The Socioeconomic Gradient to Late-Life Depression: a
Potential Pathway of Fruit and Vegetable Intake
Evidence from the Survey of Health, Ageing and Retirement in Europe
Written by:
M.L. Noordveld
s2026414
Supervisor:
Dr. J.O. Mierau
14
thof June, 2018
2 Abstract:
Depression is a prevalent condition that is predicted to be the largest contributor to the global burden of disease by 2030. Action is needed to reduce this burden. Depression prevalence is highest in late adulthood and to depression prevalence there is a clear socioeconomic gradient. This research investigates one potential transmission mechanism through which socioeconomic status affects late-life depression. With data from the Survey of Health, Ageing and Retirement in Europe, a logistic regression estimation indicates that sufficient fruit and vegetable intake reduces the chance of a depression. Moreover, the results indicate that fruit and vegetable intake could indeed be a pathway through which socioeconomic status affects late-life depression. To these effects, no significant sex differences were found. Additional sensitivity analyses indicate little support for the explanation that insufficient fruit and vegetable intake is the result of financial limitations. A potential solution to increase fruit and vegetable consumption and reduce depression prevalence might be found in nutritional knowledge.
Keywords:
Health economics, mental health, late-life depression, socioeconomic gradient, diet quality, fruit and vegetables
3
1. Introduction
In 2017, 4.4% of the world population suffered from a depression, which comprises 300 million people (World Health Organization, 2017). Mathers et al. (2008) predict that by 2030, depression will be the largest contributor to the global burden of disease. Past literature often indicated an inverted u-shape of depression prevalence over the life cycle, where prevalence was highest in young adulthood and around the age of 60 and decreased again after this age (Mirowsky & Ross, 1992; Gutiérrez-Lobos et al., 2002). However, recent data indicates that prevalencerates of depression are the highest for the age group from 55 to 74 years old (World Health Organization, 2017). Especially in women, prevalence rates remain high in older ages. This research focuses on depression in late adulthood, namely on late-life depression. Late-life depression has severe consequences for the individual such as higher risk of cardiovascular diseases (Van der Kooy et al., 2007), early retirement (Karpansalo et al., 2005), declined cognitive and physical functioning (Blazer, 2003) and ultimately, higher risk of suicide and early morbidity (Chapman & Perry, 2008).
However, late-life depression also has large societal consequences. In 2004, the cost burden for depression in Europe was already estimated at 188 billion euro (Sobocki et al., 2006). Contributors to this burden are not only treatment costs; indirect costs due to morbidity and mortality are also included. One could argue that the actual cost burden is even higher if other costs are included such as, for example, reduced economic productivity due to early retirement and sick leave (Karpansalo et al., 2005).
Increasing prevalence and rising cost burdens of late-life depression call for government action to help prevent this serious condition. There is overwhelming evidence that a socioeconomic gradient to health and specifically to late-life depression exists (Adler et al., 1994; Lorant et al., 2003; Conti et al. 2010). However, the exact mechanisms through which socioeconomic factors affect health are not all yet established. This research investigates such a potential transmission mechanism through which socioeconomic status affects late-life depression: fruit and vegetable intake. Fruit and vegetables are an important diet component and they are proven effective in reducing disease burdens of, for example, coronary heart disease, multiple types of cancer, and depression (Lock et al. 2005). Since fruit and vegetable intake is associated to both socioeconomic status and late-life depression, it can potentially have a mediating effect in the socioeconomic gradient of late-life depression. If this is the case, promoting fruit and vegetable consumption could be a valuable policy strategy that can reduce prevalence of this condition and can help contain the associated cost burden, while it will help reduce the prevalence of other diseases at the same time. Moreover, increased fruit and vegetable consumption is a relatively inexpensive option to reduce depression prevalence and one that does not lead to side effects, as is the case with, for example, antidepressants (Papakostas, 2008).
4 predicted to rise even further (Mathers et al., 2008), preventative measures will ultimately become essential to reduce prevalence and contain costs. If evidence of a potential transmission mechanism will be found, this gives rise to interesting opportunities, not only for depression prevention, also for other diseases that have a socioeconomic gradient and are associated with fruit and vegetable consumption. In addition, this research adds to the still limited cross-national literature on late-life depression prevalence. Therefore, it can provide useful insights for the European Union and its member states. In the remainder of this paper, first a review of the literature on late-life depression, socioeconomic status, fruit and vegetable intake and their interrelatedness is provided. A methodology chapter that elaborates on the data, research method and descriptive statistics follows the literature review. Then, results are given, followed by a discussion of the findings of this study, recommendations and limitations. Lastly, a conclusion is provided.
2. Literature review
In this literature review, first the characteristics of late-life depression are explained. This explanation is followed by an overview of the literature on socioeconomic status, and by an explanation on how both concepts relate to fruit and vegetable intake and to each other. Following this overview, the last subsection will provide the research question and hypotheses of this research.
2.1 Characteristics of late-life depression
The World Health Organization (WHO) defines depression as ‘a common mental disorder, characterized by sadness, loss of interest or pleasure, feelings of guilt or low self-worth, disturbed sleep or appetite, feelings of tiredness and poor concentration’ (WHO, 2017, p7). Depression in late-life has specific characteristics. Late-life depression is a serious condition leading, for example, to declined physical and cognitive functioning and higher risk of early morbidity and mortality (Blazer, 2003; Chapman & Perry, 2008). In late-life, depression is often chronic and comorbid (Beekman et al., 2002; Krishnan, 2002). Over the years, certain risk factors and determinants of late life-depression have been discovered. Depression is often observed after a stroke, after or during cancer, and after or during heart diseases. Additionally, it can be genetically determined, the result of adverse life event or of a lack of social support, or the result of overall health and disability deterioration (Kennedy et al., 1990; Krishnan, 2002). One form of such an adverse life events that often occurs in old age is bereavement (Alexopoulos, 2005). Bereavement-related depression is regularly confused with grieving or prolonged grief and therefore it is often undertreated (Rosenzweig et al., 1997). A year after bereavement, approximately 15% of older individuals experience depressive symptoms (Hensley, 2006).
5 setbacks. However, a different explanation is that the gap arises from reporting differences, since women are more inclined to report depressive symptoms (Acciai & Hardy, 2017).
Another important characteristic is that there is a strong socioeconomic gradient to late-life depression. This gradient will be discussed in the next subsection.
2.2 The socioeconomic gradient to late-life depression
6 expected that inequalities are not very big and that cohort effects will be moderate. Additionally, by controlling for respondents their age, birth cohort effects can be accounted for.
Earlier research already found support for the effect of educational attainment on late-life depression. Using SHARE data Ladin (2008) finds that individuals with low educational attainment have a significantly higher chance of becoming depressed in late-life. This result remained significant while controlling for other sociodemographic factors such as sex, income and cohabitation. It is plausible that this effect results from multiple mechanisms that are at play. Although literature on these mechanisms is still limited for the gradient to late-life depression, there is extensive research on the gradient to mental health and health in general. When we look at the gradient between educational attainment and mental health in later life, differences in health behaviour could be a potential explanation of health disparities. Cutler & Lleras-Muney (2010) find that 30%of the gradient can be explained by cognitive ability and knowledge. Most important is not factual knowledge, it is general cognitive capability, the way in which someone processes information which according to them, increases with education. In turn, this general cognitive capabilityimproves health behaviour. Better health behaviour results, for example, in lower chances of smoking, substantial alcohol consumption and of obesity. Additionally, they argue that education increases life opportunities and the ability to cope with adverse life advents that in turn reduce the chance of mental health problems. Conti et al. (2010), who look at the socioeconomic gradient to general health, support the finding that the ability to process information is an important determinant of the socioeconomic gradient. Moreover, they find that noncognitive capabilities are also an important determinant of health, however, they add that inadequate noncognitive capabilities can be compensated by education.
Education-dependent health behaviour could also affect fruit and vegetable intake, which will be discussed in the next subsection.
2.3 SES and fruit and vegetable intake
Diet quality is strongly related to SES, low SES is often associated with an unhealthy diet. Hulshof et al. (2013) find in their study with survey data of dietary intake in the Netherlands, that respondents with low SES consumed more meat, potatoes, visible fats, coffee and soft drinks. Respondents with higher SES consumed more vegetables and cheese and received more dietary fibre and necessary micronutrients. Turrell et al. (2002) find with data from Australia that people with low SES buy food that is not consistent with dietary recommendations. They eat too much salt, sugar and fat and do not eat enough fruit and vegetables.
7 for all ethnicities and varies between men and women. For Whites and women, this effect is the most profound. Since SES is measured as educational attainment in these studies, this could mean that there is an education effect to diet quality. And indeed, an important explanation for poor diet quality is the lack of nutritional knowledge (Turrell & Kavanagh, 2006; Beydoun & Wang, 2008a). McKinnon et al. (2012) divide nutritional knowledge into three types: knowledge of the recommendations of dietary guidelines, knowledge of the relationship between nutrition and disease and knowledge of the nutritional content of different foods. In their research they find that especially knowledge of the relationship between nutrition and disease is related to socio-economic differences to diet quality. Price differences between healthy and unhealthy nutrition are another explanation for the association between SES and diet quality, or more specifically, fruit and vegetable intake. Monsivais et al. (2012) find that differences in dietary intake between levels of SES can be largely explained by diet costs, since healthy food is more expensive than unhealthy food (high in sugar and saturated fats). However, their study is solely based on data from the United States. Mackenbach et al. (2015) find with data from the United Kingdom that there is a strong association between diet costs and fruit and vegetable intake for respondents with low SES. They argue that price reductions, for example by subsidizing fruit and vegetables, could reduce disparities in fruit and vegetable intake. Moreover, they note that further research should also investigate physical access to fruit and vegetables. Physical access differences could be the result of area deprivation or the result of supply differences between supermarket segments. However, Pechey & Monisivais (2015) argue that this association is not the result of supply differences between high-cost and low-cost supermarkets. Moreover, Black et al. (2014) argue that area deprivation, where food accessibility depends on the area where you live, is indeed a determinant of disparities in diet quality in the United States. They also argue that the United States is an exception and that this does not hold for other developed countries.
2.4 Fruit and vegetable intake and late-life depression
8 depression. With data from questionnaires that were filled out by older Americans, they find that fruit and vegetable intake is low for the individuals that reported a depression. Interestingly, they did not find a relationship between antioxidant supplements and depression, suggesting that fruit and vegetables contain essential components that positively affect mental health and cannot be replaced by supplements. There is also evidence that indicates that certain micronutrients are inversely related to depression. These micronutrients include folate, zinc and magnesium. Especially folate and magnesium are present in many fruits and vegetables, which adds to the health benefits of fruit and vegetables (Gómez-Pinilla, 2008; Jacka et al., 2012).
The consumption of fruit and vegetables is not only effective in preventing depression. Sufficient fruit and vegetable intake is also found to reduce chances of many health problems including ischemic strokes (Joshipura et al., 1999), coronary heart disease (Joshipura et al., 2001) and certain types of cancer (Riboli & Norat, 2003). However, all these diseases are known to also increase the chance of a depression in late-life (Krishnan, 2002). The multi-effectiveness of fruit and vegetable intake makes it an efficient focus area for policies aiming to reduce disease burdens. In the specific case of depression, another advantage of fruit and vegetable intake is that intake does not come with the risk of side effects. This is in contrast to other treatment options such as the use of antidepressants, which is often accompanied by substantial side effects (Papakostas, 2008).
Although there is no doubt that the consumption of fruit and vegetables has many health benefits, it is difficult to establish an intake threshold for consumption to positively affect mental health. However, for general health, such an intake threshold level is established. Lock et al. (2005) advocate for a transition back to a diet that is largely plant-based and find that the optimal consumption level of fruit and vegetables is 600 grams per day. Additionally, they estimate that about 1.8% of the global burden of disease is the result of insufficient fruit and vegetable intake. This makes fruit and vegetable consumption a significant determinant comparable to for example physical activity and obesity. They argue that policymakers should focus on a target consumption level of at least 400-500 grams per day, that is approximately five servings. This advice is in line with the advice of the WHO that advocates for a daily intake of more than 400 grams of fruit and vegetables (World Health Organization, 2003). An important note here is the fact that in old age, people often start eating less and experience a loss of appetite. This can be the result of many reasons, including for example the use of medication or reduced activity (Pilgrim et al., 2015). However, although they eat less, Pilgrim et al. (2015) argue that it is still very important to keep dietary recommendations in mind and to maintain a healthy diet, to reduce the risk of nutritional deficiencies.
2.5 Fruit and vegetable as a potential pathway from SES to late-life depression
9 in late-life. This could be a stand-alone effect. However, it could also be a transmission mechanism through which socioeconomic status affects late-life depression. This research will attempt to relate the effect of fruit and vegetable consumption on life depression to the socioeconomic gradient of late-life depression. Undoubtedly, this will not be the only explanation to the socioeconomic gradient, however, it could be one of the explanations. In section 2.3, it was explained that educational attainment improves health behaviour (Cutler & Lleras-Muney, 2010). Improved health behaviour could lead to more awareness of diet quality, leading to higher fruit and vegetable consumption. However, there are more reasons why fruit and vegetable consumption might be higher if SES is higher. Other explanations could be, for example, differences costs between healthy and unhealth food (Mackenbach et al. 2015). This research will attempt to clarify the role of fruit and vegetable intake, if it has a role, in the socioeconomic gradient to late-life depression. As visually explained in figure 1, part of the socioeconomic effect could be transmitted through fruit and vegetable consumption. This leads to the following research question that will be investigated in this paper:
What is the role of fruit and vegetable intake in the socioeconomic gradient of late-life depression?
Figure 1: Conceptual model
For fruit and vegetable intake to act as a transmission mechanism, it is important to first establish the effect of fruit and vegetable intake. Multiple studies indicated that sufficient fruit and vegetable intake reduces chances of a depression (Payne et al., 2012; Jacka et al., 2012). Therefore, it is to be expected that this also holds for late-life depression, leading to the first hypothesis:
10 The next hypothesis is then based on the research question in the sense that it predicts that fruit and vegetable intake could indeed be a potential transmission mechanism. This would imply that if the effect of fruit and vegetable intake is taken into account, the stand-alone effect of educational attainment is lower. However, since multiple other mechanisms exist through which educational attainment affects general health, it can be expected that these explanations also hold for late-life depression (Conti et al., 2010; Cutler & Lleras-Muney, 2010). The direct effect of educational attainment on late-life depression will then still be positive.
Hypothesis 2: When controlling for fruit and vegetable intake, the effect of SES on late-life depression is lower, although it is still positive.
To both fruit and vegetable consumption and late-life depression, large sex differences exist. Late-life depression is more prevalent in women (Acciai & Hardy, 2017). Moreover, evidence was found that the positive effect of fruit and vegetable intake on the occurrence of depressive symptoms is larger for women (Samieri et al., 2008). Additionally, sex differences in diet composition exist. Wardle et al. (2004) find evidence that women consume more dietary fibre, less saturated fats and more fruit and vegetables than men do. Evidence that does not seem to be confounded by cultural differences. They argue that women are more focused on weight loss and that they value healthy eating habits more than men do. All taken together, this evidence of substantial sex differences makes it almost inevitable that sex differences also exist for the potential transmission mechanism of fruit and vegetable intake. This leads to the formation of the third hypothesis.
Hypothesis 3: The transmission mechanism of fruit and vegetable intake through which SES affects late-life depression is larger for women than for men.
3. Methodology
3.1 Data
11 whether the respondent indicates that this is due to financial reasons or due to other reasons. Börsch-Supan et al. (2013) provide further information on the methodological details of SHARE.
3.1.1 Depression
Measurement of depression is based on the Euro-D scale. This scale, which was presented by Prince et al. (1999), ranks depressive symptoms on a scale from ‘0’ (not depressed) to ‘12’ (very depressed). The scale is designed to detect depression in older age specifically. The scoring was validated by clinical interviews and is based on questions about, sleep, concentration, appetite, fatigue, interest, guilt, suicidality, pessimism, depression, irritability, tearfulness and feelings of enjoyment. Scores are based on the number of symptoms experienced in the last month before the interview. The scale offers a valid comparison of depression prevalence and risk associations between different countries in Europe (Castro-Costa et al., 2008). Prince et al. (1999) find that the optimal cut-off point to predict depression with the Euro-D scale is a score between ‘3’ and ‘4’. This means that with the Euro-D scale, all scores of ‘4’ and higher are considered cases of depression. Therefore, the dependent variable depression is constructed as a binary variable where ‘1’ means a case of depression, because the scale score is a ‘4’ or higher.
3.1.2 Fruit and vegetable intake
The behavioural module of the survey contains a question where respondents indicate how often they consume fruit and vegetables. The survey question has a 5-point scale where ‘1’ means that the respondent consumes fruit or vegetables at least once a day and ‘5’ means that the respondent consumes fruit or vegetables less than once per week. Based on the findings of Lock et al. (2005) that advocate for a minimum consumption level of five servings per day, none of the scales results in an optimal consumption level. Therefore, a minimum requirement is implemented of at least one serving a day, which could be interpreted as ‘the least insufficient’. The minimum requirement we state here is a ‘1’ on the 5-point scale. Therefore, the binary variable fruit and vegetables is constructed where ‘1’ indicates a fruit and vegetable intake of at least once a day, and all other values of the scale result in a ‘0’.
3.1.3 Socioeconomic status: educational attainment
12 ‘6’ (second stage of tertiary education) indicate high education. In the model, low education is the reference category.
3.1.4 Other explanatory variables
Besides the dependent variables low education, intermediate education, high education and fruit and vegetable intake, multiple control variables are added to the model of this research. First, the demographic factors country, age and sex are included. There exist strong country differences in depression prevalence (Kok et al., 2012). Therefore, country fixed effects are added to the model. Since women are more likely to become depressed, it is important to control for sex differences (Acciai & Hardy, 2017). The variable sex is ‘1’ if the respondent is a female and ‘0’ if the respondent is a male. Age measures the age of the respondent at the time of the interview. Since depression is not linear in age, age fixed effects are included to account for cohort and age effects (Kessler et al., 1994; Lynch, 2003; World Health Organization, 2017). Second, to control for the effect of overall health and disability deterioration, found to be an important depression determinant by Kennedy et al. (1990), functional disability indicates whether respondents experienced limitations in their normal activities for health reasons over the last six months.
Third, to prevent a bias of omitted variables, the model also includes various lifestyle factors that are known to affect late-life depression; smoking and physical activity (Poulin et al., 2005; Lindwall & Larsman, 2011). The binary variable smoking measures if the respondent is a smoker. To control for physical activity, two variables are included that indicate how often respondents engage in physical activity. Moderate physical activity measures the weekly frequency of activities that require moderate energy such as walking or gardening. The variable has four scales where a ‘1’ indicates a frequency of more than once a week and a ‘4’ indicates that such activity is hardly ever conducted by the respondent. Heavy physical activity indicates how often the respondent engages in activities such as physical labour or exercising. The division of scales is the same for this variable. For both variables, the scales are added separately as binary variables.
3.2 Research method
13 sensitivity analyses to establish the robustness of the estimations and to explain the potential pathway of fruit and vegetable intake further.
3.3 Descriptive statistics
The dataset consists of 94,044 respondents, of which 25,625 respondents were excluded since they did not have an interview to answer the survey modules of the fifth wave of SHARE. All respondents who reported an age of below 50 were excluded since the focus of this research is on depression in late adulthood. Respondents that are older than 97 are excluded for statistical purposes since very few respondents reported to be of such an age. This lead to the exclusion of another 1,226 respondents. Additionally, all observations for which one or more variables contained a missing value were excluded. For the variable depression there were 2208 missing observations and for the variable smoking 12,210 missing observations. These variables are sensitive to refusal to answer since talking about these topics can make respondents uncomfortable and answers might be socially undesirable (Christelis, 2011). However, the proportion of missing values for the independent variable is less than 10%, which means it is safe to assume that this will not result in biased estimates (Bennett, 2001).
The sample that forms the basis of this analysis contains 50,332 respondents where female respondents account for 53.97% of the population and male respondents for 46.03%. The mean age of all respondents is 66.7 years old. Table 1 provides complete descriptive statistics by sex. This table also includes the results of student’s t-test of group mean comparison. The differences in means between men and women are found to be significant for all variables. Appendix A reports a correlation matrix; this matrix shows no strong correlations between variables that require attention.
Table 1: Descriptive statistics
Male (N=23,169) Female (N=27,163)
14 Interestingly, from the descriptive statistics it can already be observed that 19.52% of all men experience a depression, compared to 32.58% of all women. Additionally, it can be observed that 74.31% of all men obtain sufficient fruit and vegetable intake compared to 83.43% of all women. The group mean comparison indicates that all means are significantly different between the subgroups men and women. Figure 2 shows the prevalence of depression measured as the percentage of total sex population per country. Estonia has the highest depression rate amongst men with 29.2% and Italy has the highest depression rate amongst women with 46%. The lowest depression rate amongst men are in Switzerland with 12.3% and amongst women in the Netherlands with 20.8%. These rates roughly indicate a gradient where depression is more prevalent in southern countries and in eastern-European countries than in northern countries. This is in line with findings of Ladin (2008) and Kok et al. (2012). Additionally, figure 1 shows that 83.43% of all women consume sufficient amounts of fruit and vegetables compared to 74.31% of all men.
Figure 2: Prevalence of depression per country
4. Results
4.1 Estimation results
In table 2, the results of the logistic regressions are presented. The coefficients are transformed to average marginal effects. For intermediate education and high education, low education is the reference category. For both moderate and heavy physical activity, less than once a week is the reference scale. All variables are significant at the 1%-level, except for one country effect. However, the differences between the two subgroups of the effect of intermediate education, high education and fruit and vegetables are not significant.
The classification results show that the model correctly predicts 81.41% of the cases for men and 71.65% of the cases for women. The pseudo R-squared of the third model has a value of 0.1325 for the subgroup men and a value of 0.1120 for the subgroup women.
15
Table 2: Average marginal effects indicating determinants of late-life depression
Logistic model
(1) (2) (3)
Male Female Male Female Male Female χ2 Intermediate education -0.04541*** (0.006378) -0.07183*** (0.006612) -0.04327*** (0.006356) -0.06835*** (0.006607) -0.02433*** (0.006026) -0.04318*** (0.006369) 0.72 High education -0.08344*** (0.007370) -0.1189*** (0.007978) -0.07934*** (0.007345) -0.1124*** (0.007983) -0.03877*** (0.006940) -0.06742*** (0.007677) 1.09
Fruit and vegetables - - -0.05389*** (0.005797) -0.07863*** (0.007239) -0.03335*** (0.005541) -0.04730*** (0.007019) 0.01 Functional disability - - - - 0.1740*** (0.005022) 0.2107*** (0.005205) 12.39*** Smoking - - - - 0.02336*** (0.005795) 0.03944*** (0.006874) 0.49
Moderate physical activity
1-3 times per month - - - - -0.08467***
(0.01152) -0.07991*** (0.01274) 3.49* Once a week - - - - -0.08112*** (0.009010) -0.08062*** (0.01016) 4.15**
More than once a week - - - - -0.09362***
(0.007152)
-0.1048*** (0.008294)
3.98**
Heavy physical activity
1-3 times per month - - - - -0.06298***
(0.009613) -0.02876*** (0.01015) 12.20*** Once a week - - - - -0.06973*** (0.008486) -0.06434*** (0.008699) 4.94**
More than once a week - - - - -0.05795***
(0.006360)
-0.04359*** (0.007028)
10.66***
Age fixed effects YES YES YES YES YES YES - Country fixed effects YES YES YES YES YES YES - Observations 23,169 27,163 23,169 27,163 23,169 27,163 - Pseudo R-squared 0.0385 0.0447 0.0421 0.0480 0.1325 0.1119 - *= p<0.1, **=p<0.05, ***=p<0.01, standard errors are between brackets
From the estimations, it becomes apparent that sufficient fruit and vegetable intake reduces the chance of a depression in late-life, and that the effect of SES is smaller when accounting for fruit and vegetable intake. This indicates that fruit and vegetable intake could indeed act as a pathway through which SES affects the chances of a depression in late-life, thereby confirming hypothesis 1 and 2.
16 With respect to the control variables, the estimations indicate that being physically limited and being a smoker both increases chances of a depression. Higher frequencies of physical activity reduce the chance of a depression. For both men and women, the impact of moderate physical activity is larger than that of heavy physical activity. The effects of functional disability and physical activity do differ significantly by sex.
4.2 Intake motivation
In the SHARE survey, if fruit and vegetable intake was low, respondents were asked a follow-up question to indicate the reason behind their low fruit and vegetable intake. Here the respondents could answer that they did not consume more fruit and vegetables due to financial reasons or due to other reasons. This is the variable intake motivation. The variable takes the value ‘1’ if intake is low due to financial reasons and the value ‘0’ if intake is low due to other reasons. In this subsection, the reasons for low intake will be investigated further to find out whether the effect on late-life depression is different for the two groups of reasons. The question was only asked if fruit and vegetable intake was lower than three to six times per week. People that have an intake of three to six times per week are in this research considered to have an insufficient intake, however, they were not asked why in the survey. Although this leads to a small subsample of 2,910 respondents, further investigating motivations could still provide useful information about the fruit and vegetable pathway of SES. Descriptive statistics of the subsample are presented in table 3a by subgroup, and in total in table 3b.
Table 3a: Descriptive statistics subsample based on intake motivation by subgroup
Male (N=1,766) Female (N=1,144)
17
Table 3b: Descriptive statistics subsample based on intake motivation
(N=2,910)
Mean SD Min Max. Depression 0.3722 0.4835 0 1 Intake motivation 0.1275 0.3336 0 1 Low education 0.4375 0.4962 0 1 Intermediate education 0.3993 0.4898 0 1 High education 0.1632 0.3696 0 1 Age 65.64 9.697 50 96 Functional disability 0.5704 0.4951 0 1 Smoking 0.4100 0.4919 0 1 Moderate physical activity 1.864 1.162 1 4 Heavy physical activity 2.813 1.312 1 4
Due to the small sample and small variance in the independent variable intake motivation, the estimation is not divided in subgroups based on sex. Table 4 shows the results of a logistic regression estimation where intake motivation is the dependent variable. For comparison purposes, the other explanatory variables are the same as in the main estimation shown in section 4.1. Again, robust standard errors are computed. Three age groups were omitted because only one or two respondents reported this age, resulting in a total sample of 2,906 observations. The results are again presented as average marginal effects.
Intermediate education, smoking and the moderate physical activity category more than once a week are only significant at the 10%-level. All other variables are significant at the 1%-level. The classification table indicates that this model correctly predicts 72.33% of all cases. The pseudo R-squared is 0.1722.
18
Table 4: Average marginal effects indicating determinants of late-life depression with the independent variable intake motivation
Logistic model Dy/dx SD Intake motivation 0.1252*** (0.02441) Intermediate education -0.02888 (0.01846) High education -0.07428** (0.02505) Sex 0.1446*** (0.01619) Functional disability 0.1968*** (0.01652) Smoking 0.03525** (0.01754) Moderate physical activity
1-3 times per month -0.09894*** (0.03528)
Once a week -0.1243*** (0.02793)
More than once a week -0.09513*** (0.02381)
Heavy physical activity
1-3 times per month -0.06240** (0.02941)
Once a week -0.07505*** (0.02638)
More than once a week -0.08748*** (0.02195)
Age fixed effects YES - Country fixed effects YES - Observations 2,906 - Pseudo R-squared 0.1722 - *= p<0.1, **=p<0.05, ***=p<0.01, standard errors are between brackets
4.3 Income: an alternative measure of SES
19 are used. Real income is then divided by the square root of household size to obtain the variable income. This equivalence scale to transform household income to individual income was originally invented by Rainwater (1974) and is often used by the OECD (OECD, 2008). Data on income is available for a subsample of 35,607 respondents. Of this subsample, 577 observations were omitted because these respondents did not receive a Euro-D score. Another 5,957 observations were omitted due to missing values for one of the other variables. This leaves a subsample of 29,073 respondents. Income is divided in quartiles. Complete descriptive statistics for this subsample can be found in table 5. Appendix B provides correlations with the variable income.
Table 5: Descriptive statistics subsample based on income
Male Female
Mean SD Min Max. Mean SD Min Max. T-statistic Depression 0.1975 0.3982 0 1 0.3375 0.4729 0 1 -26.78*** Fruit and vegetables 0.7304 0.4438 0 1 0.8288 0.3767 0 1 -20.43***
Income 49,1774 5.5*107 0.04923 6.2*109 6.9*108 6.2*1010 0.06030 6.2*102 -1.249 Age 66.82 9.510 50 97 67.17 9.998 50 97 -2.846*** Functional disability 0.4290 0.4950 0 1 0.5016 0.5000 0 1 -12.25*** Smoking 0.2486 0.4322 0 1 0.1999 0.3999 0 1 9.939*** Moderate physical activity 1.530 0.9693 1 4 1.631 1.060 1 4 -8.411*** Heavy physical activity 2.463 1.343 1 4 2.710 1.328 1 4 -15.66*** Observations 12,661 16,412
Note: the t-statistic is the result of a student’s t-test of group mean comparisons, significant differences at the 10%, 5% and 1%-level are reported by respectively *, ** and ***.
The estimation in table 6 is again based on a logistic regression with robust standard errors. The reference groups are the same as in the previous model and results are presented as average marginal effects. Income is divided in quartiles; the fourth quartile is the reference category. Descriptive statistics of these quartiles can be found in appendix C. For the subgroup men, the age group ‘95’ is omitted since only one respondent reported this age.
As can be observed from table 6, all variables are found to be significant at the 1%-level, except for some age and country fixed effects. For the subgroup female, the pseudo R-squared is 0.1124, for the subgroup male this is 0.1275. This goodness-of-fit measure indicates that this model performs roughly as well as the model of the main estimation in section 4.1 in explaining total variation.
20 Table 6 shows that, when taking income as a measure for SES, fruit and vegetable intake still reduces the chance of a depression. However, although the effects of the income quartiles become smaller in model 2 compared to model 1, these differences are very small.
Table 6: Average marginal effects indicating determinants of late-life depression with income as an alternative measure of SES
Logistic model
(1) (2) (3)
Male Female Male Female Male Female χ2 Income Second quartile -0.06355*** (0.01013) -0.08382*** (0.01013) -0.06100*** (0.01010) -0.07881*** (0.01014) -0.03554*** (0.009747) -0.05368*** (0.009801) 0.05 Third quartile -0.01039*** (0.01084) -0.1196*** (0.01143) -0.09989*** (0.01079) -0.1125*** (0.01144) -0.05376*** (0.01047) -0.07152*** (0.01105) 0.04 Fourth quartile -0.1285*** (0.01139) -0.1420*** (0.01184) -0.1238*** (0.01131) -0.1350*** (0.01183) -0.06848*** (0.01075) -0.08858*** (0.01129) 0.15
Fruit and vegetables - - -0.06264*** (0.007737) -0.08578*** (0.009301) -0.03791*** (0.007505) -0.05401*** (0.008959) 0.00 Functional disability - - - - 0.1653*** (0.006811) 0.2104*** (0.006813) 2.88* Smoking - - - - 0.03932*** (0.007724) 0.04304*** (0.008905) 0.73
Moderate physical activity
1-3 times per month - - - - -0.08218***
(0.01590) -0.09947*** (0.01645) 0.32 Once a week - - - - -0.08400*** (0.01285) -0.08809*** (0.01334) 1.72
More than once a week - - - - -0.08199***
(0.01020)
-0.1159*** (0.1095)
0.00
Heavy physical activity
1-3 times per month - - - - -0.04629***
(0.01303) -0.03845*** (0.01313) 1.37 Once a week - - - - -0.05420*** (0.01141) -0.06349*** (0.01122) 0.40
More than once a week - - - - -0.05748***
(0.008555)
-0.04426*** (0.009162)
5.72**
21
5. Discussion
5.1 Discussion of findings
Depression is a prevalent health problem for Europeans in late adulthood. According to the Euro-D classification, 28.7% of the SHARE respondents in this sample is considered to have a depression. Between sexes there is a large difference in prevalence, 19.52% of all men and 32.59% of all women have a depression. The results of this research clearly indicate that fruit and vegetable intake can reduce the chances of getting a depression in late-life. In addition, it provides support for the hypothesis that fruit and vegetable intake could be a potential pathway through which SES affects late-life depression. To this potential pathway, however, this research did not provide evidence for differences by sex. The fact that insufficient fruit and vegetable intake is found to be a significant determinant of late-life depression highlights the importance for a diet strategy to prevent late-life depression, and possibly other diseases. Since sufficient fruit and vegetable intake also decreases chances of ischemic strokes, coronary heart diseases and certain types of cancer, strategies targeted at improving diet quality have many benefits and can become very cost-effective (Lock et al. 2005). The findings of this research are in line with previous findings of Poulin et al. (2005) and Lindwall & Larsman (2011) and also indicate the significant impact of other lifestyle factors; physical activity and smoking.
The analysis of intake motivation provides some useful insights that strengthen this conclusion. In the literature review, it was explained that there exist different explanations as to why socioeconomic status affects dietary intake. This research provides little support for the conclusion of Monsivais et al. (2012) and Mackenbach et al. (2015) that dietary intake is of poor quality because of financial reasons. Of the 2,910 respondents who answered the question about their intake motivation, only 18.89% indicated that financial limitations were the reason for low fruit and vegetable intake. 87.77% of the respondents indicated that there were other reasons for their insufficient intake. On average, respondents who report financial limitations for low fruit and vegetable intake are 12.52% more likely to become depressed in late-life compared to respondents who reported other reasons. An explanation could be the increased psychological stress due to financial constraints that increase the chance of a depression (Adler et al., 1994).
22 The results of this research support the hypothesis that fruit and vegetable intake could explain the socioeconomic differences to late-life depression. To decrease these disparities and reduce depression prevalence, action is needed. In the literature review, explanations were given why people with low SES do not consume enough fruit and vegetables. One of these reasons was because fruit and vegetables are too expensive (Mackenbach et al. 2015). A potential solution to increase fruit and vegetable consumption could than be to subsidize these foods. However, the results of the sensitivity analyses indicate that strategies that focus on the affordability of fruit and vegetables will probably be inefficient. Only a small percentage reported that they consumed too little fruit and vegetables due to financial reasons, and the potential mediating effect of fruit and vegetables was very small when SES was measured by real income. Another explanation is the lack of nutritional knowledge. Since educational attainment does show a mediating effect of fruit and vegetables to late-life depression, strategies to reduce depression prevalence and the associated cost burden should rather focus on education. Smit et al. (2006) emphasize in their epidemiological research on late-life depression that to ensure cost-effectiveness, preventive measures should be targeted at high-risk groups. In this case, measures should be targeted at people with low SES only and especially on women. Although this research does not find support for sex differences in the potential mediating effect of fruit and vegetables, it does indicate that women are much more likely to have a depression in late-life. With respect to people with low SES, it is known that they are a difficult target group to reach (Turrell, 2002). It will not be enough to provide them with information on the effects of a healthy diet. Their ability to process information and to transform this new information to changes in their health behaviour also depends on education (Cutler & Lleras-Muney, 2010). Therefore, it is important that further research will first investigate if health behaviour can be improved by providing more information about nutrition, or if a different approach is needed.
Since other lifestyle factors are found to be important determinant of a depression in late-life as well, a strategy to reduce depression prevalence should not target fruit and vegetable intake alone. Preferably, such a strategy focuses on a healthy lifestyle. That is, a lifestyle with sufficient plant-based nutrition, without smoking and with sufficient physical activity (Lock et al., 2005; Lindwall & Larsman, 2011; Poulin et al., 2005). Such a lifestyle could reduce the associated disease and cost burdens of respectively insufficient fruit and vegetable intake, insufficient physical activity, smoking and of late-life depression.
5.2 Limitations
There are a few limitations to this research that are worth mentioning. First, the dataset was constructed using interviews. As is the case with all interviews and surveys, this could lead to a self-reporting bias. In this particular case some variables, for example fruit and vegetables or smoking, might be sensitive to socially desirable answers (Christelis, 2011).
23 without a prospective study design, especially since loss of appetite can also be a symptom of depression, which is then likely to also reduce fruit and vegetable intake. Gomes et al. (2017) find evidence for this interrelationship between late-life depression and diet quality. Therefore, they advise to focus on promoting adequate diet quality to prevent depression on the one hand and on the other hand provide nutritional attention to individuals experiencing a depression. Moreover, an estimated 75% of all psychiatric disorders manifest already in childhood or during adolescence (Kessler et al., 2005). Effects of diet during this period are even proven transmittable to new generations due to epigenetic changes (Kaati et al., 2007). Further research on the effects of diet quality in early-life on late-life depression could be very interesting to clarify the relationship between nutritional intake and late-life depression.
Third, the second limitation is part of slightly larger limitation, the concern of endogeneity. Besides reverse causality, it is difficult to rule out omitted variable bias. However, to account for this bias lifestyle factors that are proven to affect late-life depression were included.
Fourth, the variable that measures real income is split into quartiles based on the total sample. This means that differences in income between countries are not accounted for, this could potentially mean that a very large portion of a the respondents from one country end up in one quartile. However, since this research does not further investigate country differences, for this research, the distribution of respondents over the quartiles was not of utmost relevance.
Fifth, since the survey question about fruit and vegetable intake is a scale variable, this limits the possibility to determine optimal levels of fruit and vegetable intake. However, the intention of this research was to investigate fruit and vegetable intake as a transmission mechanism through which SES affects late-life depression and for this goal, the data suffices. For further research, more detailed information on fruit and vegetable intake could be an interesting starting point.
6. Conclusion
24 that such policies should not focus on reducing financial disparities. Rather, they should focus on improving nutritional knowledge. Moreover, these policies should focus on people in late adulthood with low SES and especially on women.
25
7. References
Acciai, F., & Hardy, M. (2017). Depression in later life: A closer look at the gender gap. Social Science Research, 68(1), 163-175.
Adler, N. E., Boyce, T., Chesney, M. A., Cohen, S., Folkman, S., Kahn, R. L., & Syme, S. L. (1994). Socioeconomic status and health: the challenge of the gradient. American Psychologist, 49(1), 15.
Alexopoulos, G. S. (2005). Depression in the elderly. The Lancet, 365(9475), 1961-1970. Beekman, A. T., Geerlings, S. W., Deeg, D. J., Smit, J. H., Schoevers, R. S., de Beurs, E., ... & van Tilburg, W. (2002). The natural history of late-life depression: a 6-year prospective study in the community. Archives of General Psychiatry, 59(7), 605-611.
Bennett, D. A. (2001). How can I deal with missing data in my study?. Australian and New Zealand Journal of Public Health, 25(5), 464-469.
Beydoun, M. A., & Wang, Y. (2008a). Do nutrition knowledge and beliefs modify the
association of socio-economic factors and diet quality among US adults?. Preventive Medicine, 46(2), 145-153.
Beydoun, M. A., & Wang, Y. (2008b). How do socio-economic status, perceived economic barriers and nutritional benefits affect quality of dietary intake among US adults?. European Journal of Clinical Nutrition, 62(3), 303-313.
Black, C., Moon, G., & Baird, J. (2014). Dietary inequalities: what is the evidence for the effect of the neighbourhood food environment?. Health & Place, 27, 229-242.
Blazer, D. G. (2003). Depression in late life: review and commentary. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 58(3), M249-M265.
Börsch-Supan, A., Hank, K., & Jürges, H. (2005). A new comprehensive and international view on ageing: introducing the ‘Survey of Health, Ageing and Retirement in Europe’. European Journal of Ageing, 2(4), 245-253.
Börsch-Supan, A., Brandt, M., Hunkler, C., Kneip, T., Korbmacher, J., Malter, F., ... & Zuber, S. (2013). Data resource profile: the Survey of Health, Ageing and Retirement in Europe (SHARE). International Journal of Epidemiology, 42(4), 992-1001.
Castro‐Costa, E., Dewey, M., Stewart, R., Banerjee, S., Huppert, F., Mendonca‐Lima, C., ... & Tsolaki, M. (2008). Ascertaining late‐life depressive symptoms in Europe: an evaluation of the survey version of the EURO‐D scale in 10 nations. The SHARE project. International Journal of Methods in Psychiatric Research, 17(1), 12-29.
Chapman, D. P., & Perry, G. S. (2008). Peer reviewed: depression as a major component of public health for older adults. Preventing Chronic Disease, 5(1).
Christelis, D. (2011). Imputation of Missing Data in Waves 1 and 2 of SHARE. Naples, Italy: Centre for Studies in Economics and Finance (CSEF).
Conti, G., Heckman, J., & Urzua, S. (2010). The education-health gradient. American Economic Review, 100(2), 234-38.
Cutler, D. M., & Lleras-Muney, A. (2010). Understanding differences in health behaviors by education. Journal of Health Economics, 29(1), 1-28.
Gomes, A. P., Bierhals, I. O., Soares, A. G., Hellwig, N., Tomasi, E., Assunção, M. F., & Gonçalves, H. (2017). Interrelationship between diet quality and depressive symptoms in elderly. The Journal of Nutrition, Health & Aging, 22(3), 387-392.
Gómez-Pinilla, F. (2008). Brain foods: the effects of nutrients on brain function. Nature Reviews Neuroscience, 9(7), 568.
26 Hensley, P. L. (2006). Treatment of bereavement-related depression and traumatic grief. Journal of Affective Disorders, 92(1), 117-124.
Hulshof, K. F. A. M., Brussaard, J. H., Kruizinga, A. G., Telman, J., & Löwik, M. R. H. (2003). Socio-economic status, dietary intake and 10 y trends: the Dutch National Food Consumption Survey. European Journal of Clinical Nutrition, 57(1), 128.
Jacka, F. N., Maes, M., Pasco, J. A., Williams, L. J., & Berk, M. (2012). Nutrient intakes and the common mental disorders in women. Journal of Affective Disorders, 141(1), 79-85.
Joshipura, K. J., Ascherio, A., Manson, J. E., Stampfer, M. J., Rimm, E. B., Speizer, F. E., ... & Willett, W. C. (1999). Fruit and vegetable intake in relation to risk of ischemic stroke. Jama, 282(13), 1233-1239.
Joshipura, K. J., Hu, F. B., Manson, J. E., Stampfer, M. J., Rimm, E. B., Speizer, F. E., ... & Willett, W. C. (2001). The effect of fruit and vegetable intake on risk for coronary heart disease. Annals of Internal Medicine, 134(12), 1106-1114.
Kaati, G., Bygren, L. O., Pembrey, M., & Sjöström, M. (2007). Transgenerational response to nutrition, early life circumstances and longevity. European Journal of Human Genetics, 15(7), 784.
Karpansalo, M., Kauhanen, J., Lakka, T. A., Manninen, P., Kaplan, G. A., & Salonen, J. T. (2005). Depression and early retirement: prospective population based study in middle-aged men. Journal of Epidemiology & Community Health, 59(1), 70-74.
Kennedy, G. J., Kelman, H. R., & Thomas, C. (1990). The emergence of depressive symptoms in late life: the importance of declining health and increasing disability. Journal of Community Health, 15(2), 93-104.
Kessler, R. C., McGonagle, K. A., Nelson, C. B., Hughes, M., Swartz, M., & Blazer, D. G. (1994). Sex and depression in the National Comorbidity Survey. II: Cohort effects. Journal of Affective Disorders, 30(1), 15-26.
Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K. R., & Walters, E. E. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62(6), 593-602.
Kok, R., Avendano, M., d’Uva, T. B., & Mackenbach, J. (2012). Can reporting heterogeneity explain differences in depressive symptoms across Europe?. Social Indicators Research, 105(2), 191-210.
Krishnan, K. R. R. (2002). Biological risk factors in late life depression. Biological Psychiatry, 52(3), 185-192.
Ladin, K. (2008). Risk of Late-Life Depression Across 10 European Union Countries. Journal of Aging and Health, 20(6), 653-670.
Lindwall, M., Larsman, P., & Hagger, M. S. (2011). The reciprocal relationship between physical activity and depression in older European adults: A prospective cross-lagged panel design using SHARE data. Health Psychology, 30(4), 453.
Lock, K., Pomerleau, J., Causer, L., Altmann, D. R., & McKee, M. (2005). The global burden of disease attributable to low consumption of fruit and vegetables: implications for the global strategy on diet. Bulletin of the World Health Organization, 83(2), 100-108.
Lorant, V., Deliège, D., Eaton, W., Robert, A., Philippot, P., & Ansseau, M. (2003). Socioeconomic inequalities in depression: a meta-analysis. American Journal of Epidemiology, 157(2), 98-112.
Lynch, S. M. (2003). Cohort and life-course patterns in the relationship between education and health: A hierarchical approach. Demography, 40(2), 309-331.
27 Malter, F. and A. Börsch-Supan (Eds.) (2015). SHARE Wave 5: Innovations & Methodology. Munich: MEA, Max Planck Institute for Social Law and Social Policy.
Mathers, C., Fat, D. M., & Boerma, J. T. (2008). The Global Burden of Disease: Update 2004. Geneva, Switzerland: World Health Organization.
McKinnon, L., Giskes, K., & Turrell, G. (2014). The contribution of three components of nutrition knowledge to socio-economic differences in food purchasing choices. Public Health Nutrition, 17(8), 1814-1824.
Miech, R. A., & Hauser, R. M. (2001). Socioeconomic status and health at midlife: a comparison of educational attainment with occupation-based indicators. Annals of Epidemiology, 11(2), 75-84.
Mirowsky, J., & Ross, C. E. (1992). Age and depression. Journal of Health and Social Behavior, 33(3), 187-205.
Monsivais, P., Aggarwal, A., & Drewnowski, A. (2012). Are socio-economic disparities in diet quality explained by diet cost?. Journal of Epidemiology and Community Health, 66(6), 530-535.
Monsivais, P., Mclain, J., & Drewnowski, A. (2010). The rising disparity in the price of healthful foods: 2004–2008. Food Policy, 35(6), 514-520.
Mueller, C. W., & Parcel, T. L. (1981). Measures of socioeconomic status: Alternatives and recommendations. Child Development, 52(1), 13-30.
OECD. (2008). Growing unequal? Income distribution and poverty in OECD countries. Paris, France: OECD.
Papakostas, G. I. (2008). Tolerability of modern antidepressants. The Journal of Clinical Psychiatry, 69(E1), 8-13.
Payne, M. E., Steck, S. E., George, R. R., & Steffens, D. C. (2012). Fruit, vegetable, and antioxidant intakes are lower in older adults with depression. Journal of the Academy of Nutrition and Dietetics, 112(12), 2022-2027.
Pechey, R., & Monsivais, P. (2015). Supermarket choice, shopping behavior, socioeconomic status, and food purchases. American Journal of Preventive Medicine, 49(6), 868-877.
Pechey, R., & Monsivais, P. (2016). Socioeconomic inequalities in the healthiness of food choices: Exploring the contributions of food expenditures. Preventive Medicine, 88, 203.
Pilgrim, A., Robinson, S., Sayer, A. A., & Roberts, H. (2015). An overview of appetite decline in older people. Nursing Older People, 27(5), 29.
Poulin, C., Hand, D., Boudreau, B., & Santor, D. (2005). Gender differences in the association between substance use and elevated depressive symptoms in a general adolescent population. Addiction, 100(4), 525-535.
Prince, M. J., Reischies, F., Beekman, A. T., Fuhrer, R., Jonker, C., Kivela, S. L., ... & Van Oyen, H. (1999). Development of the EURO-D scale--a European, Union initiative to compare symptoms of depression in 14 European centres. The British Journal of Psychiatry, 174(4), 330-338.
Rainwater, L. (1974). What money buys: inequality and the social meanings of income. New York, United States: Basic Books, Inc.
Riboli, E., & Norat, T. (2003). Epidemiologic evidence of the protective effect of fruit and vegetables on cancer risk. The American Journal of Clinical Nutrition, 78(3), 559S-569S.
Rosenzweig, A., Prigerson, H., Miller, M. D., & Reynolds, C. F. (1997). Bereavement and late-life depression: grief and its complications in the elderly. Annual Review of Medicine, 48(1), 421-428.
28 Shavers, V. L. (2007). Measurement of socioeconomic status in health disparities research. Journal of the National Medical Association, 99(9), 1013.
Smit, F., Ederveen, A., Cuijpers, P., Deeg, D., & Beekman, A. (2006). Opportunities for cost-effective prevention of late-life depression: an epidemiological approach. Archives of General Psychiatry, 63(3), 290-296.
Sobocki, P., Jönsson, B., Angst, J., & Rehnberg, C. (2006). Cost of depression in Europe. The Journal of Mental Health Policy and Economics, 9(2), 87-98.
Turrell, G., Hewitt, B., Patterson, C., Oldenburg, B., & Gould, T. (2002). Socioeconomic differences in food purchasing behaviour and suggested implications for diet‐ related health promotion. Journal of Human Nutrition and Dietetics, 15(5), 355-364.
Turrell, G., & Kavanagh, A. M. (2006). Socio-economic pathways to diet: modelling the association between socio-economic position and food purchasing behaviour. Public Health Nutrition, 9(3), 375-383.
Van der Kooy, K., van Hout, H., Marwijk, H., Marten, H., Stehouwer, C., & Beekman, A. (2007). Depression and the risk for cardiovascular diseases: Systematic review and meta-analysis. International Journal of Geriatric Psychiatry, 22(7), 613-626.
Walsh, R. (2011). Lifestyle and mental health. American Psychologist, 66(7), 579.
Wardle, J., Haase, A. M., Steptoe, A., Nillapun, M., Jonwutiwes, K., & Bellisie, F. (2004). Gender differences in food choice: the contribution of health beliefs and dieting. Annals of Behavioral Medicine, 27(2), 107-116.
World Health Organization. (2003). Diet, nutrition, and the prevention of chronic diseases: report of a joint WHO/FAO expert consultation (WHO Technical Report Series 916). Geneva, Switzerland: World Health Organization.
