International variations in the outcome of schizophrenia and the

prevalence of depression in relation to national dietary practices:

an ecological analysis
Malcolm Peet
BJP 2004, 184:404-408.
Access the most recent version at DOI: 10.1192/bjp.184.5.404

References This article cites 0 articles, 0 of which you can access for free at:
http://bjp.rcpsych.org/content/184/5/404#BIBL
Reprints/ To obtain reprints or permission to reproduce material from this paper, please
permissions write to [email protected]

You can respond http://bjp.rcpsych.org/letters/submit/bjprcpsych;184/5/404

to this article at
Downloaded http://bjp.rcpsych.org/ on May 13, 2014
from Published by The Royal College of Psychiatrists

To subscribe to The British Journal of Psychiatry go to:

http://bjp.rcpsych.org/site/subscriptions/
B R I T I S H J O U R N A L O F P S YC H I AT RY ( 2 0 0 4 ) , 1 8 4 , 4 0 4 ^ 4 0 8

International variations in the outcome total outcome score is a composite score

that takes into account all outcome
measures in the IPSS study. In the
of schizophrenia and the prevalence of depression DOSMED study, data on percentage of
patients never hospitalised and percentage
in relation to national dietary practices: of patients with little social impairment
were used for analysis. Where data for
an ecological analysis{ urban and rural populations were given
separately, the urban data were used for
MALCOLM PEET analysis. In addition, a ‘total best outcome’
score was calculated from all the outcome
measures listed in Table 4.10 from
Background Dietary variations are Some physical illnesses, particularly Jablensky et al (1992). This was calculated
known to predictthe prevalence of diabetes and coronary heart disease, occur by adding together the percentages of
with increased frequency in patients with patients with ‘best possible’ outcomes as
physical illnesses such as diabetes and
schizophrenia and major depression (Peet defined by Jablensky et al (1992), namely:
heart disease butthe possible influence of & Edwards, 1997; Ryan & Thakore, remitting course with full remissions; in
diet on mental health has been neglected. 2002). These physical illnesses share some psychotic episodes, 1–5% of follow-up
epidemiological features with mental dis- period; in complete remission, 76–100%
Aims To explore dietary predictors of orders. Thus, the outcome of schizophrenia of follow-up period; on no antipsychotic
the outcome of schizophrenia and the is generally better in developing countries medication throughout follow-up; never
prevalence of depression. than in more ‘developed’ nations (Hopper hospitalised; and unimpaired social func-
& Wanderling, 2000). Similarly, diabetes tioning for 76–100% of follow-up period.
Method Ecological analysis of national and heart disease are regarded as diseases The use of this composite outcome score
dietary patterns in relation to international of Western industrialised nations and are ensured that all outcome variables were
less common in developing countries taken into account. Data on international
variations in outcome of schizophrenia and
(Tucker & Buranapin, 2001). It is well variations in the prevalence of depression
prevalence of depression. recognised that the pattern of food intake in the population were taken from the study
is of substantial importance in the aetiology of Weissman et al (1996), plus data from
Results A higher national dietary
of diabetes and heart disease. In view of the Japan as used in the analysis by Hibbeln
intake of refined sugar and dairy products clinical and epidemiological association (1998).
predicted a worse 2-year outcome of between these mental and physical illnesses, The countries included in the analysis
schizophrenia. A high national prevalence it is surprising that there has been little were: Denmark, India, Colombia, Nigeria,
of depression was predicted by a low research interest in the relationship between UK, USSR, USA and Czechoslovakia for
nutrition and mental illness. The present the IPSS study; Denmark, India, Colombia,
dietary intake of fish and seafood.
study is an ecological analysis of inter- Ireland, USA, Nigeria, USSR, Japan, UK
national variations in food supply in and Czechoslovakia for the DOSMED
Conclusions The dietary predictors of
relation to epidemiological data on the study; and New Zealand, Canada, Germany,
outcome of schizophrenia and prevalence outcome of schizophrenia and on the France, USA, USSR, Taiwan and Japan for
of depression are similar to those that prevalence of depression. the depression study. No data on social
predict illnesses such as coronary heart outcome were available for Japan. Data
disease and diabetes, which are more on food usage were taken from the
METHOD FAOSTAT database (Food and Agriculture
common in people with mental health
Organization, 2002), which records the
problems and in which nutritional Existing databases on international varia- apparent national food consumption. This
approaches are widely recommended. tions in the outcome of schizophrenia, the is estimated from the total domestic pro-
Dietary intervention studies are indicated prevalence of depression and patterns of duction of food (including non-commercial
in schizophrenia and depression. food usage were identified. For the out- production and production from kitchen
come of schizophrenia, 2-year outcome gardens) plus imports, minus exports,
Declaration of interest M.P. has data from the International Pilot Study of taking into account changes in stocks such
Schizophrenia (IPSS; World Health as stored grain and minus the food lost to
received research funding from Laxdale
Organization, 1979) and the Determinants waste during commercial processing. The
Ltd, a company which is developing ethyl- of Outcome of Severe Mental Disorders following foodstuffs were included in the
eicosapentaenoic acid as a treatment for (DOSMED) study (Jablensky et al, al, 1992) analysis: meat (including beef, mutton,
psychiatric and neurological disorders. were used. From the IPSS study, data on goat, pig and poultry); fish and seafood;
mean days out of hospital and percentage eggs; dairy products (milk, cheese, butter
of patients with severe social impairment and ghee); vegetable oil, vegetables; cereals
were used as outcome variables, as well as (excluding beer); fruits (excluding wine);
a total outcome score, as described by starchy roots; refined sugar; pulses; nuts;
{
See editorial, pp. 381^382, this issue. Christensen & Christensen (1988). This coffee; and alcoholic beverages. All were

404
 DIE T AN
NDD P S YC H I AT R I C I L L NE
NE S S

expressed as supply in kilograms per capita correlation with sugar consumption was study, in which increased intake of dairy
per year. Food consumption data were seen both for outcome measures based on products predicted fewer days spent out of
taken for the year that approximated to hospital admission and those based on so- hospital. Also, alcohol consumption was
the time when the clinical studies were cial outcome. Other correlations that were found to be an additional weak but signifi-
conducted (1970 for the IPSS study, 1980 found in both schizophrenia data-sets but cant positive predictor of overall ‘total best
for the DOSMED study and 1990 for the not necessarily for both admission and outcome’ in the DOSMED study, although
depression prevalence study). It was not social outcome measures included the this accounted for only a very small part of
considered necessary to analyse more than consumption of meat and eggs (adverse the variance relative to sugar. All of these
one year because national food consump- relationship) and the consumption of pulses associations predicted the majority of
tion patterns change slowly over decades (beneficial relationship). Dairy products variance in the outcome of schizophrenia.
and there is little variation from year to and alcohol consumption were associated The sole independent predictor of
year. with a poor outcome in the IPSS study but depression prevalence emerging from the
Each foodstuff group was correlated not in the DOSMED database. With regard multiple regression analysis was the
with each of the clinical databases. Stepwise to depression, the strongest association was consumption of fish and seafood, which
multiple regression was then carried out between a high dietary intake of fish and predicted the prevalence of depression with
using the Statistical Package for the Social seafood and reduced prevalence of depres- an adjusted R2 of 0.74.
Sciences for Windows, Version 11.5, with sion. A high intake of dairy products and
the clinical measures as dependent variables sugar was associated with an increased
DISCUSSION
and dietary measures as independent prevalence of depression, whereas a high
variables. intake of starchy roots was associated with The results of this analysis allow the
a reduced prevalence of depression. hypothesis that the outcome of schizo-
The results of the multiple regression phrenia and prevalence of depression are
RESULTS analysis are shown in Table 2. The most influenced by dietary factors.
striking relationship in the schizophrenia
The results of the multiple (Pearson) databases is between sugar consumption
correlations are shown in Table 1. The and outcome. This was an independent Diet and outcome of schizophrenia
most consistent finding is that a greater predictor of poor outcome of schizophrenia The finding that the outcome of schizo-
consumption of refined sugar is associated as judged by both social and hospital phrenia is better in developing than in
with a worse outcome of schizophrenia admission criteria, as well as the total out- developed countries has never been satis-
and a greater prevalence of depression. In come score. An exception to this was the factorily explained and does not appear
the schizophrenia data-sets, a high hospital admission data from the IPSS related simply to confounding factors such

T
Table
able 1 Pearson correlation coefficients between both the schizophrenia 2-year outcome variables and depression prevalence and the composition of the national diet

Foodstuff Schizophrenia outcome Depression prevalence

IPSS DOSMED

Mean days % Patients with severe Total outcome Never % Patients with little Total best
out of hospital social impairment score hospitalised social impairment outcome score

Meat 0.86** 0.81* 0.78* 0.82*** 0.63 0.63 0.70

Fish 0.70 0.54 0.60 0.21 0.34 0.46 70.85**
Eggs 0.91*** 0.86** 0.83* 0.80** 0.60 0.65 70.12
Dairy 0.92*** 0.87** 0.88*** 0.58 0.60 0.55 0.71*
Vegetable oil 0.37 0.16 0.13 0.31 0.22 0.11 70.07
Sugar 0.91*** 0.89*** 0.94*** 0.93*** 0.82** 0.89*** 0.74*
Cereals 70.15 70.32 70.27 70.20 70.24 70.12 70.53
Pulses 70.64 70.62 70.76* 70.77** 70.60 70.67 0.11
Vegetables 0.66 0.49 0.45 0.56 0.59 0.59 70.27
Starchy roots 70.19 70.42 70.35 0.40 0.47 0.50 70.75*
Fruits 0.09 70.02 70.00 0.34 0.47 0.30 0.50
Nuts 70.04 70.30 70.31 0.09 70.26 70.28 70.01
Coffee 0.45 0.48 0.60 0.48 0.39 70.43 0.39
Alcohol 0.81* 0.70 0.72* 0.51 0.20 0.24 0.69

IPSS, International Pilot Study of Schizophrenia; DOSMED, Determinants of Outcome of Severe Mental Disorders.
Positive correlations reflect a detrimental relationship between the nutritional variable and the clinical outcome.
*P50.05; **P
**P50.01; ***P
***P50.005.

405
PEET

T
Table
able 2 Multiple regression analysis of food categories in the national diet as predictors of the outcome of schizophrenia and prevalence of depression

Dependent variable Predictor b-coefficient Adjusted R2 P

IPSS Mean days out of hospital Dairy products 70.92 0.82 0.001
% Patients with severe social impairment Sugar 0.89 0.76 0.003
Total outcome score Sugar 0.94 0.87 50.0005

DOSMED Never hospitalised Sugar 70.93 0.86 50.0005

% Patients with little social impairment Sugar 70.82 0.62 0.007
Total best outcome score Sugar 71.14 0.87 50.0005
Alcohol 0.42 0.04

Depression prevalence Fish and seafood 70.89 0.74 0.008

IPSS, International Pilot Study of Schizophrenia; DOSMED, Determinants of Outcome of Severe Mental Disorders.

as diagnostic differences and selective depression database as Hibbeln (1998), we The databases used for schizophrenia
outcome measures (Hopper & Wanderling, have confirmed the relationship between outcome are well accepted and often cited.
2000). Christensen & Christensen (1988) fish consumption and international varia- The data on major depression prevalence
reported a correlation between international tions in rates of depression, and also found reported by Weissman et al (1996) are
variations in outcome of schizophrenia that sugar consumption relates to the preva- derived from rigorous methodologies and
according to the IPSS study and the ratio lence of depression. This had been noted are among the most reliable cross-national
in the diet of animal (mainly saturated) fat previously by Westover & Marangell data available.
to fish and vegetables (mainly polyun- (2002). However, multiple regression It is well recognised that the use of
saturated) fats. This was reflected in the analysis shows that fish and seafood multiple correlations can give rise to
present study, where correlations were consumption provides the strongest and spurious associations by chance. Also,
shown between a higher consumption of most robust independent predictor of multiple regression analysis can be sensitive
meat and dairy products and a worse out- depression prevalence. to the exact regressors entered. However,
come of schizophrenia. However, strong the breadth of data used in this study has
intercorrelations are found between various enabled patterns to be identified and this
dietary constituents, and on multiple re- Limitations was strengthened by the robust findings
gression analysis it was sugar consumption Association does not prove causation. Diet from the use of multiple regression analysis.
that was the predominant predictor of poor is partly determined by sociocultural
outcome in schizophrenia. Exceptions to factors, so diet could be a proxy for other
this were that the consumption of dairy social variables. Correlation studies of this Implications
products predicted hospital admission in type, particularly when taken in conjunc- Diabetes, coronary heart disease and other
the IPSS study, and alcohol was a weak tion with other sources of information, related conditions that cluster together have
predictor of global good outcome in the have their greatest value in generating been conceptualised as manifestations of
DOSMED study. Therefore, the dominant hypotheses. Issues of cause and effect can the ‘metabolic syndrome’ (De Fronzo &
and robust finding of this analysis is the be clarified only by intervention studies in Ferrannini, 1991; Hansen, 1999). Because
predictive value of sugar consumption. which diet is manipulated, by either supple- these diseases also cluster with schizo-
mentation or restriction, and mental health phrenia and depression, it has been
outcome is assessed. proposed that these mental disorders may
Diet and prevalence of depression Correlations are only as good as the share some aetiological factors with phy-
There has been recent interest in the data-sets utilised. The national apparent sical diseases that constitute the metabolic
relationship between fish consumption and food consumption data are the best avail- syndrome (Peet & Edwards, 1997; Ryan
depression. Hibbeln & Salem (1995) noted able but they are inexact and do not neces- & Thakore, 2002). A fundamental
that the increased prevalence of depression sarily reflect the dietary practices of abnormality in the metabolic syndrome is
in the 20th century parallels the increased individuals or of population subgroups. insulin resistance (Hansen, 1999). Abnor-
rates of coronary heart disease that are This can be done only by case–control or mal glucose utilisation reflecting insulin
thought to be associated with altered diet- cohort studies in which individual dietary resistance has been demonstrated in
ary patterns, including reduced dietary practices are recorded. There are several patients with depression (Peet & Edwards,
intake of omega-3 polyunsaturated fatty large prospective studies relating diet to 1997). In schizophrenia, abnormalities of
acids. Hibbeln (1998) has subsequently subsequent heart disease, diabetes and glucose utilisation were demonstrated
demonstrated striking correlations between cancer, but there are no large prospective before the introduction of modern anti-
dietary fish intake and international varia- studies relating diet to subsequent mental psychotic drugs (Ryan & Thakore, 2002).
tions in major depression. Using the same health problems. Insulin resistance is altered by diet. Regular

406
 DIE T AN
NDD P S YC H I AT R I C I L L NE
NE S S

consumption of significant quantities of

foodstuffs with a high ‘glycaemic load’,
CLINICAL IMPLICATIONS
such as sugar and white bread, which
rapidly release glucose into the blood- & Variations in the national diet predict international variations in the outcome of
stream, gives rise to insulin resistance and
schizophrenia and the prevalence of depression.
subsequent susceptibility to diabetes and
cardiovascular disease (Ludwig, 2002). A & The dietary predictors for mental health are the same as those that are known to
high dietary intake of saturated fat leads be of causal importance for physical health problems, including coronary heart
to increased insulin resistance, whereas disease and diabetes.
substitution of saturated fat with poly-
unsaturated fat can reverse this effect & These findings allow the hypothesis that nutrition is important in the genesis and
and may be protective against future maintenance of mental ill health.
development of insulin resistance (Summers
LIMITATIONS
et al,
al, 2002). The risk of mortality from
coronary heart disease is decreased by & Data for the national diet are crude and approximate.
regular consumption of fish and pulses
(Mann, 2002). Thus, dietary patterns that & Diet is partly determined by social factors, so diet could be a proxy for other
influence insulin resistance and lead to sociocultural variables.
diseases of the metabolic syndrome are
reflected by the dietary predictors of out- & Association does not prove causation. Further dietary intervention studies are
come of schizophrenia and prevalence of required to test the hypothesis.
depression in the present study. It can there-
fore be hypothesised that insulin resistance
and associated metabolic disturbances
resulting from dietary factors may account
for the clinical association between dep- MALCOLM PEET, FRCPsych, Swallownest Court Hospital, Aughton Road, Sheffield S26 4TH, UK.
ression, schizophrenia and the physical [email protected]
Tel: +44 (0)114 2872570; fax: +44 (0)114 2879147; e-mail: malcolmpeet@
diseases of the metabolic syndrome.
(First received 24 January 2003, final revision 30 May 2003, accepted 3 September 2003)
An obvious practical consequence of
this hypothesis is the possibility that mental
health could be improved by dietary manip-
ulation. There have been few controlled
trials of nutrition in the treatment of psy- Insulin receptors are widely distributed in REFERENCES
chiatric disorders. If food with a high the brain where they are involved in cell
glycaemic load is of importance in deter- signalling and modulate the actions of other Christensen, O. & Christensen, E. (1988) Fat
consumption and schizophrenia. Acta Psychiatrica
mining the outcome of schizophrenia, then neurotransmitters (Kyriaki, 2003). Also, a Scandinavica,
Scandinavica, 78,
78, 587^591.
modification of glycaemic load or its meta- high-fat, refined sugar diet has been shown
De Fronzo, R. A. & Ferrannini, E. (1991) Insulin
bolic consequences should be beneficial in to reduce levels of brain-derived neuro- resistance: a multifaceted syndrome responsible for
the management of schizophrenia. Diets trophic factor in rat hippocampus (Molteni NIDDM, obesity, hypertension, dyslipidemia, and
with a low glycaemic load were promoted et al,
al, 2002): brain-derived neurotrophic atherosclerotic cardiovascular disease. Diabetes Care,
Care, 14,
14,
173^194.
decades ago as being potentially useful in factor promotes a variety of neuromodula-
Durany, N., Michell, T., Zochling, R., et al (2001)
the treatment of schizophrenia but have tory processes during early development
Brain-derived neurotrophic factor and neurotrophin 3
been supported only by anecdotal evidence and adulthood and has been implicated in in schizophrenia. Schizophrenia Research,
Research, 52,
52, 89^86.
(Meiers, 1973). schizophrenia (Durany et al,
al, 2001). Food and Agriculture Organization (2002) Food and
With regard to depression, the over- In conclusion, this study has shown Agriculture Organization of the United Nations
riding association was between fish and that national dietary factors predict Statistical Database: http://apps.fao.org/default.jsp
seafood consumption and prevalence of international variations in the outcome of Hansen, B. C. (1999) The metabolic syndrome X.
depression, which supports previous schizophrenia and prevalence of depres- Annals of the New York Academy of Science,
Science, 892,
892, 1^24.

epidemiological data (Hibbeln, 1998). This sion. The nutritional predictors are similar Hibbeln, J. (1998) Fish consumption and major
to those reported for other ‘Western’ depression (letter). Lancet,
Lancet, 351,
351, 1213.
leads directly to the proposition that
depression should be treatable with diseases such as diabetes and coronary Hibbeln, J. R. & Salem, N., Jr. (1995) Dietary
polyunsaturated fatty acids and depression: when
omega-3 fatty acids. This proposition has heart disease, which are more common in
cholesterol does not satisfy. American Journal of Clinical
been tested recently, with strong positive people with mental health problems. Nutrition,
Nutrition, 62,
62, 1^9.
antidepressant benefits reported and Whether the relationship between nutri- Hopper, K. & Wanderling, J. (2000) Revisiting the
confirmed (Nemets et al, al, 2002; Peet & tional factors and mental health is causal developed versus developing country distinction in
Horrobin, 2002). This is now a matter of can be determined only by intervention course and outcome in schizophrenia: results from ISOS,
the WHO collaborative follow-up project. Schizophrenia
intense further research. studies. Initial treatment studies with Bulletin,
Bulletin, 26,
26, 835^846.
It is premature to speculate on mol- omega-3 fatty acids in depression are
Jablensky, A., Sartorius, N., Emberg, G., et al (1992)
ecular mechanisms, save to note that a encouraging but other nutritional strategies Schizophrenia: manifestations, incidence and course
number of potential mechanisms do exist. should be explored. in different cultures: a World Health Organization

4 07
PEET

10 -country study. Psychological Medicine, Monograph brain-derived neurotrophic factor, neuronal plasticity, Summers, L. K., Fielding, B. A., Bradshaw, H. A., et
Supplement,
Supplement, 20,
20, 1^97. and learning. Neuroscience,
Neuroscience, 112,
112, 803^814. al (2002) Substituting dietary fat with polyunsaturated
fat changes abdominal fat distribution and improves
Kyriaki, G. (2003) Brain insulin: regulation, mechanisms Nemets, B., Stahl, Z. & Belmaker, R. H. (2002) insulin sensitivity.
sensitivity. Diabetologia,
Diabetologia, 45,
45, 369^377.
of action and functions. Cellular and Molecular Addition of omega-3 fatty acid to maintenance
Neurobiology,
Neurobiology, 23,
23, 1^25. medication treatment for recurrent unipolar depressive
Tucker, K. L. & Buranapin, S. (2001) Nutrition and
disorder. American Journal of Psychiatry,
Psychiatry, 159,
159, 477^479.
Ludwig, D. S. (2002) The Glycaemic Index. aging in developing countries. Journal of Nutrition,
Nutrition, 131,
131,
Physiological mechanisms relating to obesity, diabetes Peet, M. & Edwards, Rh. W. (1997) Lipids, depression 2417S ^2423S.
and cardiovascular disease. JAMA,
JAMA, 287,
287, 2414^2423. and physical diseases. Current Opinion in Psychiatry,
Psychiatry, 10,
10,
477^480. Weissman, M. M., Bland, R. C., Canino, C. J., et al
Mann, J. I. (2002) Diet and risk of coronary heart (1996) Cross-national epidemiology of major depression
disease and type 2 diabetes. Lancet,
Lancet, 360,
360, 783^789. Peet, M. & Horrobin, D. F. (2002) A dose-ranging and bipolar disorder. JAMA,
JAMA, 276,
276, 293^299.
study of ethyleicosapentaenoic acid in patients with
Meiers, R. L. (1973) Relative hypoglycaemia in ongoing depression despite apparently adequate
Westover, A. N. & Marangell, L. B. (2002) A cross-
schizophrenia. In Orthomolecular Psychiatry (eds D. treatment with standard drugs. Archives of General
national relationship between sugar consumption and
Hawkings & L. Pauling), pp. 452^462. San Francisco, CA: Psychiatry,
Psychiatry, 59,
59, 913^919.
major depression. Depression and Anxiety,
Anxiety, 16,
16, 118^130.
W. H. Freeman.
Ryan, M. C. & Thakore, J. H. (2002) Physical
Molteni, R., Barnard, R. J.,Ying, Z., et al (2002) A consequences of schizophrenia and its treatment: the World Health Organization (1979) Schizophrenia: an
high-fat, refined sugar diet reduces hippocampal metabolic syndrome. Life Sciences,
Sciences, 71,
71, 239^257. International Follow-up Study.
Study. New Y
York:
ork: John Wiley.

408