To our knowledge this is the first large epidemiological study to show a relationship between vegetarianism and significant depressive symptoms among adult men. Here we found that self –identification as a vegetarian was associated with an increased risk of depressive symptoms evaluated both as a continuous scale and using a cut-off of greater than 10 on the EPDS. These associations remained after adjustment for vulnerability factors for depression including a family history of depression, previous childhood psychiatric contact, highest educational level reached, housing tenure, age, ethnic origin, number of children in the household, marital status, employment status, alcohol and tobacco consumption and religious identification. There was a trend evident comparing the duration of vegetarianism and depressed mood although this failed to reach significance (data not shown).

Several possible factors previously linked to an increased risk of depressive symptoms might underlie this increased risk of depression among vegetarians. Any adverse effect on mood associated with vegetarian diet may result from contributions from multiple interactive nutrients including both type and quantity of dietary fat intake between vegetarians and non-vegetarians. For example, vegetarians are known to have lower intakes of n-3 HUFAs, vitamin B₁₂ and folate, greater consumption of nuts rich in omega-6 fatty acids which may be associated with greater risk of depression (Wolfe et al., 2009Wolfe et al., 2009) and a possibly increased likelihood of persons with depressive symptoms to change their dietary preferences. Other potential factors include high blood levels of phytoestrogens (consequent mainly on diets rich in vegetables and soya) and metabolites of pesticides (consequent upon relatively high intakes of fruit and vegetables), which are more likely to be found among vegetarians (Tordjman et al., 2016Tordjman et al., 2016). Another potential contributing factor is lower intakes of seafood (Li et al., 2016Li et al., 2016) and low omega-3 HUFA blood levels (Lin et al., 2010Lin et al., 2010) are thought to be associated with greater risk of depressive symptoms. Several meta-analyses of randomized, placebo-controlled intervention trials have found effect sizes comparable to those of pharmaceutical antidepressants, some specifically indicating that eicosapentaenoic rich preparations are significantly more effective (Grosso et al., 2014, Hallahan et al., 2016). One additional concern is that people who opt for the vegetarian diet might be intending to lose weight or that there are differences in the prevalence of obesity and overweight participants; medical conditions clearly associated with depression. However, whilst non-vegetarians had higher BMIs 25.20 (3.29) [Mean (SD)] compared to vegetarians, 24.32 (3.00) p < 0.001, depression and BMI were not associated: EPDS < 10 25.16 (3.26) EPDS 10+ 25.27 (3.50) p = 0.405.

Since exclusion of red meat primarily characterizes vegetarians, lower intakes of vitamin B₁₂ merit consideration as a contributing factor. Curiously in this study, 72% of self-reported vegans and 4.7% of vegetarians reported some current consumption of red meat (data not shown). However, although self-identification and dietary practice may differ substantially, self-reported vegetarians in this study are considerably less likely to consume red meat or poultry than non-vegetarians. This is consistent with prior reports from the UK Women's cohort study that of the 28% of subjects self-reported as being vegetarian and 1% as vegan, only 18% would be defined as 'vegetarian' from dietary habits reported in a food frequency questionnaire (Cade et al., 2004Cade et al., 2004). Our findings are also consistent with an evaluation of 1046 Australian women where lower red meat consumption was associated with nearly a doubling of risk for major depressive and anxiety disorders. A trend towards a U shaped curve, with greater risks of major depression or dysthymia among Australian women consuming more (n = 27) or less (n = 21) than recommended amounts was reported (Jacka et al., 2012Jacka et al., 2012). However these results must be interpreted with caution as they were only significant for greater red meat consumption when adjusted for traditional dietary patterns, which include meat consumption and in comparison to n = 12 women consuming recommended amounts of red meat. Other than vitamin B₁₂ and differences in fat consumption, the EPIC-Oxford study indicates that differences in nutrient intake between British vegetarians are modest compared with non- vegetarians (the few vegans excepted) (Davey et al., 2003Davey et al., 2003). The authors found that 52% of self-reported vegans and 7% of vegetarians were vitamin B₁₂ deficient (defined as serum vitamin B₁₂ < 118 pmol/l) (Gilsing et al., 2010Gilsing et al., 2010). Several observational studies have reported associations with elevations in homocystine or low vitamin B₁₂ and risk of major and minor depression (Penninx et al., 2000Penninx et al., 2000). DSM-IV diagnosis of depression was reduced by 50% by daily folic acid (2 mg), vitamin B₆ (25 mg), and vitamin B₁₂ (0.5 mg) in a randomized, placebo-controlled trial of survivors of stroke for 1–10.5 years (Almeida et al., 2010Almeida et al., 2010). The study presented here, as others, fails to resolve whether the increased risk of depressive symptoms in this population was due to lower intake of nutrients rich in red meats, among which vitamin B₁₂ is a plausible candidate. A randomized placebo controlled trial is warranted to determine whether depressive symptoms among vegetarians can be reduced by supplementing with folic acid and other B vitamins to lower homocystine levels. By maintaining sufficient B vitamin and folate levels, vegetarians may be able to have the benefits of better health, and harmony with ethical and cultural beliefs, while reducing risks of depressive symptoms.

Relatively few intervention trials have been conducted to assess the effects of assignment of a vegetarian diet on mental health outcomes and all of these studies may have methodological issues due to the expectancy of benefit and difficulty in achieving adequate blinding of dietary assignment and non-specific assessment of symptoms. In one of these studies global mood scores improved in association with carbohydrate intake and plasma tryptophan measures after six weeks on a vegetarian diet (Schweiger et al., 1986Schweiger et al., 1986), and in another improved mental health, vitality and fewer impairments in work productivity were reported after a 22 week intervention (Katcher et al., 2010Katcher et al., 2010). Perhaps reasons cited by former vegetarians for resuming omnivorous diets are also indicators of mild psychiatric effects of their dietary changes. The reasons reported included not feeling healthy and concern about their nutritional status (in addition to missing the taste of meat) (Barr and Chapman, 2002Barr and Chapman, 2002).

However, it is possible that the increased risk for elevated depressive symptoms found here is not related to any dietary differences, but is due to intrinsic differences in rates of psychiatric or personality problems comparing vegetarians to non-vegetarians. It is possible that for some proportion of the population, vegetarianism is not chosen for health, religious or ethical reasons, but is a marker for other psychiatric disorders manifesting with symptoms of both eating disorders and depressive symptoms. For example, among adolescents in Minnesota vegetarianism was associated with a cluster of problems including dissatisfaction with their bodies and unhealthy weight control practices, especially among males, and increased suicide risk indicators (Perry et al., 2001Perry et al., 2001); Turkish vegetarian adolescents had a cluster of problems including abnormal eating attitudes, low self-esteem, social physique and social trait anxiety (Bas et al., 2005Bas et al., 2005). However, it is unclear if this clustering of problems among adolescents continues through to adult populations who frequently identify health and ethical reasons for their choice of vegetarianism. Here, self-identification as a vegetarian/vegan was not influenced by a parental history of depression or by childhood contact with psychiatric services. However, subjects with depressive symptoms are well known to have alterations in dietary preference and appetite resulting in weight gain or weight loss. In contrast, Janelle and Barr (1995)Janelle and Barr (1995) found no evidence of increased risk of eating disorders among adult vegetarian women after assessing dietary restraint (conscious limitation of food intake), disinhibition, and hunger assessed by the Three-Factor Eating Questionnaire.

Some methodological limitations in the study reported here need consideration. First, there are no data on serum cholesterol or fatty acids or vitamin B₁₂ in these subjects. Second, it is possible that by correcting for variables that may be a consequence of depression such as marital status, employment status, alcohol and tobacco consumption, a realistic relationship with depression may be masked. However, variables such as these can be both potential causes and consequences of depression; nevertheless even when taking these variables into account the relationship remained. Another consideration is that self-report questions do not allow confirmation of a diagnosis of depression and a bias may result from non-response by depressed subjects, or the decision by mothers to invite their partners to opt in to the study. However the results presented here do indicate that male British vegetarians did have a greater risk for depressive symptoms after adjusting for multiple confounding variables. We also cannot rule out the possibility of reverse causation, that is that having depressive symptoms might change dietary habits and increase the likelihood of being a vegetarian. This study does not resolve the question of whether adoption of a vegetarian diet will increase, or decrease the risk of depressive symptoms or affect mental well-being or what specific nutrients, if any, may influence those risks, but does suggest that a randomized controlled trial of selected nutrients or foods may be warranted.