Mood Food: Chocolate and Depressive Symptoms in a Cross-sectional Analysis

Background  Much lore but few studies describe a relation of chocolate to mood. We examined the cross-sectional relationship of chocolate consumption with depressed mood in adult men and women.

Methods  A sample of 1018 adults (694 men and 324 women) from San Diego, California, without diabetes or known coronary artery disease was studied in a cross-sectional analysis. The 931 subjects who were not using antidepressant medications and provided chocolate consumption information were the focus of the analysis. Mood was assessed using the Center for Epidemiologic Studies Depression Scale (CES-D). Cut points signaling a positive depression screen result (CES-D score, ≥16) and probable major depression (CES-D score, ≥22) were used. Chocolate servings per week were provided by 1009 subjects. Chocolate consumption frequency and rate data from the Fred Hutchinson Food Frequency Questionnaire were also available for 839 subjects. Chocolate consumption was compared for those with lower vs higher CES-D scores. In addition, a test of trend was performed.

Results  Those screening positive for possible depression (CES-D score ≥16) had higher chocolate consumption (8.4 servings per month) than those not screening positive (5.4 servings per month) (P = .004); those with still higher CES-D scores (≥22) had still higher chocolate consumption (11.8 servings per month) (P value for trend, <.01). These associations extended to both men and women. These findings did not appear to be explained by a general increase in fat, carbohydrate, or energy intake.

Conclusion  Higher CES-D depression scores were associated with greater chocolate consumption. Whether there is a causal connection, and if so in which direction, is a matter for future prospective study.

A rich cultural tradition links chocolate consumption with putative mood benefits. Indeed, the words “chocolate” and “mood” yield over 5 690 000 Google search results (March 22, 2010). Potential means by which chocolate may mediate purported mood benefits have been described.¹ However, scientific interest has been less robust: epidemiological studies directly examining the association between chocolate consumption and mood in humans remain sparse.

One exploratory analysis of the relation of food to mood suggested a specific connection of depressive symptoms to chocolate consumption.² However, the chocolate analysis was not hypothesis driven, and a single such association could occur by chance. In addition, only women were examined and the study used a single mood instrument that is not in frequent use in clinical studies.

We examined the relationship of chocolate consumption to mood in an adult study sample including men and women, using several chocolate consumption metrics assessed against a widely recognized mood instrument.

A total of 1018 subjects (694 men and 324 women aged 20 to 85 years) from the San Diego area in California, without known cardiovascular disease, diabetes, or extremes of high or low level of low-density lipoprotein cholesterol, were screened for participation in a clinical study (examining noncardiac effects of reducing cholesterol level) that sampled broadly across the adult age range.^3,4 Of the 1018 subjects, 1009 completed both the depression instrument and a study-specific “Statin Study Questionnaire” chocolate consumption (SSQ-C) question, prior to study intervention. Of these, 78 subjects (35 men and 43 women) were using antidepressant medications and were excluded from the primary analysis. Thus, 931 subjects were the focus of this analysis.

All subjects were asked “How many times a week do you consume chocolate?”

A total of 839 subjects completed the Fred Hutchinson Food Frequency Questionnaire (FFQ) developed by the Nutrition Assessment Shared Resource of the Fred Hutchinson Cancer Research Center.^5,6 The FFQ has been validated against means from four 24-hour delay recalls and a 4-day food record⁷ and has been shown to predict objectively assessed serum nutrient concentrations.⁸ A medium serving on this assessment was considered to be 1 small bar or 1 ounce (28 g) of chocolate candy. A small serving, equal to one-half of a medium serving, and a large serving, equal to 1½ of a medium serving, were converted to the corresponding number of medium servings for the purpose of analysis. Responses were converted from daily or monthly consumption into per-month consumption to provide a unified metric. The 2 chocolate measures from the FFQ (FFQ-C) comprised the absolute frequency of any chocolate consumption (times per month) and the rate of chocolate consumption (servings per month).

To assess for specificity of the chocolate finding, consumption of several other foods and nutrients were analyzed including energy intake, carbohydrates, and fat.

Subjects completed the Center for Epidemiological Studies–Depression Scale (CES-D), a validated and widely used instrument related to mood.^9-11 A CES-D score of 16 or higher is often considered to reflect a positive screening result¹⁰ and was used as a cutoff in this study. To allow further analysis within the presumptively depressed range, a second cutoff score of 22 or higher conformed to a threshold found in 1 venue to characterize probable major depression.¹²

Subject characteristics at baseline were evaluated for the set of subjects in whom CES-D was measured, and at least 1 measure of chocolate consumption was procured.

Self-reported chocolate consumption was assessed for the total sample and stratified by sex, dichotomizing CES-D score at a threshold of 16 and 22. Both the study-specific chocolate measure (SSQ-C) and the FFQ-C items were assessed. Unpaired t tests compared mean chocolate consumption stratified by CES-D category.

Evidence of a dose-response relationship was additionally assessed (ie, whether more-depressed persons appeared to consume more chocolate) across 3 CES-D categories defined by the 2 cut points (CES-D scores of 16 and 22).

To evaluate whether the effect was nonspecific (eg, related to total energy intake, carbohydrates, or fat), we examined additional FFQ items of potential interest to determine the specificity of the finding.

Of the subjects, the mean (range) age was 57.6 (20.0-85.5) years and body mass index (calculated as weight in kilograms divided by height in meters squared) was 27.8 (17.3-49.6); 70.1% were male, 80.4% were white, and 58.8% were college graduates. The mean CES-D score was 7.7, with a median of 6.0. The CES-D scores ranged from 0 to 45 (maximum possible score, 60). Mean chocolate consumption values were 6.0 medium servings per month and 6.0 times per month using the FFQ-C and 2.0 servings per week using the SSQ-C. Subjects with a CES-D score of 16 or higher reported significantly more chocolate consumption compared with those with lower CES-D scores (Figure 1). These findings were significant with all 3 measures of chocolate consumption frequency (SSQ-C times per week; FFQ-C times per month; and FFQ-C servings per month). Significance was evident with a cut point of 16 as shown, as well as with a cut point of 22 (data not shown). Findings were significant separately for men and women (P = .03 for men and P = .04 for women using the FFQ-C times-per-month measure). The FFQ-C, the measure with a greater history of use, gave closely similar results whether times or servings per month were used.

A nonparametric test of trend for a CES-D score lower than 16, 16 to 21, and 22 or higher was also significant for the total sample (P = .01) and for men separately (P = .03), but not significant for women separately (P = .07), using the FFQ-C times-per-month measure. There were fewer women than men in the study, providing lower statistical power for women; however, women clearly contribute to power for the total sample (Figure 2). Similar results were found for the other 2 measures as shown in Figure 2.

To evaluate specificity of the finding, other foods and nutrients that might be hypothesized to underlie the appearance of a chocolate-mood relationship were assessed (Table). In contrast to the findings for chocolate, differences in consumption of fat, energy, or carbohydrates by CES-D group were not significant, suggesting relative specificity of the chocolate finding. Of note, mean (SD) body mass index did not differ by CES-D category: 27.7 (4.3) (CES-D score <16) vs 27.8 (4.5) (CES-D score ≥16) (P = .74). Moreover, there was no difference in consumption of other presumed antioxidant foods such as fish, coffee, caffeine, and fruits and vegetables between the 2 groups.

Depressed mood was significantly related to higher chocolate consumption. Findings were similar in men and women. Higher CES-D scores within the provisionally depressed range (above the depression screen threshold) were associated with still greater chocolate consumption. Several nutrient factors that could be theorized to drive the appearance of a putative chocolate-mood association, such as caffeine, fat, carbohydrate, and energy intake, bore no significant correlation with mood symptoms, suggesting relative specificity of the finding.

This finding is an important addition to existing literature associating mood with chocolate consumption. A previous study incidentally noted an apparent chocolate-depression relationship²; that study was confined to women, used a single measure of chocolate consumption, and used a less-recognized measure of mood symptoms. It was an exploratory study of sweets in general; the relationship of mood with chocolate consumption was identified incidentally and thus could have reflected a chance association among multiple comparisons. Ecological analysis has previously suggested a relation of increased chocolate consumption to an outcome related to depressed mood: a study of European countries showed that higher national chocolate consumption was correlated with an increased suicide rate.¹³ However, the ecological fallacy¹⁴ (applicable where data at the level of the group rather than individual are linked) is a concern with this study design. Another study linked the “desire for chocolate” with higher depressive symptom ratings but not suicidal ideation.¹⁵ The present study focuses on actual consumption of chocolate.

A prior study found a connection between caffeine consumption and better mood in women.¹⁶ Although phosphodiesterase inhibitors or methylxanthines such as caffeine and theobromine (found in chocolate) could be postulated to engender an association of chocolate with mood, some maintain that the content of caffeine and theobromine in chocolate is too low to account for potential effects of chocolate.¹⁷ We observed no association of caffeine with mood in our sample.

This study has several limitations. First, the study was conducted for a different initial purpose; however, the initial proposal specified that the baseline database would be used for cross-sectional analyses relating predictors to outcomes such as mood and behavior. Moreover, the present analysis was performed in a hypothesis-driven fashion: the determination to examine the chocolate-mood association was made prior to examination of the chocolate data, the mood data, or evaluation of any association between nutrients and mood. In addition, the findings buttress and substantially extend previous findings examining chocolate and mood, and several factors improve the authority of the findings, including consistency of findings for men and women and with several measures of chocolate consumption, use of a widely recognized depression rating, and demonstration of relative specificity of the finding. Second, these data are based on self-report of diet, and chocolate and other nutrient consumption could have error or bias. However, self-reported diet data are widely used, and the inherent limitations of self-report are similar to those of any nutritional epidemiologic study based on recall. Third, although the reliability and validity of the CES-D have been confirmed,¹⁰ and it was designed for use in broadly sampled groups such as those herein⁹ (and has subsequently been demonstrated to be valid in an extensive range of settings), it is still a screening test and does not indicate any Diagnostic and Statistical Manual of Mental Disorders diagnosis of a mood disorder. Fourth, although the present study assessed and presumptively excluded a confounding contribution by key macronutrient elements such as fats, energy intake, and carbohydrates, it did not assess some potential mediating factors such as phenylethylamine, anandamine, or theobromine levels, which will be left for future studies. Moreover, different chocolate products that may differ in characteristics are not distinguished.

This is a cross-sectional study, precluding inferences about causality or even directionality. Thus, findings are compatible with any of several possibilities. First, depression could stimulate chocolate cravings as “self-treatment” if chocolate confers mood benefits, as has been suggested in recent studies of rats.¹⁸ Second, depression may stimulate chocolate cravings for unrelated reasons, without a treatment benefit of chocolate (in our sample, if there is a “treatment benefit,” it did not suffice to overcome the depressed mood on average). Third, from cross-sectional data the possibility that chocolate could causally contribute to depressed mood, driving the association, cannot be excluded. Fourth, a physiological factor (eg, oxidative stress or inflammation) could drive both depression and chocolate cravings. Fifth, more complex relationships could be hypothesized; for instance, chocolate could have intrinsic mood-elevating effects driving chocolate cravings, but artificial trans fats (which inhibit omega-3 fatty acid production^19-21 with expected mood worsening consequences^22,23) accompany many chocolate products and may neutralize or reverse such benefits on average. Alternatively (or consistent with the prior statement), analogous to alcohol, there could be short-term benefits of chocolate to mood with longer-term untoward effects. Distinguishing among these possibilities will require different study designs.

In conclusion, people with higher depression ratings consume more chocolate. Future studies are required to elucidate the foundation of the association and to determine whether chocolate has a role in depression, as cause or cure.

Correspondence: Beatrice A. Golomb, MD, PhD, Department of Medicine, University of California, San Diego, 9500 Gilman Dr, MC 0995, La Jolla, CA 92093-0995 (bgolomb@ucsd.edu).

Accepted for Publication: November 3, 2009.

Financial Disclosure: None reported.

Funding/Support: The procurement of the data was funded by grant RO1 HL63055 from the National Heart, Lung, and Blood Institute, National Institutes of Health, and by the University of California, San Diego General Clinical Research Center (National Institute of Health grant MO1 RR00827).

Additional Contributions: Joel Dimsdale, MD, and Cheryl Rock, PhD, RD, provided comments on earlier drafts of the manuscript.