Content uploaded by Grethe S Tell
Author content
All content in this area was uploaded by Grethe S Tell on Jun 29, 2018
Content may be subject to copyright.
The Association Between Habitual Diet Quality and the Common Mental
Disorders in Community-Dwelling Adults: The Hordaland Health Study
FELICE N. JACKA,PHD, ARNSTEIN MYKLETUN,PHD, M ICHAEL BERK,PHD, INGVAR BJELLAND,MD,PHD, AND GRETHE S. TELL,PHD
Objective: Recent evidence suggests a role for diet quality in the common mental disorders depression and anxiety. We aimed to
investigate the association between diet quality, dietary patterns, and the common mental disorders in Norwegian adults. Methods: This
cross-sectional study included 5731 population-based men and women aged 46 to 49 and 70 to 74 years. Habitual diet was assessed
using a validated food frequency questionnaire, and mental health was measured using the Hospital Anxiety and Depression Scale.
Results: After adjustments for variables including age, education, income, physical activity, smoking, and alcohol consumption, an
a priori healthy diet quality score was inversely related to depression (odds ratio [OR] = 0.71, 95% confidence interval [CI] = 0.59Y0.84)
and anxiety (OR = 0.77, 95% CI = 0.68Y0.87) in women and to depression (OR = 0.83, 95% CI = 0.70Y0.99) in men. Women scoring
higher on a healthy dietary pattern were less likely to be depressed (OR = 0.68, 95% CI = 0.57Y0.82) or anxious (OR = 0.87, 95% CI =
0.77Y0.98), whereas men were more likely to be anxious (OR = 1.19, 95% CI = 1.03Y1.38). A traditional Norwegian dietary pattern was
also associated with reduced depression in women (OR = 0.77, 95% CI = 0.64Y0.92) and anxiety in men (OR = 0.77, 95% CI =
0.61Y0.96). Awestern-type diet was associated with increased anxiety in men (OR = 1.27, 95% CI = 1.14Y1.42) and women (OR = 1.29,
95% CI = 1.17Y1.43) before final adjustment for energy intake. Conclusions: In this study, those with better quality diets were less
likely to be depressed, whereas a higher intake of processed and unhealthy foods was associated with increased anxiety.
Key words: depression, anxiety, diet, nutrition.
FFQ = food frequency questionnaire; HADS = Hospital Anxiety and
Depression Scale; OR = odds ratio; CI = confidence interval; SD =
standard deviation; Kj = kilojoule; PA = physical activity; SES =
socioeconomic status; MI = myocardial infarction.
INTRODUCTION
The alarming and oft-cited statistic offered by the World
Health Organization is that depression will become the
second most common cause of disability in the world by 2020,
despite the increased energy devoted to recognition and treat-
ment. Indeed, the outcomes of depression have not improved
markedly despite this attention, suggesting that other factors
may be influencing the burden of depression. Diet and nutrition
affect physiological factors that underpin depression, such as
inflammation, oxidative processes, and brain plasticity and
function, and the stress-response system and, as such, may play
a role in the genesis and course of this illness (1). Recent epi-
demiological evidence points to a relationship between diet
quality and depression in both adults (2Y6) and adolescents
(7,8). In our recent study in Australian women, healthy dietary
practices were associated with a reduced likelihood of both
clinically diagnosed depressive and anxiety disorders, whereas
unhealthy dietary habits were associated with an increased
likelihood of major depressive disorder, dysthymia, and psy-
chological symptoms (2). Similarly, both Nanri et al. (5) and
Kuczmarski et al. (6) have reported inverse relationships be-
tween diet quality and self-reported depressive symptoms in
community-dwelling adults. Moreover, prospective data from
the Seguimiento Universidad de Navarra cohort study in Spain
have shown an inverse association between the level of adher-
ence to a Mediterranean dietary pattern and the risk for incident
depression in more than 10,000 adults (3), while the Whitehall
II cohort study reported an increased risk for self-reported de-
pression after 5 years for those adhering more strongly to a
‘‘western’’ diet pattern and a reduced risk for those following a
‘‘whole foods’’ diet pattern (4), with odds ratios (ORs) very
similar to those reported in our Australian study (2).
We aimed to extend our recent findings (2) in a larger sample
of both men and women from Western Norway, in a study that
included assessments of both depression and anxiety. Because
dietary patterns are likely to differ between countries, another
aim was to investigate whether local patterns of dietary con-
sumption were related to mental health outcomes. Furthermore,
the use of the Hospital Anxiety and Depression Scale (HADS)
affords a measure of the common mental disorders that is
specifically exclusive of somatic symptoms, thus enabling an
investigation of the relationship between diet and psychological
symptoms that is not obscured by physical health. We hy-
pothesized that there would be inverse associations between
diet quality, measured by both an a priori diet quality score and
dietary pattern analyses, and the likelihood of both depression
and anxiety in community-dwelling Norwegians.
METHODS
Study Population
The Hordaland Health Study was conducted from 1997 to 1999. A sub-
sample of this study, comprising 9187 participants from four selected com-
munities and born in the years 1925 to 1927 or 1950 to 1951, were invited to
participate in a dietary substudy. Of these, 7074 subjects (77%) agreed to
Psychosomatic Medicine 73:483Y490 (2011) 483
0033-3174/11/7306Y0483
Copyright *2011 by the American Psychosomatic Society
From the Department of Clinical and Biomedical Sciences (F.N.J., M.B.),
The University of Melbourne, Barwon Health, Geelong, Australia; The Nor-
wegian Institute of Public Health (A.M.), Oslo; Departments of Education and
Health Promotion (A.M.), Public Health and Primary Health Care (I.B., G.S.T.),
University of Bergen, Bergen, Norway; Orygen Research Centre (M.B.), The
University of Melbourne; and Mental Health Research Institute (M.B.), Melbourne,
Australia.
Address cor respondence and reprint requests to Felice N. Jacka, BA, PhD,
PgGrad Med Sci, Department of Clinical and Biomedical Sciences, The
University of Melbourne, Barwon Health, PO Box 281, Geelong 3220, Australia.
E-mail: felice@barwonhealth.org.au
The Hordaland Health Study was conducted from 1997 to 1999 as a col-
laboration among the National Health Screening Service, the University of
Bergen in Norway, the University of Oslo in Norway, and local health services
in the Bergen region. F.J. is funded by a National Health and Medical Research
Council Postdoctoral Fellowship. A.M. is funded by the Norwegian Institute of
Public Health.
The funding providers played no role in the design or conduct of the study;
collection, management, analysis, and interpretation of the data; or in prepa-
ration, review, or approval of the article.
Received for publication October 3, 2010; revision received April 14, 2011.
DOI: 10.1097/PSY.0b013e318222831a
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
participate and underwent a brief health examination that included ques-
tionnaires regarding diet, mental health, and other lifestyle and medical para-
meters. In total, 6140 subjects (87%) completed the relevant dietary
assessments. Individuals with very low (G2000 kJ [n= 41]) or very high
(915,000 kJ [n= 140]) estimated daily energy intakes were excluded from the
analyses, as were participants who had not provided mental health data (n=
228). There were missing data on smoking (n= 23) and physical activity (PA,
n= 152), and these were replaced with the mode. The final sample comprised
2477 men and 3254 women. All participants gave written, informed consent to
participate. The study was granted approval by the Western Norway Regional
Committee for Medical Research Ethics and by the Norwegian Data
Inspectorate.
Exposure Variables
Dietary intakes were assessed with a self-administered, optically readable food
frequency questionnaire (FFQ) (9), which has been validated in Norwegian
populations (10) and was designed to obtain information on usual food intake
during the past year. This comprehensive questionnaire comprised questions re-
garding 169 items and included frequency alternatives (from once per month to
several times per day),the number of food units eaten, and theportion sizes. It also
assessedthe number and timing of meals per day. The foods were grouped to reflect
typical Norwegian meals. Using the estimated portion sizes, intakes of each food
were converted to daily equivalents (grams per day) for statistical analyses. Energy
intakes, in kilojoules per day, were also computed from the FFQ data.
A Priori Diet Quality Score
A diet quality score was derived from the FFQ and comprised the summed
tertile rankings of six food groups, chosen a priori, with high scores indicating
a more healthy diet. Food groups included in this score were vegetables, fruit,
low-fat dairy, whole grains, fish, and nonprocessed red meat. With the exception
of red meat, each food group comprised the sum of intakes of foods, categorized
into tertiles, with the highest tertile given the highest score (i.e., 3). The ex-
ception was red meat, wherein the middle tertile was given the highest score.
TABLE 1. The Hordaland Health Study 1997 to 1999: Characteristics of the Study Participants
46Y49 y 70Y74 y
Men (n= 1229) Women (n= 1728) Men (n= 1248) Women (n= 1526)
Diet quality score (range, 6Y18), M (SD) 12.7 (2.3) 12.0 (2.4) 12.4 (2.5) 11.2 (2.7)
Healthy diet factor score, median (IQR) j0.3 (j0.8 to 0.4) 0.2 (j0.4 to 0.9) j0.4 (j0.9 to 0.2) j0.3 (j0.7 to j0.4)
Western diet factor score, M (SD) 1.0 (1.0) 0.2 (0.8) j0.3 (0.70) j0.7 (0.6)
Traditional diet factor score, M (SD) 0.1 (0.9) j0.5 (0.8) 0.7 (1.0) j0.1 (0.9)
HADS-D score, median (IQR) 3 (1Y5) 2 (1Y5) 3 (1Y6) 3 (1Y5)
HADS-A score, median (IQR) 4 (2Y6) 4 (2Y7) 3 (1Y5) 4 (2Y7)
Energy, M (SD), kJ/d 10,017 (2330) 7774 (2048) 8374 (2225) 6533 (2048)
Waist-hip ratio, M (SD) 0.91 (0.06) 0.80 (0.06) 0.95 (0.06) 0.83 (0.07)
Systolic BP, M (SD), mm Hg 130.6 (14.8) 124.4 (15.4) 145.7 (19.4) 147.8 (21.9)
Education, n(%)
Primary 198 (16.1) 390 (22.6) 450 (36.1) 858 (56.2)
Secondary 497 (40.4) 741 (42.9) 508 (40.7) 502 (32.9)
1Y3 y higher 235 (19.1) 290 (16.8) 157 (12.6) 101 (6.6)
Q4 y higher 299 (24.3) 307 (17.8) 133 (10.7) 65 (4.3)
Annual income (NOK), n(%)
0Y99,900 4 (0.3) 31 (1.8) 43 (3.4) 124 (8.1)
100,000Y249,000 1268 (51.2) 620 (35.9) 972 (77.9) 1215 (79.6)
250,000Y499,000 673 (27.2) 561 (32.5) 167 (13.4) 102 (6.7)
Q500,000 489 (19.7) 601 (18.5) 66 (5.3) 85 (5.6)
Nonsmokers, n(%) 841 (68.4) 1139 (65.9) 1049 (84.1) 1312 (86.0)
Smokers, n(%) 388 (31.6) 589 (34.1) 199 (15.9) 214 (14.0)
PA per wk, n(%)
No PA or light PA G1 h/wk 126 (10.3) 157 (9.1) 118 (9.5) 208 (13.6)
Light PA Q1 h/wk 169 (13.8) 389 (22.5) 468 (37.5) 758 (49.7)
Hard PA G2 h/wk 745 (60.6) 975 (56.4) 481 (38.5) 477 (31.3)
Hard PA Q2 h/wk 189 (15.4) 207 (12.0) 181 (14.5) 83 (5.4)
Alcohol consumption, n(%)
Nondrinker 183 (14.9) 477 (27.6) 460 (36.9) 878 (57.5)
1Y15 units/fortnight 889 (72.3) 1200 (69.4) 687 (55.0) 630 (41.3)
Q16 units/fortnight 157 (12.8) 51 (3.0) 101 (8.1) 18 (1.2)
HADS-D Q8, n(%) 118 (9.6) 131 (7.6) 122 (9.8) 150 (9.8)
HADS-A Q8, n(%) 172 (14.0) 341 (19.7) 99 (7.9) 267 (17.5)
M = mean; SD = standard deviation; IQR =interquartile range; HADS-D = Hospital Anxiety and Depression Scale for depression; HADS-A = HADS for anxiety;
BP = blood pressure; NOK = Norwegian kroner; PA = physical activity.
F. N. JACKA et al.
484 Psychosomatic Medicine 73:483Y490 (2011)
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
This was done to reflect Australian national healthy eating recommendation for
the moderate consumption of nonprocessed red meats (11).
Dietary Patterns
To complement the diet quality score, an a posteriori dietary pattern analysis,
using a principal components method with varimax rotation, was undertaken to
identify patterns of dietary intakes derived from the FFQ data. In contrast to an
a priori diet quality score, which imposes a predetermined set of values regarding
what constitutes a healthy diet, an a posteriori dietary pattern analysis allows
an examination of dietary intakes as they exist within the sample data, using
validated and reliable methods (12). The scree plot of eigenvalues indicated that
there were three main dietary factors explaining 19.5% of the variance in dietary
intakes. These three factors were retained as variables in the data set and named as
healthy,western,andtraditional (Norwegian), respectively (13). A healthy dietary
pattern comprised (in order of magnitude of loadings) vegetables, salads, fruits,
rice, pasta, cereals, fish, wine, and nonprocessed meats, whereas a western pattern
comprised foods including meat and liver, processed meats, pizza, salty snacks,
chocolates, sugar and sweets, soft drinks, margarine, mayonnaise and other
dressings, french fries, beer, coffee, cake, and ice cream. A traditional Norwegian
dietary pattern consisted of foods such as fish and shellfish, potatoes, fruits,
vegetables, butter and margarine, milk and yogurt, bread, pasta, rice, meat, meat
spreads, legumes, and eggs.
Outcome: HADS
Symptoms of depression and anxiety were self-reported using the HADS (14).
The HADS is a self-administered questionnaire comprising 14 four-point Likert
scaled items, seven for anxiety (HADS-A subscale) and seven for depression
(HADS-D subscale), encompassing the past week. The HADS is a widely used
instrument specifically designed to identify possible and probable cases of de-
pression and anxiety in epidemiological research and specialist care (15). It has
high internal consistency, with Cronbach >values of 0.80 for anxiety and 0.76
for depression subscales (16). The HADS specifically excludes measures of vege-
tative symptoms and focuses on psychological aspects of depression such as dys-
phoria and anhedonia. In this study, scores on HADS-A and HADS-D were summed
to give a total continuous score for each subscale that had the potential range from
0 to 21. For those participants with fewer than three missing values on the HADS,
scores were replaced with individual means. Case-level depression was subsequently
defined as a HADS-D score of 8 or higher, whereas case-level anxiety was defined as
a HADS-A score of 8 or higher (17).
Potential confounding factors were identified apriori, based on theoretical
considerations and the previous literature. These included total energy intake,
gender, age group (47Y49 or 70Y74 years), income, education, PA, current
smoking, and alcohol use. Income consisted of four categories ranging from 0 to
500,000 Norwegian kroner or more per annum. Education included the categories
of basic schooling, finished high school, 1 to 3 years in college/university, and 4
or more years in university. Leisure-time PA was assessed by a self-report ques-
tionnaire and registered as the number of hours per week, with light activity
meaning no sweating or being out of breath and hard activity meaning sweating
and being out of breath. These data were categorized to comprise four categories
(no PA or light PA, G1 h/wk; light PA, Q1 h/wk;hard PA,G2 h/wk; and hard PA, Q2
h/wk). Current smoking was defined as individuals reporting the smoking of
cigarettes, cigars, or pipes on a daily basis. Alcohol use was measured as units of
alcohol consumed per fortnight and categorized into three levels: no drinking,
drinkingbetween 1 and 15 units per fortnight, and drinking more than 15 units per
fortnight. All covariates were entered as categorical variables, with exception of
energy consumption, whichwas encoded as a continuousvariable. Waist-hip ratio,
blood pressure, self-reported diabetes (yes/no), and myocardial infarction (MI)
(yes/no) were also tested in the models as potential mediators or confounders
(18,19).
Statistical Analysis
Differences in characteristics between those with and without categorical de-
pression or anxiety were tested using ttests or the W
2
test. Correlations between
measures of diet quality and other variables were assessed using Pearson ror
Spearman r
s
statistics. Diet quality scores and the western and traditional factor
scores were normally distributed; however, the healthy dietary pattern was skewed
and therefore transformed using a natural log transformation. All exposure vari-
ables were subsequently standardized as zscores and expressed as standard de-
viation (SD) units. Age group, gender, and PA levels were testedas potential effect
modifiers. Because gender, but not age group or PA, was identified as an effect
modifier in the relationships between both diet quality scores (p=.02)anda
healthy dietary pattern (pG.001), and the mental health outcomes, subsequent
analyses were stratified by gender.
Diet quality scores and dietary factor scores (entered simultaneously) were the
exposure variables of interest. Multivariate logistic regression analyses with 95%
confidence intervals (CIs) were used to assess the relationships of the dietary
exposurevariables to the likelihood of case-level depression and anxiety. For each
regression analysis, the following were entered sequentially following the primary
exposure variables to assess confounding: age, socioeconomic factors (education
and income), health behaviors (PA, alcohol consumption, and smoking), and en-
ergy intake. All variables were retained in analyses on theoretical grounds. Waist-
hip ratio, blood pressure, diabetes, and MI were also tested in the models as
potential mediators or confounders.
RESULTS
Characteristics
The characteristics of study participants by age group and
gender are presented in Table 1.
Diet Quality Score and Depression
The results of logistic regression analyses with diet quality
score as the exposure variable are presented in Table 2. For men,
each 1-SD increase in diet quality score was associated with a
TABLE 2. The Hordaland Health Study 1997 to 1999: Logistic Regression Analyses: A Priori Diet Quality Score
(Standardized zScore) and Case-Level Depression and Anxiety in Men and Women
Men (n= 2477) Women (n= 3254)
A Priori Diet Quality Score OR (95% CI) pOR (95% CI) p
Depression (HADS-D Q8)
Unadjusted 0.83 (0.73Y0.95) .006 0.65 (0.58Y0.74) G.001
Fully adjusted
a
0.83 (0.70Y0.99) .034 0.71 (0.59Y0.84) G.001
Anxiety (HADS-A Q8)
Unadjusted 1.04 (0.91Y1.18) .57 0.86 (0.79Y0.94) .001
Fully adjusted
a
1.00 (0.85Y1.17) .98 0.77 (0.68Y0.87) G.001
a
Adjusted for age, income, education, physical activity, smoking, alcohol consumption, and energy consumption.
OR = odds ratio; CI = confidence interval; HADS-D = Hospital Anxiety and Depression Scale for depression; HADS-A = HADS for anxiety.
DIET QUALITY AND MENTAL HEALTH IN ADULTS
Psychosomatic Medicine 73:483Y490 (2011) 485
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
17% reduction in the OR for case-level depression, both before
and after adjustments for potential confounders including age,
income, education, PA, smoking, alcohol consumption, and en-
ergy intake. For women, each 1-SD increase in diet quality score
was associated with a 35% reduction in the OR for case-level
depression before adjustments and a 29% reduction after
adjustments. The nonparametric correlation between diet quality
scores and continuous depression scores was r
s
=j0.04 (p=.04)
for men and r
s
=j0.09 (pG.001) for women.
Diet Quality Score and Anxiety
There was no association between diet quality score and the
OR for case-level anxiety in the men. In women, each SD
increase in diet quality score was also associated with a 14%
reduction in the OR for case-level anxiety before adjustments
and a 23% reduction after adjustments (Table 2). The non-
parametric correlation between diet quality scores and con-
tinuous anxiety scores was r
s
=0.04(p= .03) for men and
r
s
=j0.04 (p= .01) for women. Further adjustments for waist-
hip ratio, blood pressure, diabetes, and MI did not affect the
described relationships.
Dietary Factor Scores and Depression
In logistic regression analyses with dietary factor scores en-
tered simultaneously as exposure variables against case-level
depression, there were no associations between dietary patterns
anddepressioninmenbeforeadjustments(Table3).After
adjustments for potential confounding variables, each 1-SD
increaseinatraditional dietary factorscorewas associatedwitha
23% reduction in the OR for depression. In women, a healthy
dietary pattern was associated with reduced OR for case-
level depression, both before (43%) and after (32%) adjust-
ments. There were no significant associations detected between
western or tra ditional dietary pa tterns and case-level depression
in women.
Dietary Factor Scores and Anxiety
In analyses with dietary factor scores entered simultaneously
as exposure variables against case-level anxiety, both a healthy
TABLE 3. The Hordaland Health Study 1997 to 1999: Logistic Regression Analyses: Dietary Patterns
(Standardized zScore) and Case-Level Depression
Men (n= 2477) Women (n= 3254)
Dietary Pattern Case-Level Depression,
a
OR (95% CI) pCase-Level Depression,
a
OR (95% CI) p
Unadjusted model
Healthy 0.94 (0.82Y1.08) .27 0.57 (0.49Y0.66) G.001
Western 1.03 (0.91Y1.16) .81 1.03 (0.90Y1.20) .54
Traditional 0.96 (0.74Y1.09) .66 0.90 (0.77Y1.05) .09
Fully adjusted model
b
Healthy 1.02 (0.87Y1.19) .92 0.68 (0.57Y0.82) G.001
Western 0.87 (0.68Y1.11) .25 1.25 (0.93Y1.68) .27
Traditional 0.77 (0.61Y0.96) .02 0.99 (0.76Y1.29) .51
a
Case-level depression was defined as a Hospital Anxiety and Depression Scale for depression score of 8 or higher.
b
Adjusted for age, income, education, physical activity, smoking, alcohol consumption, and energy consumption.
OR = odds ratio; CI = confidence interval.
TABLE 4. The Hordaland Health Study 1997 to 1999: Logistic Regression Analyses:
Dietary Patterns (Standardized zScore) and Case-Level Anxiety
Men (n= 2477) Women (n= 3254)
Dietary Pattern Case-Level Anxiety,
a
OR (95% CI) pCase-Level Anxiety,
a
OR (95% CI) p
Unadjusted model
Healthy 1.14 (1.01Y1.29) .04 0.84 (0.77Y0.93) .001
Western 1.27 (1.14Y1.42) G.001 1.29 (1.17Y1.43) G.001
Traditional 0.92 (0.81Y1.05) .29 0.88 (0.79Y0.97) .007
Fully adjusted model
b
Healthy 1.19 (1.03Y1.38) .02 0.87 (0.77Y0.98) .01
Western 1.09 (0.87Y1.36) .52 1.18 (0.96Y1.45) .36
Traditional 0.86 (0.70Y1.05) .18 0.77 (0.64Y0.92) .001
a
Case-level anxiety was defined as a Hospital Anxiety and Depression Scale for anxiety score of 8 or higher.
b
Adjusted for age, income, education, physical activity, smoking, alcohol consumption, and energy consumption.
OR = odds ratio; CI = confidence interval.
F. N. JACKA et al.
486 Psychosomatic Medicine 73:483Y490 (2011)
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
and western dietary pattern were associated with increased
OR for anxiety in men, before adjustments (Table 4). The positive
association between a western pattern score and anxiety per-
sisted after adjustments for age, income, education, and health
behaviors (OR = 1.19, 95% CI = 1.03Y1.37) but was atten-
uated by final adjustment for energy intake. However, the positive
association of a healthy dietary factor score and anxiety per-
sisted after all adjustments; each 1-SD increase in healthy factor
score is being associated with a 19% increase in the OR for
anxiety, although this association was apparent only in the highest
quintile of healthy diet score. There was no association between a
traditional Norwegian dietary pattern and anxiety in men.
In women, each dietary pattern was associated with case-
level anxiety before adjustments: both a healthy and a tra-
ditional pattern demonstrated an inverse association, and a
western pattern demonstrated a positive association (Table 4).
After adjustments, each 1-SD increase in healthy factor score
was associated with a 16% reduction in the OR for anxiety,
while each 1-SD increase in a traditional factor score was
associated with a 23% reduction in the OR for anxiety. After
adjustments for age, income, education, and health behaviors,
a western dietary pattern was associated with an increase in
the OR for anxiety (OR = 1.32, 95% CI = 1.17Y1.48), but
again, this relationship was attenuated by final adjustments
for energy intake. Further adjustments for waist-hip ratio,
blood pressure, diabetes, and MI did not materially affect the
described relationships.
DISCUSSION
In this large, population-based study of diet quality and the
common mental disorders, a healthier diet, measured with an
a priori diet quality score, was associated with a reduction in
the likelihood of both depression and anxiety in women and
with a reduced likelihood for depression in men. However,
although a healthy dietary pattern, measured by factor analysis,
was also associated with reduced depression and anxiety in
women, such a dietary pattern was associated with increased
anxiety and showed no associations with depression in men.
On the other hand, men scoring higher on a traditional Norwegian
dietary pattern, with higher intakes of fish, fruit, vegetables, and
dairy products, were less likely to be depressed; women scoring
higher on this factor were less likely to be anxious. A western
dietary pattern was associated with increased anxiety, but not
depression, in both men and women, although this associa-
tion was attenuated once overall energy intake was taken into
account.
These results are largely concordant with our previous re-
search demonstrating inverse associations between measures of
diet quality and the common mental disorders in women (2)
and adolescents (8). In our previous study, conducted in a
randomly selected, population-based sample of 1046 adult
women, a dietary pattern comprising vegetables, fruit, beef,
lamb, fish, and whole-grain foods was associated with a re-
duced likelihood of clinically diagnosed depressive and anx-
iety disorders, whereas a dietary pattern comprising processed
and ‘‘unhealthy’’ foods (western) was associated with an in-
creased likelihood of psychological symptoms, as well as ma-
jor depressive disorder and dysthymia before adjustments for
energy intake. Increased a priori diet quality scores were also
associated with reduced psychological symptoms (2). In an-
other study, both a lower adherence to the consumption of
foods comprising a healthy diet, and an increased consumption
of unhealthy and processed foods, were associated with in-
creased odds for self-reported symptomatic depression in more
than 7000 young adolescents. These relationships demonstrated
dose-response patterns and remained robust after adjustment for
a wide range of potential confounding factors (8). These results
also support our previous finding of inverse associations be-
tween dietary magnesium and depressive symptoms in the same
cohort of Norwegian adults (20). Magnesium is found in many
of the foods that comprise a diet promoted as healthy, such as
whole grains, legumes, and green leafy vegetables; and mag-
nesium intake was highly correlated with the diet quality score
in the current study (data not shown), suggesting that magne-
sium intake is a good proxy for a healthy diet.
Where these results diverged noticeably from our previous
findings was in the positive relationship observed between a
healthy dietary pattern and anxiety in men. The increase in the
likelihood of anxiety was seen in those in the top quintile of
healthy diet score, with no relationship observed between anxiety
and the other four quintiles of a healthy diet score. There may be
two possible explanations for this finding. The first is that of a
type I error; with analyses conducted separately for each gender,
with several different exposure variables and two outcome vari-
ables, the risk of statistically spurious findings is increased. An-
other explanation relates to reverse causality. It is plausible to
hypothesize that increased anxiety may be related to an increased
adherence to dietary practices designed to protect against detri-
mental health outcomes; in another recent large-scale study
conducted in Norway, case-level anxiety, when comorbid with
case-level depression, was related to a reduction in mortality (21).
This was in contrast to the noxious effects of depression on
mortality. This finding is consistent with the contention that
anxiety may protect against detrimental health outcomes by en-
couraging increased health awareness and practices.
Another divergence relates to the association between high
scores on a western dietary pattern and mental health outcomes.
In our previous Australian study, we reported a positive associ-
ation between such a dietary pattern and both psychological
symptoms and depressive disorders, but not anxiety (2). In this
study, the relationship was seen with anxiety, but not depression.
It is not clear as to why this might be so; however, the use of the
HADS to measure self-reported depression and anxiety, rather
than a clinical interview measuring major depression, dysthymia,
and clinical anxiety disorders, such as previously used, may ex-
plain these differing associations. Concordant with this study,
however, we had previously noted an attenuation of the signifi-
cant positive relationships between a western dietary pattern
and clinical depressive disorders once overall energy intake was
taken into account. Theoretically, it is the absolute amount of
unhealthy food consumed that is of most relevance to mental
DIET QUALITY AND MENTAL HEALTH IN ADULTS
Psychosomatic Medicine 73:483Y490 (2011) 487
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
health, rather than the quantity as a proportion of overall diet,
because of its impact on important brain proteins (22). The sec-
ond possibility may relate to statistical considerations: variables
sharing a correlation coefficient greater than 0.6 have been shown
to result in unstable and potentially unreliable outcomes when
entered together in logistic regression analyses (23), and the
western dietary pattern exceeded this degree of correlation with
energy intake in both the previous and current study.
Why differential associations were observed for women and
men in our study is not clear. Previous studies focusing on the
dietary consumption of U-3 fatty acids and/or fish and depression
commonly report stronger associations in women than in men
(e.g., Timonen et al. [24]) and it is plausible to postulate that
response biases in dietary reporting may account for some of
these differences. There may also be biologic explanations for
these gender differences, wherein in gonadal hormones result in
differential responses to possible mediating factors such as oxi-
dative stress and inflammation (25).
Strengths and Limitations
The study had several important strengths, including the
large and representative sample comprising both genders and
two different age groups, the use of a comprehensive dietary
assessment designed for the Norwegian population, the inclu-
sion of a wide range of likely confounding variables, and the
testing of medical conditions and parameters as possible medi-
ating factors. Diet quality was also operationalized using both
apriorimeasures (diet quality score) and naturalistic observa-
tions (dietary pattern analysis).
The obvious limitation in this study is the cross-sectional
study design. This precludes the possibility of determining cause
and effect relationships. Reverse causality, wherein dietary intakes
are a function of mood state, is one possible explanation for our
findings. The two recent well-powered and conducted studies
undertaken in both the Seguimiento Universidad de Navarra
cohort in Spain and the Whitehall II cohort did not support re-
verse causality as an explanation for the relationships observed
between diet and the risk for incident depression (3,4); how-
ever, longitudinal studies that explicitly examine this issue are
required to rule out this explanation for the demonstrated rela-
tionships. A recent Australian study of depression in primary
care settings indicated that nearly 30% of individuals with de-
pressive symptoms voluntarily improved their dietary quality in
an attempt to improve symptoms (personal e-mail conversation
with Professor Jane Gunn on July 14, 2010). At the same time,
profound appetite changes, including both increased and de-
creased appetite, are a common feature of major depressive epi-
sodes. Such differential changes in diet as a consequence of
mental health symptoms increase the difficulty in adequately
accounting for such relationships.
Another potential limitation relates to measurement error,
which is a recognized feature of any measures used to assess
diet. Food frequency questionnaires are known to be associ-
ated with substantial measurement error, and it is recognized
that these errors can seriously attenuate the power of nutri-
tional epidemiological studies to estimate the association of
diet with disease outcomes (26). However, dietary measure-
ment error associated with FFQs is known to lead to an un-
derestimation, rather than an overestimation, of the true
association with measured outcomes, therefore potentially ob-
scuring the true relationship between diet and disease (26).
Statistical techniques that reduce the impact of measurement
error, such as structural equation modeling, may be useful in
quantifying the relationships between diet and mental health in
future research.
Confounding
A critical consideration in this nascent field of research is
the possibility of residual and/or unmeasured confounding. Diet
is influenced by socioeconomic status (SES) and educational
status, with those higher on these factors reporting higher diet
quality diets. Although we used measures of SES, education, and
other health behaviors as covariates in this study, there is the
possibility that we did not adequately account for these important
variables and that the results are due to residual or unmeasured
confounding. There may also have been a bias in reporting,
whereby those with higher SES and/or educational levels were
more likely to report a healthier diet. In this study and in our
previous studies (2,8), measures of SES and education had little
impact on the reported associations. Similarly, in previous pro-
spective studies (3,4), the associations between diet quality and
incident depression were not confounded by these factors. We
were also careful to assess waist-hip ratio, diabetes, and blood
pressure as both potential confounders and variables on the causal
pathway (18,19) and found no impact of these factors on the
relationships of interest. However, residual confounding cannot
be ruled out as an explanation for all of these results. Bias may
have also resulted from differential dietary reporting by those
with depressive symptoms. Moreover, there may be unmeasured
factors, such as maternal depression and/or personality factors
that account for these reported relationships. Future studies
should include these potential explanatory variables in addition to
stringent measures of SES.
Mechanisms
Depressive illness is underpinned by several biologic factors:
immune system dysfunction (27), associated oxidative stress
(28), the stress-response system (29), biochemistry (30), and
genetics (31). Diet modulates each of these factors (1), particu-
larly immune and oxidative processes (e.g., [32,33]) and bio-
chemical parameters (22), and these pathways may be mediators
in the relationship between diet quality and depressive illnesses.
Although the relationship between dietary quality and anxiety is
less amenable to mechanistic explanations, the impact of diet on
gene expression (34) and the stress-response system (35,36),
along with the well-documented overlap and comorbidity be-
tween depression and anxiety, all offer potential explanations
for the demonstrated relationships.
CONCLUSIONS
We have demonstrated cross-sectional relationships between
measures of healthy and unhealthy diets, and the common mental
F. N. JACKA et al.
488 Psychosomatic Medicine 73:483Y490 (2011)
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
disorders in middle-aged and elderly men and women. These
relationships could not be fully explained by age group, gender,
income, education, body habitus, medical parameters, or life-
style behaviors other than diet; however, reverse causality and/or
residual or unmeasured confounding cannot be ruled out as
explanations for these findings. Nevertheless, these results sup-
port findings from recent studies in adults (2Y4) and adolescents
(8,7), are largely consistent with the hypothesis that poor qual-
ity diet may be a risk factor for the common mental illnesses,
and support the potential clinical and public health importance
(37) of the emerging field of behavioral nutriomics.
Conventional therapeutic approaches focus on biologic and
psychological aspects of depression and largely ignore the ac-
cumulating body of evidence that lifestyle components play a
role as predisposing and perpetuating factors. The evidence base
regarding exercise as a treatment strategy for depression is well
developed (38); however, there is no equivalent evidence base
for dietary modification to date. As a result, lifestyle modifica-
tion is not routinely incorporated as a part of the management
of depression and therefore fails to address these potentially
reversible factors. Studies examining the impact of dietary im-
provement on existing mental illness are now urgently warranted,
as well as quality longitudinal studies confirming the role of
diet quality in the genesis of the common mental disorders.
Such research may afford the development of both prevention
and treatment strategies for the common mental disorders in-
corporating lifestyle improvements.
REFERENCES
1. Jacka FN, Berk M. Food for thought. Acta Neuropsychiatr 2007;19:
321Y3.
2. Jacka FN, Pasco JA, Mykletun A, Williams LJ, Hodge AM, O’Reilly SL,
Nicholson GC, Kotowicz MA, Berk M. Association between western and
traditional diets and depression and anxiety in women. Am J Psychiatry
2010;167:305Y11.
3. Sanchez-Villegas A, Delgado-Rodriguez M,Alonso A, Schlatter J, Lahortiga
F, Majem LS, Martinez-Gonzalez MA. Association of the Mediterranean
dietary pattern with the incidence of depression: the Seguimiento Uni-
versidad de Navarra/University of Navarra follow-up (SUN) cohort. Arch
Gen Psychiatry 2009;66:1090Y8.
4. Akbaraly TN, Brunner EJ, Ferrie JE, Marmot MG, Kivimaki M, Singh-
Manoux A. Dietary pattern and depressive symptoms in middle age. Br J
Psychiatry 2009;195:408Y13.
5. Nanri A, Kimura Y, Matsushita Y, Ohta M, Sato M, Mishima N, Sasaki S,
Mizoue T. Dietary patterns and depressive symptoms among Japanese men
and women. Eur J Clin Nutr 2010;64:832Y9.
6. Kuczmarski MF, Cremer Sees A, Hotchkiss L, Cotugna N, Evans MK,
Zonderman AB. Higher Healthy Eating Index-2005 scores associated with
reduced symptoms of depression in an urban population: findings from the
Healthy Aging in Neighborhoods of Diversity Across the Life Span
(HANDLS) Study. J Am Diet Assoc 2010;110:383Y9.
7. Oddy WH, Robinson M, Ambrosini GL, TA OS, de Klerk NH, Beilin LJ,
Silburn SR, Zubrick SR, Stanley FJ. The association between dietary
patterns and mental health in early adolescence. Prev Med 2009;49:
39Y44.
8. Jacka FN, Kremer PJ, Leslie E, Berk M, Patton G, Toumbourou JW,
Williams JW. Associations between diet quality and depressed mood in
adolescents: results from the Healthy Neighbourhoods Study. Aust N Z J
Psychiatry 2010;44:435Y42.
9. Konstantinova SV, Vollset SE, Berstad P, Ueland PM, Drevon CA, Refsum
H, Tell GS. Dietary predictors of plasma total homocysteine in the Hor-
daland Homocysteine Study. Br J Nutr 2007;98:201Y10.
10. Nes M, Frost Andersen L, Solvoll K, Sandstad B, Hustvedt BE, Lovo A,
Drevon CA. Accuracy of a quantitative food frequency questionnaire ap-
plied in elderly Norwegian women. Eur J Clin Nutr 1992;46:809Y21.
11. National Health and Medical Research Council (NHMRC). Dietary
Guidelines for Australian Adults. A Guide to Healthy Eating. Common-
wealth of Australia, National Health and Medical Research Council.
Canberra: Australian Government Publishing Service; 2003.
12. Hu FB, Rimm E, Smith-Warner SA, Feskanich D, Stampfer MJ, Ascherio
A, Sampson L, Willett WC. Reproducibility and validity of dietary pat-
terns assessed with a food-frequency questionnaire. Am J Clin Nutr
1999;69:243Y9.
13. Konstantinova SV, Tell GS, Vollset SE, Ulvik A, Drevon CA, Ueland PM.
Dietary patterns, food groups, and nutrients as predictors of plasma choline
and betaine in middle-aged and elderly men and women. Am J Clin Nutr
2008;88:1663Y9.
14. Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta
Psychiatr Scand 1983;67:361Y70.
15. Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the
Hospital Anxiety and Depression Scale. An updated literature review.
J Psychosom Res 2002;52:69Y77.
16. Mykletun A, Stordal E, Dahl AA. Hospital Anxiety and Depression (HAD)
scale: factor structure, item analyses and internal consistency in a large
population. Br J Psychiatry 2001;179:540Y4.
17. Mykletun A, Bjerkeset O, Dewey M, Prince M, Overland S, Stewart R.
Anxiety, depression, and cause-specific mortality: the HUNT study. Psy-
chosom Med 2007;69:323Y31.
18. Rivenes AC, Harvey SB, Mykletun A. The relationship between abdominal
fat, obesity, and common mental disorders: results from the HUNT study.
J Psychosom Res 2009;66:269Y75.
19. Engum A, Mykletun A, Midthjell K, Holen A, Dahl AA. Depression and
diabetes: a large population-based study of sociodemographic, lifestyle,
and clinical factors associated with depression in type 1 and type 2 dia-
betes. Diabetes Care 2005;28:1904Y9.
20. Jacka FN, Overland S, Stewart R, Tell GS, Bjelland I, Mykletun A. Asso-
ciation between magnesium intake and depression and anxiety in commu-
nity-dwelling adults: the Hordaland Health Study. Aust N Z J Psychiatry
2009;43:45Y52.
21. Mykletun A, Bjerkeset O, Overland S, Prince M, Dewey M, Stewart R.
Levels of anxiety and depression as predictors of mortality: the HUNT
study. Br J Psychiatry 2009;195:118Y25.
22. Molteni R, Barnard RJ, Ying Z, Roberts CK, Gomez-Pinilla F. A high-
fat, refined sugar diet reduces hippocampal brain-derived neurotrophic
factor, neuronal plasticity, and learning. Neuroscience 2002;112:803Y14.
23. McGee D, Reed D, Yano K. The results of logistic analyses when the
variables are highly correlated: an empirical example using diet and CHD
incidence. J Chronic Dis 1984;37:713Y9.
24. Timonen M, Horrobin D, Jokelainen J, Laitinen J, Herva A, Rasanen P.
Fish consumption and depression: the Northern Finland 1966 birth cohort
study. J Affect Disord 2004;82:447Y52.
25. Kher A, Wang M, Tsai BM, Pitcher JM, Greenbaum ES, Nagy RD, Patel
KM, Wairiuko GM, Markel TA, Meldrum DR. Sex differences in the
myocardial inflammatory response to acute injury. Shock 2005;23:
1Y10.
26. Kipnis V, Midthune D, Freedman L, Bingham S, Day NE, Riboli E, Ferrari
P, Carroll RJ. Bias in dietary-report instruments and its implications for
nutritional epidemiology. Public Health Nutr 2002;5:915Y23.
27. Pasco JA, Nicholson GC, Williams LJ, Jacka FN, Henry MJ, Kotowicz
MA, Schneider HG, Leonard BE, Berk M. Association of high-sensitivity
C-reactive protein with de novo major depression. Br J Psychiatry
2010;197:372Y7.
28. Ng F, Berk M, Dean O, Bush AI. Oxidative stress in psychiatric disorders:
evidence base and therapeutic implications. Int J Neuropsychopharmacol
2008;11:851Y76.
29. Leonard BE. The HPA and immune axes in stress: the involvement of the
serotonergic system. Eur Psychiatry 2005;20(suppl 3):S302Y6.
30. Shimizu E, Hashimoto K, Okamura N, Koike K, Komatsu N, Kumakiri C,
Nakazato M, Watanabe H, Shinoda N, Okada S, Iyo M. Alterations of
serum levels of brain-derived neurotrophic factor (BDNF) in depressed
patients with or without antidepressants. Biol Psychiatry 2003;54:70Y5.
DIET QUALITY AND MENTAL HEALTH IN ADULTS
Psychosomatic Medicine 73:483Y490 (2011) 489
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
31. Sullivan PF, Neale MC, Kendler KS. Genetic epidemiology of major de-
pression: review and meta-analysis. Am J Psychiatry 2000;157:1552Y62.
32. Esposito K, Nappo F, Marfella R, Giugliano G, Giugliano F, Ciotola M,
Quagliaro L, Ceriello A, Giugliano D. Inflammatory cytokine concentra-
tions are acutely increased by hyperglycemia in humans: role of oxidative
stress. Circulation 2002;106:2067Y72.
33. Esposito K, Ciotola M, Giugliano D. Mediterranean diet, endothelial func-
tion and vascular inflammatory markers. Public Health Nutr 2006;9:1073Y6.
34. Kornman KS. Interleukin 1 genetics, inflammatory mechanisms, and
nutrigenetic opportunities to modulate diseases of aging. Am J Clin Nutr
2006;83:475SY83S.
35. Kamara K, Eskay R, Castonguay T. High-fat diets and stress responsivity.
Physiol Behav 1998;64:1Y6.
36. Hamazaki K, Itomura M, Huan M, Nishizawa H, Sawazaki S, Tanouchi M,
Watanabe S, Hamazaki T, Terasawa K, Yazawa K. Effect of omega-3 fatty
acid-containing phospholipids on blood catecholamine concentrations in
healthy volunteers: a randomized, placebo-controlled, double-blind trial.
Nutrition 2005;21:705Y10.
37. Freeman MP. Nutrition and psychiatry. Am J Psychiatry 2010;167:244Y7.
38. Stathopoulou G, Powers M, Berry A, Smits J. Exercise interventions for
mental health: a quantitative and qualitative review. Clin Psychol Sci Prac
2006;13:179Y93.
F. N. JACKA et al.
490 Psychosomatic Medicine 73:483Y490 (2011)
Copyright © 2011 by the American Psychosomatic Society. Unauthorized reproduction of this article is prohibited.
