Depression in middle and older adulthood: the role of immigration, nutrition, and other determinants of health in the Canadian longitudinal study on aging

The sample of this study included the baseline comprehensive set of participants (n = 30,097) of the Canadian Longitudinal Study on Aging (CLSA). The CLSA is a prospective study that includes biological, medical, psychological, social, lifestyle, and economic measures of Canadians aged 45 to 85 years old who will be followed for 20 years [73]. Participants in the comprehensive sample were administered in-home interviews and a wide range of physical assessment data was collected from them at dedicated data collection sites [74]. Those who did not have their anthropometrics assessed were excluded (n = 2935), yielding a final sample size of 27,162. Further details about the study protocol can be found at: https://​www.​clsa-elcv.​ca.

All analyses were completed using SPSS Version 22 [87]. As a modification of the model-based approach, normalized trimmed inflation weights were applied to provide estimates that are representative of the national population. The standardized weights were derived by dividing the trimmed inflation weight of each unit used in the analysis by the (unweighted) average of the survey weights of all the analyzed units.

Descriptive information and inferential statistics were generated by Chi-square tests using weighted means. Adjusted odds ratios (aOR) were derived from binary logistic regression to examine associations between immigrant status and depression while adjusting for the covariates. Six stepwise models were analyzed by entering demographic and socioeconomic variables (Model 1), physical health measures (Model 2), health behavior variables (Model 3), over-nutrition and poor nutrition status indicators (Models 4 and 5), and dietary intake measures (Model 6) respectively. The final model was adjusted for all aforementioned variables (Model 7). The interaction variable of sex by immigration status was significant and therefore separate logistic regression analyses were conducted for women and men.

Table 1 presents details about the proportion of men and women in the sample who were experiencing depressive symptoms. Overall, the sample mainly consisted of Canadian born residents (n = 22,423, 82.6%), 56–65 years (n = 16,008, 58.9%), earning between C$50,000–$99,999 annually (n = 9031, 33.2%), in a married or common-law relationship (n = 18,859, 69.4%), who had earned a post-secondary diploma or degree (n = 21,115, 77.7%). Among the demographic, social, and economic factors, significant associations were found for men and women by income, relationship status, and education level categories (p’s < 0.001). For the age, significant association was only found among men (p < 0.001).

In terms of health-related characteristics, the majority had at least one health condition (n = 22,297, 82.1%), some type of hypertension or taking anti-hypertensive medication (n = 17,285, 63.6%), and reported having no chronic pain (n = 20,569, 75.7%). Most respondents did not drink excess alcohol (n = 17,552, 64.6%) and did limited physical activity (n = 23,264, 85.6%). There were significant associations between all physical health and health behavior characteristics (p’s < 0.05–0.001). For nutrition status indicators, there tended to be a higher proportion of the sample having characteristics that were indicative of over-nutrition. All body composition measures indicated that the majority tended to have excess weight based on BMI measures (n = 18,893, 69.7%), high visceral adiposity based on WHR (n = 17,898, 65.9%), high body fat percent (n = 16,752, 61.7%), and elevated disease risk based on BMI and WHR measures (n = 19,198, 70.8%). There were significant associations for all over-nutrition indications (p’s < 0.05–0.001). While there were significant associations found for indicators of poor nutrition that included grip strength (p < 0.001) and nutritional risk (p < 0.001), most of the sample did not have the characteristics that would be indicative of under-nutrition. For men,

however, there was a high proportion who had sarcopenia (n = 11,282, 41.5% of the total sample) (χ²(1) = 57.4, p < 0.001). For men, significant association were also found for the anemia screen (χ²(2) = 15.6, p < 0.001) measurement, however, most of the men screened negative (n = 10,448, 38.5% of the total sample).

For both men and women, most of the 12 dietary intake measures were significantly associated with depression. Overall, the sample tended to have low intakes of fiber (≤2 servings/day; n = 21,617, 79.6%), however there was only a significant association found for women (χ² = 21.8 (3), p < 0.001). Most of the sample also had low intakes of pulses and nuts (≤1 serving/day; n = 15,027, 55.3%) as well as fruits and vegetables (≤4 servings/day; n = 17,193, 63.3%) and there were significant associations found for both of these dietary intake measures (p’s < 0.001). For the intakes of calcium containing foods, most of the sample consumed low amounts of those with high vitamin D content (≤2 servings/day; n = 18,345, 67.5%), and a significant association was only found for men (χ²(3) = 19.4, p < 0.001). For sources with low vitamin D content, most of the sample consumed some level of these (n = 22,194, 81.7%) with significant associations indicated (p’s < 0.001). For dietary sources of the omega-3 fatty acids, most consumed fish (n = 24,891, 91.6%, p’s < 0.001); few consumed omega-3 eggs (n = 20,058, 73.8%) which only showed significant association for women (χ²(1) = 9.6, p < 0.05). About half of the sample on average consumed four or more fat sources (n = 13,543, 49.9%). There was a significant association between number of fat sources consumed daily and depression among men (χ²(3) = 18.1, p < 0.001), but not among women. More than two-thirds of the sample’s average daily intakes of pure fruit juice, salty snacks, and pastries were less than once daily and of these pure fruit juice (χ²(3) = 12.6, p < 0.05) and pastries (χ² (2) = 7.75, p < 0.05) showed significant association for women. On a weekly basis, most of the sample on average consumed less than two-thirds of a chocolate bar (n = 25,938, 95.5%) and significant association was indicated (p’s < 0.001). The immigrant sample was categorized according to years since immigration with those who had lived in Canada less than 20 years considered as recent and mid-term immigrants. Those who had resided in Canada 20 or more years were considered as long-term immigrants. On average, the recent and mid-term immigrant group had resided in Canada almost 12 years (Mean = 11.8, SD ± 4.6). The average years since immigration for females (Mean = 12.3, SD ± 4.5) was slightly higher than males (Mean = 11.5, SD ± 4.7). Mean years since immigration for the long-term group was 47 years (Mean = 47.1, SD ± 12.1) with the average being slightly higher for males (Mean = 47.3, SD ± 12.1) compared to females (Mean = 46.9, SD ± 12.2).

Among women, immigrant status was associated with depression. This finding is consistent with longitudinal investigations of new immigrants which show increases in reported sadness and loneliness [29], but is inconsistent with cross-sectional studies that report lower levels of depressive symptoms when compared to native born Canadians [25‐28]. Differences in the results may be attributed to the older age of this sample (45–85 years old) and how depression was measured and what potential confounders were accounted for. The immigrant women in this study may have reported depression as a result of the substantial stress associated with settling in a new country such as having insufficient income [40], overcoming language barriers [41, 42], facing discrimination, adapting to a different culture [43], reduced social support networks [44], and having their education and work experiences unrecognized [45].

Contrary to the assumption of a health advantage of immigrants, our results showed no association with depression for men and found that middle-aged and older immigrant women are more likely to screen positive for depression compared to those who were born in Canada. The association of heightened depression was not attenuated when 32 covariates including socioeconomic status, health behaviors, nutrition status, and dietary habits were taken into account, suggesting that these may not be the main determinants of depression among immigrants. One possible explanation is that based on cumulative advantage-disadvantage theory [88‐90], exposure to migration and acculturation may contribute to stress over the life course and poorer mental health status despite having similar socioeconomic status and lifestyles as middle-aged and older non-immigrants.

Our results showed that middle-aged and older immigrant women who lived in Canada less than 20 years have a higher likelihood of depression compared to both long-term immigrant women and native-born Canadian women. Similar findings were also indicated in a longitudinal study where deterioration of self-rated health occurred among females and ethnic minority immigrants over 4 years after arriving in Canada [68]. Our study also suggested that the trajectory of deterioration in mental health among middle-aged and older immigrants might not be a linear relationship. As suggested by Wu and colleagues [28], a reverse U shape association may occur between length of stay and depression among immigrants. This would be consistent with the disillusionment model which suggests that newly arrived immigrants may experience “euphoria of arrival” and have similar or better health status than native-born residents. Subsequently, they enter a period of disillusionment and experience difficulties in the receiving country, and then eventually, adapt to the new environment [91, 92]. Further study examining longitudinal associations will better ascertain these relationships.

Similar to previous research [14‐18], the consumption of fruits and vegetables was protective for depression. Various anti-inflammatory and anti-oxidant components in fruits and vegetables may account for this relationship. Several minerals and vitamins (e.g., magnesium, zinc, selenium) present in fruits and vegetables may reduce plasma concentrations of C-reactive protein, a marker of low-grade inflammation associated with depression [93]. Antioxidants, such as vitamin C, vitamin E, and folic acid, are involved in endothelial cell signaling cascades which reduce the effects of oxidative stress on mental health [94, 95].

Interestingly, in this study chocolate intake was associated with depression. Cocoa, which is found in chocolate, has polyphenolic compounds that have been reported to modulate mental health [96] due to their roles in metal ion (e.g., iron, copper) chelation, modulation of antioxidant enzyme antioxidant ability, and anti-inflammation [97]. The reported effects of cocoa with mental health conditions such as depression are still equivocal [98]. It is also important to note that the chocolate assessed in this study referred to all types including milk and dark chocolate; the latter has 2–3 times more of the beneficial flavanol-rich cocoa. Furthermore, the relationships between chocolate consumption and depression may be impacted by whether it is consumed mindfully or not [99]. Finally, the cross-sectional nature of this study prohibits the determination of causality. Some individuals crave chocolate and its self-soothing qualities when they are depressed [100].

For men, high fiber intakes were associated with depression. High fiber foods contain phytates which bind with trace minerals such as iron, zinc, and manganese and reduces their bioavailability [101]. With increased high fiber intakes less of these nutrients, which are critical for mental health, may be absorbed which can subsequently contribute to depression symptoms [102]. Dietary fiber intake can alter the human gastrointestinal (GI) microbiota, and GI microbe composition is associated with mood in adults. Recent research has suggested that these relationships differ by sex [103].

Surprisingly, higher fat intakes tended to be protective for depression in men. This finding, which seems counterintuitive, may be due to the fact that the measure of fat included all fatty acid types and did not partition them into saturated and unsaturated fat; each type has different relationships with depression [104]. Recent exploratory research suggests that the absorption of bile acid may be impaired in chronic stress [105‐107] and that intakes of high-fat diets under stress may induce cholesterol metabolism and help to attenuate stress-related outcomes [108].

Similar to previous findings [104], intakes of omega-3 polyunsaturated fats were inversely associated with depression, however this relationship was only shown for men. This appears to be the first study to suggest the relationship between omega-3 fatty acid intake and depression may differ by gender. However, previous investigations have found sex differences for depression outcomes based on the ratio of polyunsaturated fat to saturated fat consumed [109]. Increased omega-3 fatty acid concentration in the diet may influence central nervous system cell membrane fluidity, and phospholipid composition, which may alter the structure and function of the embedded proteins and affect serotonin and dopamine neurotransmission [110].

The disproportionate associations of fruit juice and salty snack intakes with depression found in women may be due to their heightened susceptibility to factors that promote inflammation [111]. The high content of naturally occurring sugars in 100% fruit juice may cause negative health effects similar to those of other sugar-sweetened beverages [112]. Some studies have indicated that sodium reabsorption may be associated with psychological activity. Specifically, the inhibition of renal sodium reabsorption is compensated for by stimulation of the salt appetite and vice versa [111]. Exposure to stress can activate the hypothalamic-pituitary-adrenal (HPA) axis leading to release of corticosteroids which bind to two types of receptors in the brain: the mineralocorticoid receptor and the glucocorticoid receptor. Individuals with low mineralocorticoid receptor function may be more susceptible to depression [113].

Consistent with previous research, our results support that among middle-aged and older adults, those who are women, working-age (45–65 years), married, have lower education, and have less annual income are more likely to experience depression. Previous research has consistently shown that being female and being unmarried are associated with more depressive symptoms [114, 115], while research findings about the interaction between sex and age on depressive symptoms remained inconsistent [116, 117]. The association of depression with lower socioeconomic status is consistent across both cross-sectional investigations [3, 118] and longitudinal research of older adults in the Netherlands [119] and examination of depression and education attainment over the life course in the US [120]. Our study provides further evidence of the gradient effect between income level and depression.

In addition to demographic, social, and economic characteristics, associations between physical health factors and depression were also supported by previous findings. Consistent with a meta-analysis focusing on older adults, presence of chronic disease is an independent risk factor of depression [121]. Longitudinal studies of older adults also indicate that experiencing pain is associated with depression [122]. Having other physical conditions, such as hypertension, are also associated with depression among older adults [123].

With the exception of the association between smoking and depression in men, our findings of no other significant associations between depression and health behaviors were not consistent with previous studies [124‐126]. Other investigations have indicated that the associations found between smoking and depression and between smoking and heavy drinking are bidirectional [127]. The different results may be due to the variability in measures of substance use that were administered. The lack of association found between physical activity and depression may be due to confounding by factors such as presence of a health condition or having an insufficient sample who were really active [128].

The relationships between depression and nutrition status indicators were both similar and dissimilar to other studies. Like other investigations, our bivariate analyses showed association between excess body weight and depression. However, the multivariable logistic regression results did not show this relationship. This may suggest that the association was attenuated by other factors such as presence of health conditions, or, as indicated by other studies, the relationship between BMI and depression in older adults may be curvilinear [129]. Similar to previous studies, measures of poor nutrition status were associated with depression in older adults [130]. Poor nutrition is closely related to frailty, which can impact overall physical and psychological functioning [131].

Generally speaking, for women, more factors tended to be associated with depression. This may be due to women having heightened susceptibility to inflammation and autoimmune responses [132, 133] which can elevate risk for depression [134, 135]. For women, inflammation tends to affect their feelings of social disconnection to a greater extent than men [136]. In addition, factors such as relationship distress and obesity, which are both associated with depression risk, tend to have greater association with inflammation for women than for men [137]. Women experience several risk factors for inflammation at higher rates than men, including low physical inactivity and childhood adversity [105, 111, 138]. The fluctuation of women’s reproductive hormones throughout the life span, also has implications for inflammation and depression. For example, during the menopausal transition, women experience pronounced hormonal fluctuations which result in diminished estrogen levels following menopause [132, 133] and elevated levels of inflammation [134, 135]. Collectively, these findings suggest that women’s susceptibility to inflammation can contribute to the observed sex differences for depression.

To the best of our knowledge, this is the first Canadian study to comprehensively assess associations between depression and various health determinants, including immigration status and nutritional intake. However, the interpretation of the results should be viewed with caution. Given that the data was cross-sectional, the direction of the relationships between depression and the various measures cannot be ascertained. There are many patho-physiological processes that are involved in depression and thus it was impossible to account for all variables that may affect the relationships between depression and different health determinants. The use of self-report measures for many of the variables made misreporting and misclassification possible. This may be particularly the case for the lifetime measure used for smoking status and for the physical activity measurement which only assessed activity in the previous 7 days. Unfortunately, due to limited sample size of relatively recent and mid-term immigrants, larger cohorts of immigrants were analyzed (i.e., < 20 years and ≥20 years) and therefore relationships within more defined sub-groups could not be examined. Suggestions for future investigations include oversampling different immigrant groups to enable sub-population analysis, examining longitudinal data to explore relationship trajectories between depression and health determinants post-settlement, and conducting qualitative work to gain important insights into the quantitative findings.