test for trends and tested departures from linearity applying likelihood ratio tests. We generated an unadjusted and two adjusted models (see below) by country, selecting confounders on the basis of our a priori hypothesis and consistently with previous population-based studies to allow comparisons. In adjusted model 1 we controlled for age (continuous variable), gender (females vs. males), educational level (none, some, primary, secondary and tertiary), number of household assets (continuous variable), marital status (never married, married/cohabiting, widowed and divorced/separated), overall cognitive score (continuous variable), self-reported clinically diagnosed diabetes, coronary heart disease and stroke (yes/no) and total number of physical illnesses (continuous variable). To disaggregate the possible effect of socio-demographic from lifestyle factors, adjusted model 2 further allowed for physical activity level (very, fairly, not very, not at all physically active), weekly meat intake (never, some days, most days, every day), fruits and vegetables weekly portions consumed (continuous variable) and units of alcohol drunk (continuous variable). The country-specific differential contribution of confounders was explored performing backward stepwise estimations. We combined the country-specific pairs of PRs calculated in the crude model and in model 1 and 2 into a series of fixed-effect method meta-analyses to obtain the pooled estimates of the associations between fish intake levels and ICD-10 depression status, after having formally tested heterogeneity with Cochrane Q statistics (on appropriate degrees of freedom) and calculated I2 Higgins to determine the percentage of between country differences not due to chance. [52].

On inspection EURO-D scores were skewed and over-dispersed with an apparent zero-inflation. To investigate this zero-inflation, we modeled the effect of country on EURO-D scores using zero-inflated negative binomial (ZINB) regression. With this method the assumption is made that a subgroup of participants exists who always has zero counts regardless (the so called ‘nay-saying’). In zero-inflated models, this group is referred to as the ‘certain zeros’. We calculated the likelihood of being a ‘certain zero’ by country with a logit specification and used negative binomial models in the non-certain zero group to determine the effect on the EURO-D scores of country alone (dummy variable) and then adjusting for compositional variables: age, gender, education, household assets, self-reported diabetes, coronary-heart disease and stroke, number of physical illnesses, marital status, overall cognitive score and physical activity level. In doing so, we sought to account for the different psychometric properties of the EURO-D scale across the study sites.

Next we measured the associations between fish intake and depressive symptomatology entering dietary fish categories (independent variable) and EURO-D scores (dependent/outcome variable) in ZINB models, similarly to other population-based studies that used comparable depression scale and likewise had to deal with excessive zero values. [53] We used Vuong tests to formally test the goodness of fit of the ZINB over standard negative binomial models. [54] We obtained relative risks (RR) by country first in unadjusted models and then controlling for confounders as in model 1 and 2 (see above) and we estimated degree of heterogeneity among site-specific estimates using Higgins I2 with 95% C.I. as appropriate.


The achieved sample was 15,022, response rates were over 80% in all countries. Data on depression diagnosis and fish consumption were available for 14,926 participants (99.4% of the total). Some socio-demographic differences between countries especially in age, education, co-habitation and asset ownership were identified, and differences between those with and without depression were marked (Table 1). ICD-10 depression prevalence ranged from 0.5% in China to 13.8% in Dominican Republic, and EURO-D depression prevalence was consistently higher ranging from 2.8% in China to 41.5% in India. Self-reported clinically diagnosed stroke prevalence varied from 1.6% in India to 8.7% in the Dominican Republic, while self-reported non-insulin dependent diabetes mellitus was particularly common in Mexico and Cuba (21.7% and 18.6% respectively). Overall in Venezuela and the Dominican Republic more people suffered from three or more physical illnesses than in the other countries. As previously reported 10/66 dementia rate varied from 6.3% to 11.7% between countries. [55].

Table 1. Participants Socio-Demographic and Health Characteristics.

In all countries, most participants reported fish consumption on some days of the week (as shown in table 2). In all countries except India those reporting the consumption of fish on most days of the week had lower proportions of ICD-10 depression and lower EURO-D scores compared to those reporting to never eat fish. Moreover, most of the variation in the distribution of EURO-D scores across countries was accounted for by China and to a lesser extent Cuba. The very high proportion of zero scores (zero-inflation) (81.1%) in China lowered the country mean score (Table 2).

Table 2. ICD-10 depression cases and EURO-D scores (and number of zero scores) by fish consumption categories.

The first ZINB model assessed the effect of country on EURO-D scores and showed that most of the variation arose from zero-inflation while the variation was markedly smaller in the count part of the model. After adjustment, for both models the effect of the compositional differences between countries in socio-demographic and health characteristics was modest (Table 3).

Table 3. Between-country variation in zero inflation1 and euro-d total score counts as modelled by zero-inflated negative binomial regression, before and after adjustment for compositional factors.

Association between Prevalence of Depression and Dietary Fish

In unadjusted analysis, compared to those who ate fish some days of the week, there was a general tendency for rates of prevalent depression to be higher among those who reported never eating fish and lower among those who reported eating fish most days of the week (Table 4). India and Mexico were exceptions to this general pattern and for ICD-10 criteria there were too few cases in China to estimate parameters. Similarly, EURO-D scores were highest amongst those who reported never eating fish and the general pattern was consistent between ICD-10 diagnosis and EURO-D scores in all countries except Peru. The observed tendency of an inverse association of fish consumption with ICD-10 depression prevalence and EURO-D scores was largely attenuated when adjusted for participants’ socio-demographic and health status (model 1) and lifestyle characteristics (model 2). Stepwise estimations of the covariates coefficients included in model 2 failed to identify a common pattern among countries.

Table 4. Crude and adjusted robust prevalence ratios (PRs) (with 95% confidence intervals, CI) from Poisson regressions for the association between ICD-10 depression and fish consumption by country and pooled estimates from fixed-effect models meta-analyses.

Fixed-effect meta-analytical combinations of country-specific PRs (and 95% CI) substantially confirmed the patterns observed at the individual country level. In the unadjusted analysis, compared to those who eat fish some days of the week there was a significant decreased risk of prevalent ICD-10 depression among those who eat fish on most days (pooled PR = 0.73; 95% CI: 0.59 to 0.91). Adjusting for socio-demographic, health and lifestyle characteristics markedly attenuated all associations in the pooled analyses (model 2). Across countries a trend emerged towards an increased risk of depressive symptoms (EURO-D criteria only) among those who eat fish on most days compare to those who eat fish some days (Table 5). However, while between-country heterogeneity was moderate and not significant for ICD-10 criteria, it was marked for EURO-D scores and consistently increased from less to more heterogeneity for the adjusted models, such that meta-analysis of the latter estimates were deemed inappropriate (Table 5).

Table 5. Relative risks (95% confidence intervals, CI) for the association between euro-D total score and fish consumption from zero-inflated negative binomial (ZINB) models and between country heterogeneity estimates.


We conducted catchment area surveys of representative samples of 15,022 older people in Cuba, Dominican Republic, Venezuela, Mexico, Peru, India and China. We achieved high response rates and applied identical standardized validated protocols in each study site. [50] We used diagnoses of depression cross-culturally validated in older people and face-to-face dietary assessments. [41] Our assessments were well tolerated and confirmed by proxy informants (generally a close relative) where necessary. We identified differences among the seven countries in socio-demographic and health characteristics, and exposure and outcome status. We applied a strict diagnostic criteria of depressive illness (ICD-10) and scores of a broader criterion (EURO-D) to capture depressive symptomatology of less severe or precursor “depressed mood” cases. Overall we found that the associations of high fish consumption with ICD-10 prevalent depression and of low fish consumption with EURO-D scores for depressive symptoms were almost entirely explained by socio-demographic, lifestyle and health characteristics.

While the country-specific estimates of the associations of fish consumption with prevalent depression were overall homogeneous across the Latin American countries, results from India were markedly different. From cross-sectional studies such as this it is not possible to define causality and the unexpected finding in India warrants further research. For example we have recently reported an increased risk of mortality among depressed individuals in our Indian study site, [56] which opens the possibility that survival bias may have modified our associations; namely shorter survival among depressed participants who eat less fish may have occurred such that the high fish consumption in prevalent cases appeared spuriously high. Alternatively, we cannot exclude that depression may lead to an increase of fish consumption in India, differently from all other countries. The paucity of cases of depression in China did not allow appropriate comparisons, and also warrants exploration when data from the incidence phase of the 10/66 project will be available.

The study has some limitations which mainly relate to the study design and dietary assessment. Results based on an all-catchment area sampling procedure should be generalised with caution and only to populations similar to the ones under study. Observational studies hint at but cannot prove causality and are prone to residual confounding. Depression status may influence diet and lifestyle and reverse causality cannot be excluded in cross-sectional. [57].

While there is biological plausibility to support the relationship between fish consumption and depression, [58] the reliability of reported fish consumption as a proxy of n-3 LC PUFAS intake [59] or status [60] has been questioned. Our dietary assessments also provided no information about type of fish consumed or cooking methods which may further modify n-3 LC PUFAS intake. Formal validation of dietary measures in large epidemiological studies is complex especially in multi-country studies such as ours. We validated our dietary assessment using concurrent measures gathered within our study and showed consistent and plausible dietary patterns and important associations with health outcomes. [48] A recent exercise to create detailed lists of common fish and fish names in each study site confirmed that the definition of fish given in our study protocol remains valid across countries.

Information bias is less likely in face-to-face interviews used in our study than in widely used food frequency questionnaires, [61] and systematic underreporting found for example in obese subjects, [62] is also unlikely in our study. Errors in measures of dietary consumption are likely to have occurred at random in our study leading to an underestimation of the true effect. In our study, fish consumption was fairly consistently associated with higher education and better socio-economic conditions, [48] and when we included these potential confounding factors in our models the association of fish consumption with depression was significantly attenuated in all study sites.

Our depression diagnoses are based on symptoms referred to the month prior the interview and we did not distinguish between long life depression and late life depression. However, a sensitivity analysis that excluded those who reported a past history of clinically diagnosed depression did not alter substantially our results (data not shown). Our diagnostic tools have been cross-culturally validated [41] and the internal validity of our study is strong, nevertheless the large between-country variation in zero inflation in the EURO-D scores (independent of compositional factors) represents an important finding, with a large positive association with zero inflation in China and inverse associations in all other countries interpreted as a cultural tendency to so-called nay-saying in China and yea-saying in the other settings compared to our reference country Cuba (where access to health care is universal and number of psychiatrists per inhabitant highest). Beyond this, between countries comparisons are appropriate and indeed the low prevalence of depression in China compared to figures from several European countries [63] deserves further investigation. For instance it may be that by favouring a more “etic” vs. a less “emic” approach, we might have involuntarily introduced a cultural bias. Finally, complete data for all the covariates were available for 90.5% of the whole sample. Those with and without missing information did not differ for depression prevalence (p = 0.258) and fish consumption patterns (p = 0.09) and we ran all analyses on the same smallest sample sizes by country allowing direct comparisons across models. However while generally robust, some of our estimates from Venezuela should be interpreted conservatively due to high missing data on alcohol consumption (n = 803). When we repeated the analyses including these participants (and excluding alcohol consumption from model 2) 95% C.I. were smaller and estimates similar, for example the PRs of ICD-10 depression for those who never eat fish compared to those who eat fish some days were 1.17 (95% CI: 0.52, 2.65), 0.97 (95%CI: 0.43, 2.19) and 0.52 (95%CI: 0.34, 2.78) for the unadjusted model and model 1 and 2 respectively.

To our knowledge this is the first population-based study to focus exclusively on late-life depression and fish consumption, and the largest study ever reported in which depressive status was ascertained by applying two validated diagnostic criteria. Importantly, this is also the first study that compared data collected using a standardised protocol from a set of culturally and geographically diverse low and middle income countries. While our results differ from some previous reports[20][24], [64], [65] they replicate the findings from two similar large studies of participants aged 50+ years in Finland. [27], [66] Moreover, our results are broadly consistent with studies that defined more than two categories of fish consumption [25], [26], [28], [67] and that adjusted for a similar range of potential confounding variables. [25], [35].

Inconsistency of results between observational studies may be due to differences in methodology, populations studied and diagnostic procedures, but it could indeed reflect a genuine absence of an association of dietary fish consumption with depressive status. The latter interpretation is consistent with null experimental findings on the effect of omega-3 supplementation to improve depressive symptoms in populations that are not deficient [68] and is in line with the findings of a recent updated systematic review of similar randomized controlled trials. [29] It has been hypothesized that gene polymorphisms, [69] that alter n-3 LC PUFAS absorption and metabolism, may explain some of this inconsistency, [70], [71] but the multi-factorial nature of depression suggests that genes may in fact interact with lifestyle factors and the environment in a very complex manner, which may over-shadow the role of fish and n-3 LC PUFAS.

Our results for the first time extend to low and middle income countries the general finding from large population-based and intervention studies that there is no evidence of an association of fish consumption with depression in later life. Experimental studies are surely needed to clarify any underlying biochemical and physiological mechanisms with which n-3 LC PUFAS may interact with mood disorders and depression. However, while fish represents a healthy dietary choice,[72][74] our results do not support the recommendation to increase fish consumption to prevent late life depression among older people in countries with low and middle incomes. Other modifiable risk factors should be targeted and prioritized and other actions taken [6] to address the forecast global epidemic of depression.


We are sincerely indebted to the older adults who took part enthusiastically in the 10/66 study and to their family members who warmly welcomed our interviewers and assessors in their households.

Author Contributions

Conceived and designed the experiments: EA MJP. Performed the experiments: MG DA YH KSJ JLR AS ALS JW. Analyzed the data: EA FLL. Contributed reagents/materials/analysis tools: EA. Wrote the paper: EA. Extensively revised the manuscript through all stages and provided relevant advice: ADD CPF CS.


  1. 1. Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJ (2006) Global and regional burden of disease and risk factors, 2001: systematic analysis of population health data. Lancet 367: 1747–1757.
  2. 2. Reynolds FC, Kupfer DJ (1999) Depression and Aging: A Look to the Future. Psychiatr Serv 50: 1167–1172.
  3. 3. Miranda JJ, Kinra S, Casas JP, Davey Smith G, Ebrahim S (2008) Non-communicable diseases in low- and middle-income countries: context, determinants and health policy. Trop Med Int Health 13: 1225–1234.
  4. 4. Phillips MR, Zhang J, Shi Q, Song Z, Ding Z, et al. (2009) Prevalence, treatment, and associated disability of mental disorders in four provinces in China during 2001–05: an epidemiological survey. Lancet 373: 2041–2053.
  5. 5. Chisholm D, Flisher AJ, Lund C, Patel V, Saxena S, et al. (2007) Scale up services for mental disorders: a call for action. Lancet 370: 1241–1252.
  6. 6. Patel V, Simon G, Chowdhary N, Kaaya S, Araya R (2009) Packages of care for depression in low- and middle-income countries. PLoS Med 6: e1000159.
  7. 7. WHO (2008) mhGAP : Mental Health Gap Action Programme : scaling up care for mental, neurological and substance use disorders. Geneva: World Health Organization.
  8. 8. Whyte EM, Rovner B (2006) Depression in late-life: shifting the paradigm from treatment to prevention. Int J Geriatr Psychiatry 21: 746–751.
  9. 9. Tiemeier H (2003) Biological risk factors for late life depression. Eur J Epidemiol 18: 745–750.
  10. 10. Mendlewicz J (1976) The age factor in depressive illness: some genetic considerations. J Gerontol 31: 300–303.
  11. 11. Calder PC (2003) Do omega-3 fatty acids ease the way to silent cell death? Nutrition 19: 472–473.
  12. 12. Yao JK, Sistilli CG, van Kammen DP (2003) Membrane polyunsaturated fatty acids and CSF cytokines in patients with schizophrenia. Prostaglandins Leukot Essent Fatty Acids 69: 429–436.
  13. 13. Uauy R, Dangour AD (2006) Nutrition in brain development and aging: role of essential fatty acids. Nutr Rev 64: S24–33; discussion S72–91.
  14. 14. Hibbeln JR (1998) Fish consumption and major depression. Lancet 351: 1213.
  15. 15. James MJ, Gibson RA, Cleland LG (2000) Dietary polyunsaturated fatty acids and inflammatory mediator production. Am J Clin Nutr 71: 343S–348S.
  16. 16. Haag M (2003) Essential fatty acids and the brain. Can J Psychiatry 48: 195–203.
  17. 17. Peet M (2004) International variations in the outcome of schizophrenia and the prevalence of depression in relation to national dietary practices: an ecological analysis. Br J Psychiatry 184: 404–408.
  18. 18. Hibbeln JR (2002) Seafood consumption, the DHA content of mothers’ milk and prevalence rates of postpartum depression: a cross-national, ecological analysis. J Affect Disord 69: 15–29.
  19. 19. Noaghiul S, Hibbeln JR (2003) Cross-national comparisons of seafood consumption and rates of bipolar disorders. Am J Psychiatry 160: 2222–2227.
  20. 20. Tanskanen A, Hibbeln JR, Hintikka J, Haatainen K, Honkalampi K, et al. (2001) Fish consumption, depression, and suicidality in a general population. Arch Gen Psychiatry 58: 512–513.
  21. 21. Tanskanen A, Hibbeln JR, Tuomilehto J, Uutela A, Haukkala A, et al. (2001) Fish consumption and depressive symptoms in the general population in Finland. Psychiatr Serv 52: 529–531.
  22. 22. Timonen M, Horrobin D, Jokelainen J, Laitinen J, Herva A, et al. (2004) Fish consumption and depression: the Northern Finland 1966 birth cohort study. J Affect Disord 82: 447–452.
  23. 23. Barberger-Gateau P, Jutand MA, Letenneur L, Larrieu S, Tavernier B, et al. (2005) Correlates of regular fish consumption in French elderly community dwellers: data from the Three-City study. Eur J Clin Nutr 59: 817–825.
  24. 24. Silvers KM, Scott KM (2002) Fish consumption and self-reported physical and mental health status. Public Health Nutr 5: 427–431.
  25. 25. Jacka FN, Pasco JA, Henry MJ, Kotowicz MA, Nicholson GC, et al. (2004) Dietary omega-3 fatty acids and depression in a community sample. Nutr Neurosci 7: 101–106.
  26. 26. Suzuki S, Akechi T, Kobayashi M, Taniguchi K, Goto K, et al. (2004) Daily omega-3 fatty acid intake and depression in Japanese patients with newly diagnosed lung cancer. Br J Cancer 90: 787–793.
  27. 27. Hakkarainen R, Partonen T, Haukka J, Virtamo J, Albanes D, et al. (2004) Is low dietary intake of omega-3 fatty acids associated with depression? Am J Psychiatry 161: 567–569.
  28. 28. Miyake Y, Sasaki S, Yokoyama T, Tanaka K, Ohya Y, et al. (2006) Risk of postpartum depression in relation to dietary fish and fat intake in Japan: the Osaka Maternal and Child Health Study. Psychol Med 36: 1727–1735.
  29. 29. Appleton KM, Rogers PJ, Ness AR (2010) Updated systematic review and meta-analysis fo the effects of n-3 long-chain polyunsaturated fatty acids on depressed mood. Am J Clin Nutr 91: 757–770.
  30. 30. Beekman AT, Deeg DJ, van Tilburg T, Smit JH, Hooijer C, et al. (1995) Major and minor depression in later life: a study of prevalence and risk factors. J Affect Disord 36: 65–75.
  31. 31. Beekman AT, Kriegsman DM, Deeg DJ, van Tilburg W (1995) The association of physical health and depressive symptoms in the older population: age and sex differences. Soc Psychiatry Psychiatr Epidemiol 30: 32–38.
  32. 32. Hackett ML, Anderson CS (2005) Predictors of depression after stroke: a systematic review of observational studies. Stroke 36: 2296–2301.
  33. 33. Dunn AJ, Swiergiel AH, de Beaurepaire R (2005) Cytokines as mediators of depression: what can we learn from animal studies? Neurosci Biobehav Rev 29: 891–909.
  34. 34. Bremmer MA, Beekman AT, Deeg DJ, Penninx BW, Dik MG, et al. (2008) Inflammatory markers in late-life depression: results from a population-based study. J Affect Disord 106: 249–255.
  35. 35. Appleton KM, Peters TJ, Hayward RC, Heatherley SV, McNaughton SA, et al. (2007) Depressed mood and n-3 polyunsaturated fatty acid intake from fish: non-linear or confounded association? Soc Psychiatry Psychiatr Epidemiol 42: 100–104.
  36. 36. Prince M, Ferri CP, Acosta D, Albanese E, Arizaga R, et al. (2007) The protocols for the 10/66 dementia research group population-based research programme. BMC Public Health 7: 165.
  37. 37. Prince M, Acosta D, Chiu H, Scazufca M, Varghese M (2003) Dementia diagnosis in developing countries: a cross-cultural validation study. Lancet 361: 909–917.
  38. 38. Copeland JR, Dewey ME, Griffiths-Jones HM (1986) A computerized psychiatric diagnostic system and case nomenclature for elderly subjects: GMS and AGECAT. Psychol Med 16: 89–99.
  39. 39. Broe GA, Akhtar AJ, Andrews GR, Caird FI, Gilmore AJ, et al. (1976) Neurological disorders in the elderly at home. J Neurol Neurosurg Psychiatry 39: 362–366.
  40. 40. Sosa AL, Albanese E, Prince M, Acosta D, Ferri CP, et al. (2009) Population normative data for the 10/66 Dementia Research Group cognitive test battery from Latin America, India and China: a cross-sectional survey. BMC Neurol 9: 48.
  41. 41. Prince M, Acosta D, Chiu H, Copeland J, Dewey M, et al. (2004) Effects of education and culture on the validity of the Geriatric Mental State and its AGECAT algorithm. Br J Psychiatry 185: 429–436.
  42. 42. Guerra M, Ferri CP, Sosa AL, Salas A, Gaona C, et al. (2009) Late-life depression in Peru, Mexico and Venezuela: the 10/66 population-based study. Br J Psychiatry 195: 510–515.
  43. 43. WHO (1987) Clinical Descriptions and Diagnostic Guidelines, MNH/MEP/87.1. In: WHO , editor. Geneva: World Health Organization.
  44. 44. Copeland JR, Prince M, Wilson KC, Dewey ME, Payne J, et al. (2002) The Geriatric Mental State Examination in the 21st century. Int J Geriatr Psychiatry 17: 729–732.
  45. 45. Prince MJ, Reischies F, Beekman AT, Fuhrer R, Jonker C, et al. (1999) Development of the EURO-D scale–a European, Union initiative to compare symptoms of depression in 14 European centres. Br J Psychiatry 174: 330–338.
  46. 46. Castro-Costa E, Dewey M, Stewart R, Banerjee S, Huppert F, et al. (2008) Ascertaining late-life depressive symptoms in Europe: an evaluation of the survey version of the EURO-D scale in 10 nations. The SHARE project. Int J Methods Psychiatr Res 17: 12–29.
  47. 47. Morris JC (1997) Clinical Dementia Rating: A Reliable and Valid Diagnostic and Staging Measure for Dementia of the Alzheimer Type. International Psychogenatrics 9: 173–176.
  48. 48. Albanese E, Dangour AD, Uauy R, Acosta D, Guerra M, et al. (2009) Dietary fish and meat intake and dementia in Latin America, China, and India: a 10/66 Dementia Research Group population-based study. Am J Clin Nutr 90: 392–400.
  49. 49. Duke University (1978) Multidimensional Functional Assessment: the OARS Methodology; Development DUCftSoAaH, editor. Durham, NC: Duke University.
  50. 50. Prince MJ, de Rodriguez JL, Noriega L, Lopez A, Acosta D, et al. (2008) The 10/66 Dementia Research Group’s fully operationalised DSM-IV dementia computerized diagnostic algorithm, compared with the 10/66 dementia algorithm and a clinician diagnosis: a population validation study. BMC Public Health 8: 219.
  51. 51. Maclure M, Greenland S (1992) Tests for trend and dose response: misinterpretations and alternatives. Am J Epidemiol 135: 96–104.
  52. 52. Higgins JP, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21: 1539–1558.
  53. 53. Tracy M, Hobfoll SE, Canetti-Nisim D, Galea S (2008) Predictors of depressive symptoms among israeli jews and arabs during the Al aqsa intifada: a population-based cohort study. Ann Epidemiol 18: 447–457.
  54. 54. Vuong Q (1989) Likelihood ratio tests for model selection and non-nested hypotheses. Econometrica 57: 307–333.
  55. 55. Llibre Rodriguez JJ, Ferri CP, Acosta D, Guerra M, Huang Y, et al. (2008) Prevalence of dementia in Latin America, India, and China: a population-based cross-sectional survey. Lancet 372: 464–474.
  56. 56. Jotheeswaran AT, Williams JD, Prince MJ (2010) Predictors of mortality among elderly people living in a south Indian urban community; a 10/66 Dementia Research Group prospective population-based cohort study. BMC Public Health 10: 366.
  57. 57. DiMatteo MR, Lepper HS, Croghan TW (2000) Depression is a risk factor for noncompliance with medical treatment: meta-analysis of the effects of anxiety and depression on patient adherence. Arch Intern Med 160: 2101–2107.
  58. 58. Hofler M (2005) The Bradford Hill considerations on causality: a counterfactual perspective. Emerg Themes Epidemiol 2: 11.
  59. 59. Burdge GC, Finnegan YE, Minihane AM, Williams CM, Wootton SA (2003) Effect of altered dietary n-3 fatty acid intake upon plasma lipid fatty acid composition, conversion of [13C]alpha-linolenic acid to longer-chain fatty acids and partitioning towards beta-oxidation in older men. Br J Nutr 90: 311–321.
  60. 60. Visioli F, Rise P, Barassi MC, Marangoni F, Galli C (2003) Dietary intake of fish vs. formulations leads to higher plasma concentrations of n-3 fatty acids. Lipids 38: 415–418.
  61. 61. Margetts B, Nelson MC (1997) Design Cocepts in Nutritional Epidemiology. Oxford: Oxford University Press.
  62. 62. Goris AH, Westerterp-Plantenga MS, Westerterp KR (2000) Undereating and underrecording of habitual food intake in obese men: selective underreporting of fat intake. Am J Clin Nutr 71: 130–134.
  63. 63. Castro-Costa E, Dewey M, Stewart R, Banerjee S, Huppert F, et al. (2007) Prevalence of depressive symptoms and syndromes in later life in ten European countries: the SHARE study. Br J Psychiatry 191: 393–401.
  64. 64. Appleton KM, Woodside JV, Yarnell JW, Arveiler D, Haas B, et al. (2007) Depressed mood and dietary fish intake: direct relationship or indirect relationship as a result of diet and lifestyle? J Affect Disord 104: 217–223.
  65. 65. Colangelo LA, He K, Whooley MA, Daviglus ML, Liu K (2009) Higher dietary intake of long-chain omega-3 polyunsaturated fatty acids is inversely associated with depressive symptoms in women. Nutrition 25: 1011–1019.
  66. 66. Suominen-Taipale AL, Turunen AW, Partonen T, Kaprio J, Mannisto S, et al. (2010) Fish consumption and polyunsaturated fatty acids in relation to psychological distress. Int J Epidemiol 39: 494–503.
  67. 67. Sanchez-Villegas A, Delgado-Rodriguez M, Alonso A, Schlatter J, Lahortiga F, et al. (2009) Association of the Mediterranean dietary pattern with the incidence of depression: the Seguimiento Universidad de Navarra/University of Navarra follow-up (SUN) cohort. Arch Gen Psychiatry 66: 1090–1098.
  68. 68. Grenyer BF, Crowe T, Meyer B, Owen AJ, Grigonis-Deane EM, et al. (2007) Fish oil supplementation in the treatment of major depression: a randomised double-blind placebo-controlled trial. Prog Neuropsychopharmacol Biol Psychiatry 31: 1393–1396.
  69. 69. Appleton KM, Rogers PJ, Ness AR (2008) Is there a role for n-3 long-chain polyunsaturated fatty acids in the regulation of mood and behaviour? A review of the evidence to date from epidemiological studies, clinical studies and intervention trials. Nutr Res Rev 21: 13–41.
  70. 70. Covault J, Pettinati H, Moak D, Mueller T, Kranzler HR (2004) Association of a long-chain fatty acid-CoA ligase 4 gene polymorphism with depression and with enhanced niacin-induced dermal erythema. Am J Med Genet B Neuropsychiatr Genet 127B: 42–47.
  71. 71. Brookes KJ, Chen W, Xu X, Taylor E, Asherson P (2006) Association of fatty acid desaturase genes with attention-deficit/hyperactivity disorder. Biol Psychiatry 60: 1053–1061.
  72. 72. Hostenkamp G, Sorensen J (2009) Are fish eaters healthier and do they consume less health-care resources? Public Health Nutr. pp. 1–8.
  73. 73. Kaushik M, Mozaffarian D, Spiegelman D, Manson JE, Willett WC, et al. (2009) Long-chain omega-3 fatty acids, fish intake, and the risk of type 2 diabetes mellitus. Am J Clin Nutr 90: 613–620.
  74. 74. Virtanen JK, Mozaffarian D, Chiuve SE, Rimm EB (2008) Fish consumption and risk of major chronic disease in men. Am J Clin Nutr 88: 1618–1625.