Meat and mental health: A meta-analysis of meat consumption, depression, and anxiety

Abstract

In this meta-analysis, we examined the quantitative relation between meat consumption or avoidance, depression, and anxiety. In June 2020, we searched five online databases for primary studies examining differences in depression and anxiety between meat abstainers and meat consumers that offered a clear (dichotomous) distinction between these groups. Twenty studies met the selection criteria representing 171,802 participants with 157,778 meat consumers and 13,259 meat abstainers. We calculated the magnitude of the effect between meat consumers and meat abstainers with bias correction (Hedges’s g effect size) where higher and positive scores reflect better outcomes for meat consumers. Meat consumption was associated with lower depression (Hedges’s g = 0.216, 95% CI [0.14 to 0.30], p < .001) and lower anxiety (g = 0.17, 95% CI [0.03 to 0.31], p = .02) compared to meat abstention. Compared to vegans, meat consumers experienced both lower depression (g = 0.26, 95% CI [0.01 to 0.51], p = .041) and anxiety (g = 0.15, 95% CI [-0.40 to 0.69], p = .598). Sex did not modify these relations. Study quality explained 58% and 76% of between-studies heterogeneity in depression and anxiety, respectively. The analysis also showed that the more rigorous the study, the more positive and consistent the relation between meat consumption and better mental health. The current body of evidence precludes causal and temporal inferences.

Keywords:

• anxiety
• depression
• meat
• mental health
• vegan
• vegetarianism
• sex

Many have unskillfully written upon the preservation of health; particularly by attributing too much to the choice, and too little to the quality of meats.

Francis Bacon

Introduction

In 2017, the World Health Organization (WHO) reported that mental illness was the leading cause of disability worldwide (World Health Organization 2017). They estimated that over 300 million people suffer from depression and over 260 million people suffer from anxiety (Friedrich 2017; Kessler et al. 2005). These estimates reflect a substantial increase in the number of people living with mental disorders over the past two decades (Vos et al. 2015; World Health Organization 2017). Given that dietary intake is considered a major contributor to mental health, it is important to provide rigorous quantitative evidence on diet—mental health relations to inform both public policy and clinical practice.

In parallel with increments in mental disorders, vegetarianism and veganism (i.e., avoidance of meat or all animal products, respectively) have become more popular and prevalent worldwide. The drivers of these corresponding trends are varied and include ethical, environmental, and animal rights-related reasons, (Kerschke-Risch 2015; Kerschke-Risch 2016; Rosenfeld 2019; Whorton 1994) as well as attempts to treat or ameliorate mental illnesses via dietary restrictions. As a result, the question of whether meat consumption or avoidance is associated with better mental health has become a controversial issue in public health and nutrition science (Jacka et al. 2012; Mofrad et al. 2021; Zeraatkar et al. 2019).

To date, the research has been equivocal and our recent systematic review summarized these findings (Dobersek et al., 2020). Briefly, we found that the majority of studies showed that meat abstainers (vegetarians and vegans) had substantially higher rates or risk of depression, anxiety, and/or self-harm (e.g., suicide) (Baines, Powers, and Brown 2007; Hibbeln et al. 2018; Matta et al. 2018; Michalak, Zhang, and Jacobi 2012). Additionally, these groups were more likely to be prescribed medication for mental-health issues (Baines, Powers, and Brown 2007). However, some studies suggested the opposite conclusion (Beezhold et al. 2015; Beezhold, Johnston, and Daigle 2010), and the findings regarding other less clearly defined concepts (e.g., mood states, affective well-being, stress perception, quality of life) were less obvious (Beezhold et al. 2015; Beezhold and Johnston 2012; Beezhold, Johnston, and Daigle 2010; Boldt et al. 2018; Pfeiler and Egloff 2018; Wirnitzer et al., 2018).

Our analysis showed that these contradictory findings were engendered by numerous factors (Dobersek et al., 2020). These included the inconsistent categorization of vegans and vegetarians, poor operationalization of psychological outcomes, biased recruitment and sampling strategies (Beezhold and Johnston 2012; Beezhold, Johnston, and Daigle 2010; Lindeman 2002; Perry et al. 2001; Timko, Hormes, and Chubski 2012), invalid dietary assessment protocols (e.g., food frequency questionnaires, FFQs) (Archer, Pavela, and Lavie 2015; Archer, Hand, and Blair 2013; Archer, Lavie, and Hill 2018a; Archer, Marlow, and Lavie 2018b, 2018c), and statistical and communication errors (e.g., failure to correct for multiple comparisons, inappropriate use of causal language). In summary, study rigor and quality appeared to be the greatest generator of inconsistent findings (Dobersek et al., 2020).

While our systematic review (Dobersek et al. 2020) provided rigorous evidence suggesting that meat consumers had lower rates or prevalence of mental health issues, we did not examine the strength of this relation. Therefore, the purpose of this meta-analytic review was twofold. First, to provide a comprehensive assessment of the quantitative relation between meat consumption or avoidance and mental health, and second, to examine the effect of study quality on this relationship. To avoid inconsistent definitions of meat abstention, we captured only studies that offered a clear dichotomy between individuals who reported consuming meat (i.e., omnivores) and those who reported to be meat abstainers (i.e., vegans, vegetarians). Our focus was limited to the two most prevalent mental-health outcomes worldwide: depression and anxiety.

Methods

Search strategy

The search strategy was adopted from our previous systematic review (Dobersek et al. 2020) and updated to include all papers published through June 18^th, 2020. Five online databases (i.e., PubMed, PsycINFO, CINAHL Plus, Medline with full text, and Cochrane Library) were searched using a broad set of keywords for primary research examining depression and anxiety in meat consumers and meat abstainers. Supplemental File 1 provides details of the search terms and strategy. Two teams of two persons independently conducted the searches and imported them into reference-managing software (EndNote X9, 2019). After excluding duplicate articles, both teams independently identified potentially relevant articles by screening their titles and abstracts. Full texts of potentially relevant articles were then assessed by both teams and the lead investigator (UD) working independently. Reference lists from previous reviews and research papers were also examined. The teams and lead investigator then met to arrive at a consensus on the inclusion/exclusion criteria for each paper (e.g., depression/anxiety, strict definitions of dietary intake, etc.). Disputes were adjudicated by discussion, with the final decision made by the lead investigator. Consensus was obtained for all included articles.

Study inclusion criteria

The studies were included if 1) they were written in English, 2) the authors provided a clear distinction between participants who reported eating meat (i.e., meat consumers; e.g., omnivores) and those who avoided meat consumption (i.e., meat abstainers; e.g., vegans, vegetarians), 3) they included depression and/or anxiety as psychological outcomes, and 4) they provided enough statistical information to calculate the effect sizes (ESs). No design, geographical, cultural, or time-period (publication-date) restrictions were set.

Study exclusion criteria

Studies were excluded if they 1) were written in a non-English language, 2) did not assess depression or anxiety, 3) did not provide (or we were unable to acquire) information needed to calculate ESs, and 4) examined meat consumption as a continuous or multi-level variable. This latter focus was necessary for the following reasons. The use of self-reported (memory-based) dietary assessments (FFQ) induces nonquantifiable measurement error due to intentional and/or non-intentional misreporting and the invalidity of pseudo-quantification (assigning nutrient and caloric values to memories of consumed foods and beverages) (Archer, Lavie, and Hill 2018a; Archer, Marlow, and Lavie 2018b; Archer, Pavela, and Lavie 2015). This nonquantifiable error precludes examining meat consumption as a continuous variable. Therefore, by examining meat consumption as a dichotomous (yes/no) variable, we potentially eliminated the nonquantifiable error and the error propagated via pseudo-quantification.

Table 1 provides the detailed description of inclusion/exclusion criteria according to a Population, Exposure/Intervention/Comparison, Outcomes, and Study Design (PE/I/COS) framework (Brown et al. 2006; Huang, Lin, and Demner-Fushman 2006).

Table 1. Detailed description of selection criteria according to the PE/I/COS Framework (Brown et al. 2006; Huang, Lin, and Demner-Fushman 2006).

Search Strategy	Details
Inclusion criteria	P: Adults and adolescents
	E/I/C: Clear distinction between meat consumers and meat abstainers
	O: Depression and anxiety assessed via self-report, use of prescription medication, doctor’s diagnosis, or diagnostic interview
	S: Observational studies (e.g., cross-sectional studies, longitudinal studies, case control), or non-/randomized control trials (RCTs)
Exclusion criteria	P: N/A
	E/I/C: No clear distinction between meat consumers and meat abstainers
	E/I/C: Meat consumption as a continuous variable
	O: Other psychological outcomes and disorders (e.g., disordered eating, dementia, self-harm, well-being, etc.)
	S: Qualitative studies, reviews, letters, book chapters, articles without quantitative data
Language	English
Time filter	None
Database	PubMed, PsycINFO, CINAHL Plus, Medline with full text, Cochrane Library

Note. P = Population, E/I/C = Exposure/Intervention/Comparison, O = Outcomes, and S = Study Design.

Data extraction

Qualitative and quantitative data extraction were performed independently by both teams and overseen by the lead investigator. The information extracted from each study included overall design, participant characteristics (e.g., population, sample size, age), assessment methods for depression, anxiety, and diet, the definitions of meat consumers/abstainers, time of dietary assessment/diet adoption, factors adjusted for in the analyses, key findings, and relevant data to calculate the ESs (e.g., means, standard deviations, proportions, standard errors, p-values). The reviewers were not blinded and had full access to paper details (e.g., authors, affiliations, and journals) during data extraction and compilation. The extraction tables were examined for accuracy and completeness by the lead investigator. Once compiled, information from the selected studies was transferred to a ‘Summary Table’ (Table 2).

Table 2. A summary of extracted participant and study characteristics included in the meta-analysis in alphabetical order.

Citation and study design	Design	Participants (sample size, population, setting)	Psychological health outcome (Description/scale/ measure)	Dietary assessment (description & measure)	Time of assessment/ adoption of a diet	Factors adjusted for in analyses	Key findings	Measure of association & risk of bias
Baines, S., Powers, J., & Brown, W. J. (2007). Public Health Nutr, 10(5), 436-442.	Cross-Sectional	9113 Australian women between 22 and 27 years of age	Depression: Have you taken prescription medication for depression in the last 4 weeks? Anxiety: Have you ever been told by a doctor that you have anxiety?	Vegetarians (n = 252) excluded red meat, poultry, and fish; semi-vegetarians excluded red meat (n = 827); Non-vegetarians (n = 8034) ate red meat.	NR	Used weighting to correct for over-sampling women living in rural/remote areas	Non-vegetarians were less likely to be diagnosed with depression and anxiety than either vegetarian group. Depression and related symptoms were more commonly reported by vegetarians. Vegetarians and semi-vegetarians were more likely to take prescription medication for depression than non-vegetarians.	Proportions High quality
Bas, M., Karabudak, E., & Kiziltan, G. (2005). Appetite, 44(3), 309-315.	Cross-sectional	1205 students from Turkey; 597 females and 608 males between 17 and 21 years of age with a mean age of 21.5 (SD = 1.9)	Anxiety: - The Social Physique Anxiety Scale (SPAS) - The State-Trait Anxiety Inventory (STAI): the mean values in the normative study ranged from 36 to 41; higher scores indicate higher levels of anxiety	Participants were asked "Are you a vegetarian? Yes/No Vegetarians (n = 31), non-vegetarians (n = 1174).	NR	NR	Female vegetarians scored significantly higher on the STAI than female non-vegetarians (and on SPAS, but the relationship was NS. Male and female vegetarians scored higher on SPAS than non-vegetarians (NS).	M, SD Moderate quality
Beezhold, B., Radnitz, C., Rinne, A., & DiMatteo, J. (2015). Nutr Neurosci, 18(7), 289-296.	Cross-sectional	620 adults majority from North America (Canada, USA), and some from outside of North America; 487 females and 133 males between 25-60 years of age with a mean age of 34.83 (SD = 10.2)	The Depression Anxiety Stress Scale −21 (DASS): - Depression: normal scores 0-9 - Anxiety: normal scores 0-7	Vegetarians consumed only eggs or dairy (n = 109). Vegans consumed no animal foods (n = 283). Omnivores consumed meat, poultry, or fish (n = 228).	At least monthly	Age	Vegans reported significantly lower anxiety than omnivores. Vegans and vegetarian males had significantly lower anxiety than omnivores.	M, SD Very low quality
Beezhold, B. L., Johnston, C. S., & Daigle, D. R. (2010). Nutr J, 9, 26.	Cross-sectional	138 Seventh Day Adventists from USA 77 females and 61 males with a mean age of 43.01 (SD = 1.41)	The Depression Anxiety Stress Scale (DASS): - reported normative score for nonclinical population = 18.38 The Profile of Mood States (POMS): - normative scores: males = 14.8; females = 20.3	Vegetarians excluded all flesh foods (n = 60). Omnivores (n = 78)	NR	Confounding variables, but NS	Vegetarians reported significantly lower depression and anxiety than vegetarians.	M, SE Transferred SE into SD and combined the DASS and POMS scores Low quality
Beezhold, B. L., & Johnston, C. S. (2012). Nutr J, 11, 9.	Randomized control trial	39 adults from USA; 32 females and 7 males; age - NR	The Depression Anxiety Stress Scales (DASS) The Profile of Mood States (POMS)	Vegetarians avoided all animal products except dairy (n = 13). Fish eaters avoided meat and poultry, consumed weekly seafood (n = 13) Omnivores consumed meat, fish, and poultry (n = 13).	2 weeks	NR	Vegetarians reported significantly better mood than omnivores and fish eaters after the trial.	M, SD were provided by the authors Combined the DASS and POMS scores and used post-trial only Very low quality
Forestell, C. A., & Nezlek, J. B. (2018). Ecol Food Nutr, 57(3), 246-259.	Cross-sectional	6450* USA undergraduate students; **3707 females and 2715 males between 16 and 47 years of age with a mean age of 18.96 (SD = 1.39)	The Center for Epidemiologic Studies Depression Scale (CESD): - scores range between 0-60	“Which of the following categories best characterizes participants’ eating behavior? Options: Vegan, lacto-vegetarian, lacto-ovo-vegetarian, pesco-vegetarian, semi-vegetarian, occasional omnivore, omnivore. Vegetarians (vegans, lacto-vegetarians, lacto-ovo-vegetarians), were defined as those who restricted all meat and fish products (n = 276). Semi-vegetarians (pesco-vegetarians, semi-vegetarians, occasional omnivores) were defined as those who ate fish or white meat or red meat occasionally (n = 1191). Omnivores were defined as those who regularly ate fish and/or meat (n = 4955).	NR	NR	Vegetarians and semi- vegetarians were significantly more depressed than omnivores.	M, SD Number of participants from male/female group was calculated from the data set available on online – re-analyzed the data to check it is consistent with the results reported in the article Moderate-low quality
Goh, C. M. J., Abdin, E., Jeyagurunathan, A., Shafie, S., Sambasivam, R., Zhang, Y. J., et al. (2019). BMC geriatrics, 19(1), 1-9.	Cross-sectional	*2565 Singapore citizens & permanent residents; 1117 males, and 1448 females between 60 and 105 years of age (M = 72.7, SD = 9.53)	The Geriatric Mental State with Automated Geriatric Examination for Computer Assisted Taxonomy (GMS-AGECAT): - scores 3 or higher were classified as depression - omitted subsyndromal depression	"How often do you eat meat?" Never, some days, most days, every day Omnivores (ate meat somedays, most days, or everyday) = 2345 Vegans/vegetarians (never consume meat) = 193	NR	Sociodemographic variables and health status (i.e., smoking)	Meat-consumers are more likely to show signs of depression than meat-abstainers.	Proportions for sex were provided by the authors Low quality
Hessler-Kaufmann, J. B., Meule, A., Holzapfel, C., Brandl, B., Greetfeld, M., Skurk, T., et al. (2020).	Cross-sectional	511 adults from Germany; 187 males, 324 females between 18 and 84 (M = 43.4, SD = 18.1)	The Patient Health Questionnaire-9 (German version): - scores range between 0-27, higher score indicates more severe depressive symptoms	"I eat vegetarian (no meat, no poultry, no fish)." = vegetarians "I eat vegan (no products derived from animals)" = vegans (n of vegans + vegetarians = 49) “I mostly eat vegetarian (no red meat but sometimes poultry and/or fish)" = semi-vegetarians (n = 98) Omnivores = 364	NR	Age, sex, BMI	Semi-vegetarians had significantly higher depression than omnivores.	M, SD for sex were provided by the authors Low quality
Hibbeln, J. R., Northstone, K., Evans, J., & Golding, J. (2018). J Affect Disord, 225, 13-17.	Cross-sectional	9668 male partners of English women in early stages of pregnancy; report only age categories (<25 to 35+)	The Edinburgh Postnatal Depression Scale (EPDS): - a score above 12 indicates a high probability of severe depression (only scores above 10 were included in the original study analyses)	"How many times nowadays do you eat " followed by 17 categories of foods: Yes/Never Vegetarian/vegan (n = 350) Omnivore (n = 9318)	NR (between less than 1 to 41 years being vegetarian; 2/3 of participants were vegetarians for more than 10 years)	Social and lifestyle variables: housing tenure, number of children in the household, age, religion, family history of depression, child psychiatric contact; alcohol and tobacco consumption, marital and employment status, BMI	Vegetarians had significantly higher depression levels than non-vegetarians.	M, SD Moderate quality
Jin, Y., Kandula, N. R., Kanaya, A. M., & Talegawkar, S. A. (2019). Ethnicity & health, 1-8.	Cross-sectional	892 South Asians living USA; 420 females and 472 males between 40-83 years of age (M = 55.3, SD = 9.4)	The Center for Epidemiologic Studies Depression Scale (CESD): - scores range between 0-60 - participants with scores > 16 were considered to have depression	Study of Health Assessment and Risk in Ethnic Groups FFQ Vegetarians: no consumption of meat, poultry, or fish (n = 335) Non-vegetarians (n = 557)	Previous year	Age, sex, study site, education, income, smoking status, BMI, acculturation, intentional exercise, alcohol and energy intake, antidepressant medication use	Vegetarian diet was associated with lower odds of depression than non-vegetarian diet.	Proportions Low quality
Kapoor et al. (2017). Neuroscience, 22, 228-232.	Cross-sectional	200 adults from Pakistan; 149 males, 51 females between 20 and 40 years of age (M = 28.25, SD = 5.78)	The Hamilton Rating Scale for Depression	Vegetarians (n = 100) ate small consumption of dairy products & no animal sourced food (e.g., meat, poultry, fish, or eggs). Omnivores (n = 100) ate all types of food.	Since their childhood	NR	Greater number of vegetarians were depressed than omnivores.	Proportions Moderate quality
Lavallee, K., Zhang, X. C., Michalak, J., Schneider, S., & Margraf, J. (2019). Journal of Affective Disorders, 248, 147-154.	Cross-sectional, longitudinal	*22417 adults from Germany, Russia, USA, and China; **13006 females and 8596 males with a mean age of 39.10 (SD = 11.68)	The Depression Anxiety and Stress Scales-21 (DASS)	Participants were asked "Are you currently vegetarian?" Yes/No Vegetarians: excluded meat and/or fish (n = 3400). Non-vegetarians: included meat (n = 18603).	Currently	Marital and socioeconomic status, urbanicity, quality of health, BMI, family affluence, age, gender, education level	Cross-sectional: Vegetarian diet was related to higher levels of depression and anxiety among Chinese students. Longitudinal: Vegetarian diet was predictive for depression and anxiety in the Chinese student sample (in that vegetarians had slightly higher levels of depression and anxiety at follow-up).	M, SD were calculated from the data available online – re-analyzed the data to check it is consistent with the results reported in the article; post-findings only for German & Chinese Moderate-low quality
Lindeman, M. (2002). Ecology of food and nutrition, 41(1), 75-86.	Cross-sectional	Study 1: 308 women from Finland, between 13 and 74 years of age with a mean age of 29 (SD = 10.81)	Study 1: The Center for Epidemiologic Studies Depression Scale (CESDS-SF): - scores can range between 0-30 with higher scores indicating greater depressed moods	Study 1: Semi-vegetarians were defined as those who avoided red meat or only ate fish & vegetarian dishes (n = 69), omnivores (n = 197), vegetarians (n = 42).	NR	NR	Study 1: Semi-vegetarians and vegetarians had more symptoms of depression than omnivores.	M, SD Low quality
Matta, J., Czernichow, S., Kesse-Guyot, E., Hoertel, N., Limosin, F., Goldberg, M., . . . Lemogne, C. (2018). Nutrients, 10(11).	Cross-sectional	90380 adults from France; 48035 females and 42345 males who were between 18 and 69 years of age with a mean age of 42.78 (SD = 13.45)	The Center of Epidemiologic Studies-Depression (CES-D) scale: - scores range between 0-60 - depression was defined by scores ≥ 19	Omnivores: consumed all foods (n = 88905). Pesco-vegetarians: did not consume any meat nor poultry but consumed fish, eggs, milk, and dairies (n = 832). Lacto-ovo-vegetarians: consumed eggs, milk, and dairies, but omitted meat, poultry, and fish (n = 562). Vegans: no consumption of animal products (n = 81).	NR	Age, sex, education, household income, frequency of fruits, vegetables, legumes, and grains consumption, alcohol and tobacco consumption, physical activity, health related concerns	The odds of depressive symptoms increased when meat was excluded from the diet. The prevalence of depressive symptoms increased with the number of excluded food groups.	Proportions for sex were provided by the authors High quality
Michalak, J., Zhang, X. C., & Jacobi, F. (2012). Int J Behav Nutr Phys Act, 9, 67.	Cross-sectional	4181* adults from Germany; **2234 females and 1882 males between 18 and 65 years of age with a mean age of 38.87 (SD = 13.33)	The Munich Composite International Diagnostic Interview (65 min average interview)	"Do you currently follow a vegetarian diet (no meat) or did you follow a vegetarian diet in the past?" Answers: ‘no, never’, ‘yes, completely’, or ‘yes, predominantly’. The word ‘meat’ in German language excludes poultry. Vegetarians (n = 54) Predominantly vegetarian (n = 190) Non-vegetarians (n = 3872)	In the past 12 months	Non-vegetarians matched on sociodemographic characteristics with completely and predominantly vegetarians	Vegetarians displayed elevated prevalence rates for depressive disorders and anxiety disorders than non-vegetarians.	Proportions for sex were provided by the authors; used lifetime prevalence only High quality
Paslakis, G., Richardson, C., Nöhre, M., Brähler, E., Holzapfel, C., Hilbert, A., et al. (2020). Scientific Reports, 10(1), 1-10.	Cross-sectional	2449* adults from Germany; 1138 males and 1311 females (M = 49.6, SD = 17.1)	The Patient Health Questionnaire-4 (PHQ-4): - scores range from 0-12 with higher scores indicating greater psychological distress	*"Have you been consciously eating a vegetarian (vegan) diet for at least last 2 weeks? Vegetarians/vegans: omitting meat/all foods of animal origin (n = 133) Omnivores (n = 2311)	Last 2 weeks	Gender, age	Vegetarians reported significantly higher anxiety than omnivores. They also reported greater depression scores (NS) than omnivores.	M, SD were provided by the authors Moderate quality
Perry, C. L., McGuire, M. T., Neumark-Sztainer, D., & Story, M. (2001). Journal of Adolescent Health, 29(6), 406-416.	Cross-sectional	4746* USA adolescents; **2255 females and 2265 males between 11 and 18 years of age with a mean age of 14.9	Depressive symptoms were assessed on a 7-item standardized scale	Participants were asked "Are you a vegetarian?" They answered either Yes/No Vegetarians (n = 262); non-vegetarians (n = 4258)	43.4% participants were vegetarians for over 2 years; 56.6% participants were vegetarians for 2 years or less	School, gender	Greater number of vegetarians were depressed than non-vegetarians (NS).	Proportions were provided by the authors Moderate-low quality
Stokes, N., Gordon, C. M., & DiVasta, A. (2011). 63. Journal of Adolescent Health, 48(2), S50.	Cross-sectional	38 female adolescents meeting the DSM criteria for anorexia nervosa with a mean age of 16.5 (SD = 2.2)	The Beck Depression Inventory (BDI) The State-Trait Anxiety Inventory (STAI)	Vegetarians: no consumption of any form of meat (via author correspondence) Vegetarians (n = 17) Non-vegetarians (n = 21)	NR	NR	Vegetarians reported higher levels of anxiety than non-vegetarians. Vegetarians also reported higher levels of depression than non-vegetarians, but NS.	M, p-values Very low quality
Timko, C. A., Hormes, J. M., & Chubski, J. (2012). Appetite, 58(3), 982-990.	Cross-sectional	486 USA university students; **374 females and 111 males between 18 and 40+ years of age with a mean age of 24.94 (SD = 9.54)	The Depression, Anxiety, and Stress Scale-21 (DASS)	Self-report of vegetarian status and the Food Frequency Questionnaire Vegans: excluded animal products (n = 35). Vegetarians: ovo-, lacto- or lacto-ovo vegetarian (n = 111) Semi-vegetarians: consumed no red meat or pork, but occasionally consumed fish or poultry (n = 75) Omnivores: ate meat (n = 265)	NR	NR	NS differences between the groups on depression and anxiety.	M, SD for sex were provided by the authors Moderate quality
Velten, J., Bieda, A., Scholten, S., Wannemüller, A., & Margraf, J. (2018). BMC Public Health, 18(1), 632.	Cross-sectional & Longitudinal	15396 university students from Germany (2991) and China (12,405); 9440 females and 5956 males between 15 and 65 years (M = 21.14, SD = 2.83)	The Depression, Anxiety, and Stress Scale-21 (DASS): - scores range from 0-63	"Do you currently follow a vegetarian diet?" Answers ‘no’, ‘yes (no meat and no fish)’, ‘yes (no meat)’, ‘vegan’ Vegetarian: no meat, no fish, vegan (n = 3216) Non-vegetarians (n = 12180)	Currently	A case-control sample of the Chinese students, matched for age and gender was randomly drawn to control for sample size differences	Vegetarian diet was associated with greater mental health problems (anxiety & depression) than non-vegetarian diet.	M, SD were calculated from the data provided by the first author based on the German sample only Moderate quality

* Discrepancies between the total sample of participants and the diet groups.

** Discrepancies between the total sample of participants and the number of males/females.

Note: NR-not reported; NS-not significant; n-number of participants; SD-standard deviation; SF-short form; BDI-The Beck Depression Inventory; CES-D-The Center for Epidemiologic Studies Depression Scale; DASS-The Depression Anxiety Stress Scale; FFQ-Food Frequency Questionnaire; SPAS-The Social Physique Anxiety Scale; STAI-The State Trait Anxiety Inventory; BMI-Body Mass Index.

Study quality and risk of bias assessment

Each study included in our meta-analysis was assessed for methodologic rigor via quantitative analyses of the study quality. An outside investigator – an expert in obesity, nutrition, and epidemiological science – and the lead investigator worked independently while employing a 100-point scale of predetermined criteria that was developed for our previously published systematic review (Dobersek et al., 2020).

Specifically, each study was evaluated for design, sampling and recruitment, specification and analysis of outcome, and interpretation and communication of results (please see Supplemental File 2). For the interpretation purpose, the studies were categorized as high quality (scores 90–100), moderate quality (scores 60–89), moderate-to-low quality (scores 40–59), low quality (scores 20–39), and very low quality (scores 0–19).

The statistical method of estimation of effect sizes and analyses

The statistical techniques to compute the ESs were adopted from Borenstein and colleagues (Borenstein et al. 2011; Borenstein, Rothstein, and Cohen 2005). We calculated the magnitude of the effect (i.e., effect size) from the d family: ESs or standardized mean differences between meat consumers and meat abstainers on depression and anxiety. Specifically, for study i we computed $$d_i=\frac{({\overline{Y}}_i^A\mathrm{ }-\mathrm{ }{\overline{Y}}_i^M)}{S_i^p},$$ where ${\overline{Y}}_i^A$ is the mean score on depression or anxiety for meat abstainers, ${\overline{Y}}_i^A$ is the analogous mean for meat consumers, and $S_i^p$ is the weighted pooled standard deviation across the two groups. Because both outcomes have negative valences, higher scores reflect more negative outcomes; negative d_i values thus reflect poorer outcomes for meat consumers and more favorable outcomes for meat abstainers. Conversely, positive d_i values reflect better outcomes for meat consumers. If means and standard deviations were not available, we used other statistics (e.g., standard errors, p or t values, proportions) and transformed the data into the d effect size using the appropriate formulas, or used the available data published with the articles, or contacted the authors for the relevant information. If the same type of outcome (e.g., anxiety) was assessed multiple times via different measures in the same study (e.g., DASS-A and POMS-A, SPAS and STAI), we calculated the simple average of the effect sizes and adjusted the variance using the formula below to address the problem of non-independent information (Gleser and Olkin 2009): $$V\left(\overline{d}\right)=\mathrm{ }\frac{1}{4}\mathrm{ }(\sum v_i+\mathrm{ }2Cov(d_i,d_{i^{\mathrm{\text{'}}}}\left)\right)$$

where v_iis the variance of d_i and $$\mathit{Cov}\left(d_i,d_{i^{\mathrm{\text{'}}}}\right)=\left(\frac{1}{n_i^A}+\mathrm{ }\frac{1}{n_i^M}\right)r_{j\mathrm{ }j\mathrm{\text{'}}}+\mathrm{ }\frac{r_{j\mathrm{ }j\mathrm{\text{'}}}{d}_j{d}_{j\mathrm{\text{'}}}}{2(n_i^A+n_i^M)}$$

Additionally, given that Cohen’s d tends to overestimate the effects, we used Hedges’s g effect size (i.e., a quantitative measure of the magnitude between meat abstainers and meat consumers) with bias correction (Hedges 1981; Hedges and Olkin 1985). All ESs were interpreted according to Cohen’s rules of thumb, with an ES of 0.20 indicating a small effect, 0.50 a medium effect, and > 0.80 a large effect (Cohen 1988). Confidence intervals (CIs) were all computed at the 95% level.

We performed a test of heterogeneity using the Cochran’s Q statistic (Hedges and Olkin 1985) to inform us about variation in ESs, and used the I² statistic to examine what proportion of the observed variance reflected differences in true ESs rather than sampling error (Higgins and Thompson 2002; Huedo-Medina et al. 2006). Given that studies included in our meta-analysis were gathered from the published literature and are potentially based on multiple populations, we used a random-effects model as the pooling method (Borenstein 2019; Borenstein et al. 2011; Higgins and Thompson 2004). Publication bias was assessed by inspection of the funnel plot, Egger linear regression test, and Begg rank correlation test (Begg and Mazumdar 1994; Egger et al. 1997). To provide a graphical overview of the ESs for each study, we constructed forest plots.

We examined the differences between meat abstainers or vegans and meat consumers on depression and anxiety separately. To estimate potential differences between the groups, we conducted a subgroup analysis by sex. Additionally, a univariate meta-regression using the restricted maximum likelihood estimator (Dempster, Rubin, and Tsutakawa 1981) for between-studies variation with the Knapp-Hartung adjustment (Hartung and Knapp 2001) was used to explore potential sources of heterogeneity using a study quality as covariate (Higgins and Thompson 2004). Finally, we performed a cumulative meta-analysis to examine the impact of individual studies on the results. Studies were added sequentially to a random-effects model using the continuous study quality score (ordered from highest to the lowest quality). All ESs and analyses were calculated and performed using the Comprehensive Meta-Analysis program Version 3.0. (Borenstein, Rothstein, and Cohen 2005). A p-value less than .05 was considered significant for all tests.

Results

Literature search and characteristics of included studies

Our initial search resulted in 7623 potentially relevant articles. After removing duplicates, we screened titles and abstracts of 7354 papers for inclusion/exclusion criteria. This resulted in 115 full-text articles which were read fully and critically assessed. This qualitative screening resulted in 20 papers published from 2001 to April 2020 that met our inclusion/exclusion criteria. These included 17 cross-sectional, 2 mixed cross-sectional and longitudinal studies, and 1 RCT. The total sample included 171,802 participants (93,257 females and 77,410 males) with 157,778 meat consumers and 13,259 meat abstainers. Three studies included only females (Baines, Powers, and Brown 2007; Lindeman 2002; Stokes, Gordon, and DiVasta 2011) and one study included only males (Hibbeln et al. 2018). Eight studies included samples from the U.S. or North America (Beezhold et al. 2015; Beezhold and Johnston 2012; Beezhold, Johnston, and Daigle 2010; Forestell and Nezlek 2018; Jin et al. 2021; Perry et al. 2001; Stokes, Gordon, and DiVasta 2011; Timko, Hormes, and Chubski 2012), eleven were from non-U.S. countries (e.g., Europe, Asia, and Oceania) (Baines, Powers, and Brown 2007; Baş, Karabudak, and Kiziltan 2005; Goh et al. 2019; Hessler-Kaufmann et al. 2020; Hibbeln et al. 2018; Kapoor et al. 2017; Lindeman 2002; Matta et al. 2018; Michalak, Zhang, and Jacobi 2012; Paslakis et al. 2020; Velten et al. 2018), and one study included samples from multinational-cohorts (Lavallee et al. 2019). The sample sizes ranged from 38 to 90,380 participants between 11 and 105 years of age.

To assess mental health, one study used diagnostic interviews (Michalak, Zhang, and Jacobi 2012) and one used doctors’ diagnoses and prescription medications (Baines, Powers, and Brown 2007). The rest used self-reported questionnaires (e.g., DASS, POMS, SPAS, STAI, etc.) (Baş, Karabudak, and Kiziltan 2005; Beezhold et al. 2015; Beezhold and Johnston 2012; Beezhold, Johnston, and Daigle 2010; Forestell and Nezlek 2018; Goh et al. 2019; Hessler-Kaufmann et al. 2020; Hibbeln et al. 2018; Jin et al. 2021; Kapoor et al. 2017; Lavallee et al. 2019; Lindeman 2002; Matta et al. 2018; Paslakis et al. 2020; Perry et al. 2001; Stokes, Gordon, and DiVasta 2011; Timko, Hormes, and Chubski 2012; Velten et al. 2018). The studies used a variety of self-reported FFQs to assess dietary intake. Table 2 reports details on participants and study characteristics. As per the Preferred Reporting Items for Systematic Review and Meta-Analyses statement (PRISMA) (Moher et al. 2009), results from each stage of the search are displayed in Figure 1.

Figure 1. PRISMA Flow-Chart: Search Result.

Methodologic quality analysis

The studies ranged from very low (7.5) to high quality (100) with a mean score of 52.5. For interpretation purposes, they were placed in five categories based on their methodologic quality score: 3 had very low quality (Beezhold et al. 2015; Beezhold and Johnston 2012; Stokes, Gordon, and DiVasta 2011), 5 had low quality (Beezhold, Johnston, and Daigle 2010; Goh et al. 2019; Hessler-Kaufmann et al. 2020; Jin et al. 2021; Lindeman 2002), 3 had moderate-to-low quality (Forestell and Nezlek 2018; Lavallee et al. 2019; Perry et al. 2001), 6 had moderate quality (Baş, Karabudak, and Kiziltan 2005; Hibbeln et al. 2018; Kapoor et al. 2017; Paslakis et al. 2020; Timko, Hormes, and Chubski 2012; Velten et al. 2018), and 3 had high quality (Baines, Powers, and Brown 2007; Matta et al. 2018; Michalak, Zhang, and Jacobi 2012). Inter-rater correlations between reviewers were very high (ICC = .979; 95% CI: .948-.992, p < .001). Several issues contributed to variation in study quality. Specifically, most studies used cross-sectional designs (n = 19), subjective and/or self-reported dietary (n = 19) and psychological assessment methods (n = 19), biased recruitment (n = 6), non-representative sampling (n = 11), and overinterpretated results (n = 9). For details regarding limitations, please see Table 3 and our systematic review for a discussion on these issues (Dobersek et al., 2020).

Table 3. Major limitations and the number of studies associated with each.

Major limitations		Number of studies
Design & Protocol	Cross-sectional design	19*
	Lack of randomization	1
	Potential biases (social desirability, reactivity, observer-expectancy)	1
	Lack of manipulation check	1
Recruitment & Sampling	Biased recruitment strategies/sample	6
	Non-representative/limited/convenience sample	11
	Low response rate/not reported	4
	Variability in sampling strategies within the study	1
	Oversampling specific groups (e.g., vegans)	1
Assessment: Exposure & Outcome	Self-reported psychological outcome	19
	Self-reported dietary status	19
	Questionnaires not validated for the sample/ no information	3
	Variability in administration of the questionnaires within the study	2
	Limited/lack of description	2
Statistical Assessment	No intra-group comparisons	1
	No control for multiple comparisons/ no information	3
	Groups different at the baseline	1
	Lack of/poor description	3
	Categorizing continuous variables	1
Interpretation & Communication Issues	Inappropriate causal/biased language	4
	Over-/mis-/selective reporting (e.g., failure to recognize regression to the mean, floor effect)	3
	Overinterpretation	9

Note: *Two studies also included longitudinal designs.

Main findings

Depression and anxiety of meat abstainers and meat consumers

The analysis of 19 studies of depression showed a small but significant effect where meat abstainers had higher levels of depression than meat consumers (g = 0.216, 95% CI [0.14, 0.30], p < .001; 95% prediction interval [-0.09, 0.52]). Figure 2 shows the forest plot for the bias-corrected Hedges’s g values and their associated 95% CIs. Heterogeneity was significant and considerable [Q(18) = 74.01, p < .001; I² = 75.68] and the funnel plot (Figure 3), Egger’s linear-regression test (intercept = −0.31, t(17) = 0.41, 95% CI [-1.93, 1.31], p = .362), and Begg’s test (Kendall’s tau = −.094, p = .288) suggested no evidence of asymmetry of effects.

Figure 2. Forest plot for Hedges’s g and its associated 95% confidence interval (CI) for differences in depression between meat abstainers and meat consumers. Studies are arranged from high to low quality score.

Figure 3. Funnel plot of standard error and Hedges’s g for differences in depression between meat abstainers and meat consumers.

The analysis of 11 studies of anxiety showed a small significant effect where meat abstainers had higher levels of anxiety than meat consumers (g = 0.17, 95% CI [0.03, 0.31], p = .02; 95% prediction interval [-0.28, 0.62]). Figure 4 shows the forest plot for each bias-corrected Hedges’s g and its associated 95% CIs. Heterogeneity was again significant [Q(10) = 54.58, p < .001; I² = 81.50] and the funnel plot (Figure 5), Egger’s linear-regression test (intercept = − 0.28, t(9) = 0.26, 95% CI [-2.64, 2.08], p = .40), and Begg’s test (Kendall’s tau < .01, p = .50) suggested no evidence of publication bias.

Figure 4. Forest plot for Hedges’s g and its associated 95% CI for differences in anxiety between meat abstainers and meat consumers. Studies are arranged from high to low quality score.

Figure 5. Funnel plot of standard error and Hedges’s g for differences in anxiety between meat abstainers and meat consumers.

Depression and anxiety of vegans and meat consumers

The analysis of 5 studies showed a small significant effect where vegans had higher depression levels than meat consumers (g = 0.26, 95% CI [0.01, 0.51], p = .041; 95% prediction interval [-0.62, 1.15]). Heterogeneity was significant [Q(4) = 17.02, p = .002, I² = 76.50], and Egger’s linear regression test (intercept = 3.60, t(3) = 1.08, 95% CI [-6.97, 14.16], p = .18), and Begg’s test (Kendall’s tau = 0.30, p = .23) suggested no evidence of publication bias.

Our analysis of 3 studies showed a small non-significant effect between vegans and meat consumers on anxiety (g = 0.15, 95% CI [-0.40, 0.69], p = .598; 95% prediction interval [-6.64, 6.92]). Heterogeneity was again significant [Q(2) = 25.80, p < .01, I² = 92.25], and Egger’s linear-regression test (intercept = 7.93, t(1) = 1.22, 95% CI [-74.72, 90.57], p = .219) and Begg’s test (Kendall’s tau < 0.001, p = .50) showed no evidence of publication bias. However, the small size of this set of studies makes generalization difficult.

Sex differences in contrasts of meat abstainers/vegan versus meat consumers on depression and anxiety

Sex did not modify the sizes of the differences between meat abstainers and meat consumers on depression, F(1, 22) = 1.34, p = .25, or anxiety, F(1, 11) = 0.77, p = .38. Also, no sex differences were detected in the comparisons of vegans versus meat consumers on depression, F(1, 6) = 0.62, p = .431, or anxiety, F(1, 2) = 0.80, p = .372.

Meta-regression and cumulative meta-analyses

A random-effects univariate meta-regression analysis of 19 studies of depression showed that study quality (as a continuous variable) had a significant positive role in between-study heterogeneity, F(1, 17) = 8.46, beta coefficient = 0.005, p = .01, R² = 0.58 (I² = 61.88%, Q(17) = 44.59, p < .001). Similarly, a random-effects univariate meta-regression of 11 studies of anxiety suggested that study quality also had a significant positive impact on between-study heterogeneity, F(1, 9) = 6.68, beta coefficient = 0.006, p = .03, R² = 0.76 (I² = 67.17, Q(9) = 27.37, p = .001). In both cases significant residual variation was unexplained, per high I² values. Please see Figures 6 and 7 for the bubble plots for the relations between study quality, depression, and anxiety.

Figure 6. Bubble plot showing a regression line and prediction interval for study quality on depression between meat consumers and meat abstainers. Bubble size reflects within-study variance.

Figure 7. Bubble plot showing a regression line and prediction interval for study quality on anxiety between meat consumers and meat abstainers. Bubble size reflects within-study variance.

Our cumulative meta-analysis of depression demonstrated that moderate to high quality studies (Timko, Hormes, and Chubski 2012 though Michalak, Zhang, and Jacobi 2012 in Supplemental File 3, Figure 8) showed greater differences whereas moderate-low to very low-quality studies (Perry et al. 2001 through Beezhold and Johnston 2012) showed smaller differences between meat consumers and abstainers. A similar pattern was observed for anxiety, where studies of moderate to high quality (Timko, Hormes, and Chubski 2012 through Michalak, Zhang, and Jacobi 2012) demonstrated greater effect and studies with moderate-low to very low quality (Lavallee et al. 2019 through Beezhold and Johnston 2012) showed a smaller effect between meat consumers and meat abstainers (Supplemental File 3, Figure 9).

Descriptive analyses between effect sizes and study quality

To depict the relations between study quality and effect size, we created two figures (please see Supplemental File 4). It is important to note that for both depression and anxiety, all the higher quality studies favored meat consumers, whereas the relation was equivocal in the lower quality studies. Additionally, high quality studies included greater number of participants and were more homogenous compared to other studies. For details, please see Supplemental File 4.

Discussion

This meta-analysis extends the findings of our prior systematic review (Dobersek et al. 2020) by presenting a quantitative evaluation of the relation between meat consumption/abstention and mental health. It included 171,802 participants aged 11 to 105 years, from varied geographic regions, including Europe, Asia, North America, and Oceania. The findings show a significant association between meat consumption/abstention and depression and anxiety. Specifically, individuals who consumed meat had lower average depression and anxiety levels than meat abstainers. We also showed that vegans experienced greater levels of depression than meat consumers. Sex did not modify these relations. Study quality explained a significant proportion of between-studies heterogeneity and a cumulative meta-analysis confirmed these findings. Specifically, the higher the study quality, the more positive the benefit of meat consumption.

Our results may explain the equivocal nature of prior research. In contrast to our clear findings (both past (Dobersek et al. 2020 and present), other systematic reviews and meta-analytic results were inconsistent or contradictory. These equivocal results suggested that vegetarians, and in some cases vegans had lower levels of depression or anxiety (Askari et al. 2020; Iguacel et al. 2020; Lai et al. 2014; Li et al. 2017; Liu et al. 2016; Nucci et al. 2020; Zhang et al. 2017). As detailed in our systematic review (Dobersek et al. 2020), numerous factors explain these inconsistent conclusions. Briefly, most prior studies employed invalid or unreliable assessment protocols to measure exposures and outcomes (i.e., diet and mental health, respectively). For example, it is well established that dietary recalls and FFQs produce physiologically implausible and non-falsifiable (pseudo-scientific) data (Archer, Pavela, and Lavie 2015; Archer, Hand, and Blair 2013; Archer, Lavie, and Hill 2018a; Archer, Marlow, and Lavie 2018b, 2018c). Thus, the disparity between self-reported and actual dietary intake may render definitive conclusions impossible when analyzing meat consumption as a continuous rather than dichotomous variable (Archer, Pavela, and Lavie 2015; Archer, Hand, and Blair 2013; Archer, Lavie, and Hill 2018a; Archer, Marlow, and Lavie 2018b, 2018c).

With respect to mental health, the most rigorous research relied on physician-diagnosed disorders using the Diagnostic and Statistical Manual of Mental Disorders (DSM) (Michalak, Zhang, and Jacobi 2012) (APA 2013) rather than self-reported (subjective) assessments with untested validity. The use of tools with questionable validity can lead to ambiguous findings and limited cross-study analyses.

Another major design error was the use of biased and selective sampling strategies. Several of the included studies recruited samples from vegan and vegetarian websites, social-networking groups, communities, and restaurants. We surmise this may have substantially biased data collection and may skew self-reported variables and findings if participants with a high degree of emotional or ideological commitment to their dietary behaviors intentionally or unconsciously misreport. An antecedent of this error may be a form of confirmation bias in which the flawed sampling confirms the investigators’ ideology or expectations rather than providing dispassionate data and results.

Finally, statistical and communication errors were ubiquitous. These included the failure to correct for multiple comparisons and the inappropriate use of causal language which can lead to invalid results, interpretations, and conclusions. In summary, given that these errors are widespread in the literature, valid conclusions from previous reviews that failed to examine study quality are not possible.

In the present meta-analysis, these errors taken together are related to significant between-studies variation in effect sizes. Study quality explained 58% and 76% of between-studies heterogeneity in the differences in depression and anxiety, respectively. Furthermore, our analyses (see Figures 2, 4, 6, 7, 10, and 11) demonstrated that higher quality studies showed a more positive and consistent relation between meat consumption and mental health. Higher quality studies had much larger sample sizes.

Finally, limited reporting of participant characteristics prevented an examination of several covariates (e.g., BMI, age of diet adoption/length of diet, clinical history, socioeconomic status, culture) that could potentially contribute to between-studies heterogeneity.

Strengths and limitations

Strengths

This meta-analysis had several strengths. First, our a priori decision to select only studies that provided a clear dichotomy between meat consumers and meat abstainers allowed for a clear and rigorous assessment. While myriad studies have examined vegetarianism along a continuum, these were excluded because the lack of a clear distinction between groups rendered inferences equivocal. This distinction is necessary because self-reported (memory-based) dietary assessments (FFQ) should not be used for quantitative analyses because of their invalidity. Any study that attempts to use FFQs as continuous variables are invalid due to nonquantifiable measurement error (Archer, Lavie, and Hill 2018a; Archer, Marlow, and Lavie 2018b; Archer, Pavela, and Lavie 2015).

Second, we limited our psychological outcomes to the most prevalent and debilitative disorders: depression and anxiety. This allowed a focused yet rigorous analysis and ameliorated the effects of poorly operationalized psychological phenomena such as disordered eating, dietary restraint, orthorexia, and neuroticism. This exclusion helps to avoid potential misclassification and concomitant pathologizing of those who simply wish to avoid specific foods or food groups (e.g., vegans). Finally, with over 170,000 participants from several geographic regions, our meta-analysis allowed for more generalizable and definitive conclusions.

Limitations

Our meta-analysis also had limitations. First, we excluded non-English-language studies. This potentially biased our results in favor of ‘Western’ norms which include meat consumption. For example, we excluded papers published in languages other than English (e.g., Japanese, Hindi). Thus, we may have omitted studies from geographic regions that follow predominantly vegetarian or plant-based dietary patterns.

Second, while our search was clearly defined and comprehensive, our inclusion criteria excluded many publications that provided data on this topic (e.g., see (Anderson et al. 2019; Barthels, Meyer, and Pietrowsky 2018; Burkert et al. 2014; Cooper, Wise, and Mann 1985; Jacka et al. 2012; Larsson et al. 2002; Li et al. 2019; Northstone, Joinson, and Emmett 2018)). Specifically, these papers were excluded because they examined constructs other than depression or anxiety (e.g., orthorexia, restrained eating behavior) or assessed meat consumption as a continuous rather than dichotomous variable. As previously stated, self-reported dietary status and FFQs lead to nonquantifiable measurement error. Nevertheless, we think that our rigorous and highly focused meta-analysis has the potential to provide stronger evidence for the medical, research, and lay communities.

Third, despite the high confidence we place in our finding that meat abstention is linked to a greater prevalence of psychological disorders, study designs precluded inferences of temporality and causality. Specifically, only two of the included studies (Lavallee et al. 2019; Velten et al. 2018) provided information on temporality. Therefore, we were unable to conclusively examine this effect. Given that there are many reasons why people abstain from meat (e.g., ethical, environmental, animal rights-related reasons), this empirical question has not been adequately addressed. However, our previous systematic review (Dobersek et al. 2020) showed conflicting evidence on the temporal relations between meat abstention and depression and anxiety. Also, conclusions on causality require evidence from rigorous RCTs. Since only one low-quality RCT met our inclusion criteria (Beezhold and Johnston 2012), no conclusions regarding causality are supported.

Finally, the results of our meta-analysis are only as valid as the data collected in the included primary studies. Given that most studies used FFQs and self-reported questionnaires, participants may have been misclassified. Merely reporting that one avoids meat is not the equivalent of actual meat abstention (Archer, Pavela, and Lavie 2015; Archer, Hand, and Blair 2013; Archer, Lavie, and Hill 2018a; Archer, Marlow, and Lavie 2018b, 2018c). In fact, self-defined vegetarians and meat abstainers may consume meat (Haddad and Tanzman 2003).

Recommendations for future directions

Future investigators should avoid the most common flaws exhibited in the included studies. First, investigators must acknowledge and address the effects of both researcher and participant biases (e.g., confirmation bias, cognitive dissonance, observer-expectancy effects/reactivity) when employing highly selective or biased samples. Individuals who are highly invested in their dietary behaviors may be predisposed to intentional and non-intentional misreporting.

Second, the use of physician-diagnosed disorders based on criteria from the DSM-V (APA 2013; Michalak, Zhang, and Jacobi 2012) is preferable to self-reported symptoms, and assists in producing more definitive results. Additionally, the severe limitations and pseudo-scientific nature of self-reported dietary data and FFQs (Archer, Pavela, and Lavie 2015; Archer, Hand, and Blair 2013; Archer, Lavie, and Hill 2018a; Archer, Marlow, and Lavie 2018b, 2018c) could be overcome in part with point-of-purchase (barcode) data (Ng and Popkin 2012). However, while these data may be less biased, they are not necessarily an accurate proxy for actual dietary consumption.

Third, the use of more rigorous study designs (e.g., RCTs) is desirable over mere observational investigations. However, it would be extremely difficult to conduct a randomized study of diets with a long enough duration to impact fundamental affective outcomes such as anxiety and depression. Furthermore, detailed participant information regarding behavioral and health-related histories and current lifestyles is essential to valid interpretation and conclusions. Finally, studies should provide complete statistical information that allow for the calculations of effect sizes. More complete reporting would enable meta-analysts to extract both effect measures and study characteristics thus allowing for exploration of potentially important but unanswered questions (e.g., how is time of diet adoption related to mental health?).

Conclusion

The purpose of this meta-analysis was to extend our previous systematic review (Dobersek et al. 2020) and provide quantitative evidence to inform clinicians, policy-makers, and future research. Our results show that meat abstention (vegetarianism or veganism) is clearly associated with poorer mental health, specifically higher levels of both depression and anxiety. Our cumulative analyses suggest that the more rigorous the study, the stronger the relation between meat abstention and mental illness. However, the current body of evidence preludes temporal and causal inferences, and none should be inferred.

Conceptualization: UD; Methodology: all authors; Formal analysis and investigation: all authors; Writing-original draft preparation: UD; Writing-review and editing: all authors; Funding acquisition: UD; Supervision: UD,

Disclosure statement

UD, SA, JA, and GW have previously received funding from the Beef Checkoff, through the National Cattlemen’s Beef Association.

Data availability

Data are available upon request.

Additional information

Funding

This study was funded in part via an unrestricted research grant from the Beef Checkoff, through the National Cattlemen’s Beef Association. The sponsor of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.