Gender differences in the longitudinal association between husbands' and wives' depressive symptoms among Korean older adults: the moderating effects of the spousal relationship

The mutual effects of depressive symptoms between couples have long been reported; however, it remains unknown whether the spousal concordance in depressive symptoms differs depending on spousal relationships. Data on 291 married couples from the Korean Social Life, Health, and Aging Project (KSHAP) were examined. The KSHAP collected global network data from the target population living in one Korean village over eight years and across five waves. A seemingly unrelated regression (SUR) model in the panel data was employed to address correlations and heterogeneity. If one spouse (husband or wife) had depressive symptoms, the other spouse tended to have depressive symptoms. However, the effect of marital relations on spousal concordance in depressive symptoms was different among husbands and wives. This study demonstrated both spousal support and spousal network aspects of spousal relationships. Depression concordance was stronger for couples with more negative marital relationship. A supportive marital relationship was associated with less concordance between spouses’ depressive symptoms for wives but not for husbands. Spousal network overlap was associated with less depression concordance for husbands; however, for wives, spousal network overlap was directly associated with more depressive symptoms and did not mediate the association with depression concordance. Our findings suggest that approaches to supporting older adults dealing with mental health disorders may involve support at both the individual and couple levels. Gender-specific strategies could also be devised to improve the mental well-being of the older population.


Introduction
Depressive symptoms are more common in older people than in the general population, and the prevalence of depression is increasing as older populations increase worldwide. In Korea, 20% of older people experience mild depressive symptoms, and 5-6% suffer from major depressive disorder [1,2]. In other countries, community-based mental health studies report a similar prevalence of mild depressive symptoms, varying between 10 and 30%. A longitudinal study found an increase in depressive symptoms, particularly among older adults [3]. Gender differences also affect the rate of depressive symptoms; older women are approximately twice as likely as men to develop major depression.
In research on the social context of depressive symptoms, attention has been given to spousal concordance. The concordance in depressive symptoms between spouses is an emotional contagion effect, indicating that depression in one spouse affects the other's depression [4][5][6]. A study of 520 Korean adult couples aged 40 or over revealed that the probability of being depressed was more than five times higher in older adults with a depressed spouse than in those without a depressed spouse [7]. A spouse's feelings inevitably influence the other's feelings [8][9][10][11] in that a husband and wife have a dependent relationship and are frequently concerned with each other's response, mood, and emotional status [12]. A spouse's depressive symptoms of demotivation, negative feelings, and emotional difficulties are negatively associated with the mental health of the other spouse [13,14]. Few longitudinal studies have been conducted, but they have consistently indicated that the concordance between husbands' and wives' depressive symptoms was positively related over time [15][16][17].
However, little is known about which couples are most likely to experience depression contagion. As a significant moderator, marital relationships may provide crucial information for understanding the developmental process of depression concordance between couples [12]. There are two different kinds of empirical evidence for the impact of marital quality on the strength and direction of interspousal concordance in depression. Some studies found that a spouse's depressive symptoms contributed to changes in the other spouse's depressive symptoms, and these findings were more robust for couples with closer relationships [11,18]. These results indicated that the emotional bond between husbands and wives strengthens depressive contagion. Other studies, however, revealed a dynamic association between husbands' depressive symptoms and their wives' depression in maritally distressed couples but not in those who were maritally satisfied. High marital quality older adults serve as a buffer from developing depressive symptoms even if one spouse was depressed [17,19]. Overall, uncertainty remains about the moderating effect of marital relationships on the association between spouses' depressive symptoms.
In this study, the determinants of marital relationships were focused on social support aspects, both supportive and negative, and social network aspects, including spousal network overlap at the couple level [20]. Social support encompasses different aspects of social networks, in which tangible and intangible goods are exchanged within social relationships [21,22]. Social support involves subjective perceptions about satisfaction with the level of support. A social network is described in objective structural terms as a web of social relationships surrounding individuals that includes factors such as size, density, and frequency of contact.
In previous studies, spousal support usually refers to spousal relationships and directly contributes to mental health [23][24][25]. Not only the positive but also the negative aspects of spousal relationships are related to mental health difficulties, including depression. Although husbands and wives with high marital quality have a lower risk of depression, the effects for have been found to be greater for wives than for husbands [26][27][28]. In a meta-analytic review of 93 cross-sectional and longitudinal studies, Proulx et al. [26] revealed that the association between marital quality and personal well-being was stronger for women than for men. It is suggested that because women are more sensitive to relational problems and have more interdependence in personal relations, marital quality has greater mental health consequences for wives than for husbands [29,30]. For this reason, the moderating effects of spousal support on depression concordance may be stronger for women.
Spousal network overlap is the degree to which the members of each spouse's individual social networks are acquainted and interact with the other spouse (illustrated in Fig. 1). Several studies have suggested that as marital relationships last longer, network overlap, or linking to a partner's network, would increase over time [31][32][33]. A high degree of spousal network overlap reflects the tendency for spouses to perform activities together or for the same action to be carried out by a spouse simultaneously. In contrast, a low degree of interconnectedness indicates that spouses perform separate tasks. Spousal network overlap also includes the social support that flows within couples and the indicator of reliability between couples [20,21,34].
There are no previous studies on the effect of spousal network overlap on spousal concordance in mental health; however, we can infer an association from social network studies. A wide body of literature on mental health indicates that men and women approach social networks differently. In older age, women maintain larger, more diverse, and more emotionally supportive networks than men [35], which means that such networks contribute to life satisfaction more for women than for men [36]. On the other hand, men often report having a high proportion of family members in their networks, and marriage plays a more pivotal role for men as they tend to rely on their wives. Likewise, men's mental health is more sensitive to changes in spouse relations, including losing a spouse [37]. Spousal network overlap indicates that couples share their personal networks and are embedded in each other's social relationships. Given the association between spousal networks and mental health, this study argues that the moderating effects of spousal network overlap on depression concordance among couples need to be tested empirically and that the effects could be gender specific.
The present study advances research on depression concordance among spouses by using the seemingly unrelated regression (SUR) model. The SUR model simultaneously examines the effect of spousal depressive symptoms and spousal relationships on respondents' depressive symptoms together with the effect of respondents' depressive symptoms and spousal relationships on spousal depressive symptoms. The SUR model proposed by Zellner [38] uses the correlation between disturbances in the equations for husbands and wives to improve the efficiency of joint estimation. Such a model has already been used extensively in research on economics [39,40], energy consumption [41], and public health [42][43][44] to date. However, only a few studies use the SUR model in the study of spousal concordance in depressive symptoms [45,46], cognitive function [46], and healthrelated behaviors [47], and we do not find any panel SUR model for spousal concordance.
We would expect heterogeneity issues from the depressive symptoms of husbands and wives within a given couple to be related in unobserved ways. Married couples share a variety of resources and environments, such as their residential environment, family income, and children, many of which may be exogenous risk factors for depressive symptoms. For example, married couples live in the same place, have contact with the same neighborhood, and are exposed to the same stressful environment in general. It is difficult to confirm that these factors truly exist in the data; however, we need to consider exogenous risk factors to reveal spousal concordance in depressive symptoms.
This study assessed the interspousal concordance in depression and the moderating effect of marital relationships on depression concordance among Korean older adults based on an 8-year longitudinal study. At the couple network level, spousal network overlap was considered for one of the marital relationships. To take advantage of the SUR model and panel data, we attempted to provide conclusions about the role of marital relationships in depression concordance between spouses. This study is important because even though an older adult's depressive symptoms are highly interconnected with those of a spouse, the process through which they are connected has not been well studied.

Study design and study population
Data from the Korean Social Life, Health, and Aging Project (KSHAP) were used. KSHAP data were collected across five waves: wave 1 (2011), wave 2 (2012), wave 3 (2014-2015), wave 4 (2015-2016), and wave 5 (2018-2019). Initiated in 2011, the KSHAP was designed to examine the entire population of adults 60 years old or older and their spouses in Township K in South Korea [48]. Township K is a typical rural community in Korea. The KSHAP involved completing a face-to-face survey with 814 adults out of the 860 target residents at baseline, with a response rate of 94.7%. In the KSHAP, a follow-up survey was also conducted with 710 out of 755 people in wave 2 (94% response rate), 591 out of 707 in wave 3 (83.6% response rate), 573 out of 593 in wave 4 (96.6% response rate), and 518 out of 578 in wave 5 (89.6% response rate). Twenty-five participants joined after the baseline, yielding a total of 839 people (316 couples) examined during all the study periods.
Our research targeted adults identified in the first, third, fourth, and fifth waves of the KSHAP panel data in which the CES-D questionnaire was administered. Analyses for this study were restricted to 291 of a total of 316 couples after excluding respondents with insufficient information regarding spousal panel data (n = 14), the marital relationship (n = 4), annual household income (n = 4), and network members (none were indicated; n = 3). The analysis included data for 291 of the 316 couples.

Depressive symptoms
Depressive symptoms were measured with the Center for Epidemiologic Studies Depression (CES-D) Scale [49]. The Korean version of the CES-D from Cho and Kim (1993) was utilized. The CES-D consisted of 20 items with a four-point scale ranging from 0 to 3 for each item (0 = rarely or none of the time, 1 = some or little of the time, 2 = moderately or much of the time, 3 = most or almost all the time). Scores ranged from 0 to 60, with high scores indicating greater depressive symptoms. A respondent's depressive symptoms were measured with continuous CES-D scores, and their spouses' depressive symptoms were dichotomized using a cutoff point to screen for an indication of mild depressive symptomatology. This study used the CES-D cutoff score of 16, which has been validated by the Diagnostic and Statistical Manual of Mental Disorders [50].

Spousal support
Spousal support was identified using the questions from Schuster et al. [51] and assessed by two subscales: supportive relations and negative relations. The first item asked, "How often can you open up to your spouse/partner if you need to talk about your worries?" and "How often can you rely on your spouse/partner for help if you have a problem?" The second item asked, "How often does your spouse/partner make too many demands on you?" and "How often does your spouse/partner criticize you?" Each question was answered by selecting one of four response options on a scale: never, hardly ever, sometimes, and often. A supportive or negative spousal relation index was standardized on the scale to a mean of 0 and a standard deviation of 1.

Network overlap proportion
To collect social network data, the KSHAP adopted an approach developed by social network researchers that involves using questions called name generators to permit the respondent to enumerate relevant social network members [52,53]. The KSHAP asked respondents to list people with whom they had discussed important matters over the last 12 months (up to five) and a spouse, if any (up to six members in total) [48,54]. This name generator is used to collect names of close, frequently met, and long-term contacts who are thought particularly important by older adults [35]. In addition to the relationships between a respondent and their social network members, the KSHAP surveyed the relationships among their social network members. Respondents indicated the frequency of communication between social network members, including with a spouse. The frequency was reported on a 9-point scale ranging from 0 (have never spoken to each other) to 8 (every day).
To construct a complete Township K network based on the respondents, the KSHAP identified the same social network members appearing in more than one social network of different respondents (i.e., duplicates). Based on the respondent's report, the KSHAP collected detailed information about social network members, including their names, genders, ages, and addresses in the smallest administrative unit, the Ri. The KSHAP assumed that two social network members were the same person if they satisfied the following four criteria: (1) at least two out of three Korean characters in their names matched, (2) their gender was the same, (3) their age difference was less than five years, and (4) their addresses were in the same Ri (see Youm et al. [48]). In a global network, the presence of the same network members among respondents can be identified.
Spousal network overlap was measured by two indicators: self-response social network (ego-centric network) data and complete network data. The self-response social network comprised the frequency of communication between a spouse and each social network member. A spouse was assumed to be connected to one of the social network members if a respondent reported communication between the spouse and the social network member at least once a week, and the number of all existing overlapping network members with a spouse was counted. The complete network was constructed by matching the number of members in the respondent's and the spouse's social networks. The number of duplicate members appearing in both networks was counted as spousal network overlap. The spousal network overlap variable ranged from 0 to 5. To control for the respondent's network size effect, the proportion of spousal network overlap was used. The number of members in both spousal networks (overlap) was divided by the social network size. For the analysis, the network overlap proportion was categorized as less than 0.5 and 0.5 or greater.

Covariates
Sociodemographic covariates included age (in years), education level (below middle school and middle school or higher), and annual household income (below the median and above the median). Annual household income was recorded in Korean won and then dichotomized at the median. Alcohol consumption was categorized as never, rarely, or once a week or more; three categories were collapsed into two (never or rarely vs. once a week or more). The total number of comorbidities was calculated to include hypertension, hyperlipidemia, diabetes, osteoporosis, and cancer; this was a continuous variable ranging from 0 to 5. Cognitive impairment was assessed using the Korean version of the Mini-Mental State Examination for Dementia Screening, and the score ranged from 0 to 30. Cognitive impairment was dichotomized using a cutoff point to screen for dementia. This was risk-adjusted following a previous standardization study in Korea [55]. The amount of social activity was a continuous variable ranging from 0 to 3 (higher values for less social activity), based on the types of activity in which the participants mainly engaged: (1) gathering with relatives or friends, (2) participating in senior center programs, or (3) attending church or religious services.

Data analysis
We used a t-test or chi-square test to compare differences in sociodemographics, depressive symptoms, comorbidities, and marital relationships between husbands and wives. Further analysis was performed using a randomeffects SUR model [38]. The SUR model simultaneously estimates the effect of spousal depressive symptoms on the respondent's depressive symptoms, together with the effects of the respondent's depressive symptoms on their spouse's symptoms. These estimates consider unmeasured and unobserved factors common to husbands and wives that are likely to affect depressive symptoms. In other words, SUR takes into account the contemporaneous correlation between the residuals of the regression equations for husbands' and wives' depression symptoms [45]. This approach improves the efficiency of the regression estimates. The SUR model in the context of unbalanced panel data used random-effects estimators. The panel data used in this study were not balanced, as husbands and wives were not surveyed in all 4 waves; they were represented in an average of 2 waves. The random-effects SUR model using unbalanced panel data was developed initially by Biørn [56], and the equation models are summarized in the supplementary material.
This approach allowed for correlation between the unobservable components of spouses' depressive symptoms. All analyses were conducted separately by gender. The main effects of spouses' depressive symptoms and three marital relationships were investigated, and the effects of the interaction between a spouse's depressive symptoms and their marital relationship were examined. Statistical analyses were carried out using Stata 15.0.

Results
Differences by gender in all study variables are presented in Table 1. The mean (± standard deviation) CES-D scores were 9.3 ± 7.6 SD for husbands and 10.4 ± 7.8 SD for wives. Based on a CES-D ≥ 16 (criteria for clinically depressive symptoms), 18.2% of husbands and 22.3% of wives had mild depressive symptoms. This difference between husbands and wives was significant. Husbands were older and were more likely to have at least a middle school education, to consume alcohol, and to have cognitive impairment than wives were. Husbands also had more supportive relations and higher network overlap than wives. Wives participated less actively in social activities and had more negative relations than husbands did. We found significant differences between husbands and wives for all the study variables.
The results from the panel SUR estimation of the model are shown in Table 2. In the adjusted model, spouses' depressive symptoms significantly increased a respondent's CES-D scores for both husbands (β = 5.11, p < 0.001) and wives (β = 6.74, p < 0.001). Regarding sociodemographic factors, a greater extent of nonparticipation in social activities was associated with higher CES-D scores for both husbands (β = 0.68, p = 0.016) and wives (β = 1.11, p < 0.001). Husbands and wives with household income below the median level were more likely to have high CES-D scores (β = 1.96, p < 0.001 for husbands, β = 1.30, p = 0.028 for wives). Husbands educated below middle school were more likely to have high CES-D scores than those with a middle school or higher education level (β = 2.28, p < 0.001). However, age was associated with an increase in CES-D scores only for wives (β = 0.13, p = 0.002). Regarding health status, cognitive impairment was associated with elevated CES-D scores for both husbands (β = 3.38, p < 0.001) and wives (β = 2.73, p = 0.005). For husbands only, a higher number of comorbidities was associated with elevated CES-D levels (β = 0.77, p = 0.003). Figure 2 illustrates the interaction effects of each of the three spousal relationships and spousal depressive symptoms on a respondent's CES-D scores. The adjusted values of husbands' and wives' CES-D scores were the predicted values from the panel SUR model containing all covariates. The results are presented in Supplementary Tables S1-S3. A supportive relation was associated with a decrease in the effect of husbands' depressive symptoms on wives' CES-D scores; however, this interaction effect was significant only for wives. As Fig. 2b shows, the gap between the straight lines changed from wide (27.9 vs 14.1 at the lowest level of supportive relations) to narrow (11.6 vs 7.0 at the highest level of supportive relations), indicating that having supportive relations decreased spousal concordance in the depressive symptoms of a wife living with a depressed husband. A negative relation was associated with a greater increase in the effects of husbands' depressive symptoms on wives' CES-D scores, and vice versa. At the highest level of negative relations in Fig. 2c and d, the adjusted values of CES-D scores were 1.7 times higher for husbands and 2 times higher for wives when living with a depressed spouse than when living with spouses who were not depressed (16.6 vs 9.8 for husbands and 22.4 vs 11.5 for wives, respectively).
Last, as Fig. 2e shows, there was no difference between the adjusted median values of a husband's CES-D scores by the proportion of network overlap when a husband lived with a nondepressed wife. However, when a husband lived with a depressed wife, the proportion of high network overlap decreased 5 points of the adjusted values of the husband's CES-D scores (15.0 vs. 9.9). In Fig. 2f, there was no interaction effect of spousal network overlap and husbands' depressive symptoms on wives' CES-D scores. However, the proportion of network overlap was directly associated with higher CES-D scores for wives in Model 2 of Table S3 (β = 1.08, p = 0.055). The results indicated the genderspecific effects of spousal network overlap on depressive symptoms.

Discussion
This study analyzed the longitudinal relations among a spouse's depressive symptoms, three types of marital relationships, and the depressive symptoms of older South Korean adults based on KSHAP population data. This study has several major findings as follows.
Having a depressed spouse was a significant risk factor for increasing an individual's depression level regardless of gender. This finding is consistent with preceding research demonstrating that emotional contagion appears to play a decisive role in spousal similarity concerning psychological distress [15][16][17]. A spouse's emotional state inevitably influences that of their spouse [8][9][10][11], as a husband and wife have a dependent relationship and are frequently concerned about the other's responses, mood, and emotional status [12]. A relationship with a depressed spouse may become more imbalanced and negative, which may cause depression in older adults and poor mental health [6,13,14,57].
The association between husbands' and wives' depressive symptoms was stronger for couples that reported a low level of supportive marital relations, but solely for the wife, and those that reported a high level of negative marital relations for both the husband and wife. There are several possible explanations for why the concordance in depressive symptoms depends on the quality of the marital relationship. First, marital satisfaction eases the burdens of care for a spouse. People with depressive symptoms can experience a persistent feeling of sadness and act in ways that are hard to cope with. A person who maintains a positive relationship with a depressed spouse is likely to accept the caregiver role and to be protected from the typical psychological burden [58,59]. In contrast, those who have a negative relationship with a depressed spouse feel intensified pressure from their inevitable role as a caregiver; thus, the relationship has a detrimental influence on the caregiving spouse. Second, marital satisfaction fulfills the need for social connectedness, thus decreasing depressive symptoms [60]. Caring for a depressed spouse imposes great restrictions on social activities and the maintenance of external relationships. Limited social activities represent a lack of social interaction opportunities, which can increase the risk factors for various mental health conditions [61,62]. However, if a couple is in a good relationship, a spouse can offer emotional resources. An intimate spousal relationship helps to satisfy an individual's basic needs, providing ways to overcome stressors through psychological support [63][64][65][66]. Therefore, even if a spouse has depressive symptoms, an individual who maintains a good relationship with them is unlikely to experience depression. Among couples who have a supportive marital relationship, the effect of spousal depression may be different for husbands and wives. In particular, only a wife's depressive symptoms appear to benefit from supportive relationships. This finding indicates that greater emotional reactivity to negative interactions is, in part, negated by the stronger healing effects of supportive interactions. Although it is difficult to determine why gender differences occur based on systematic research, there is reason to think that gender differences are involved in interpersonal sensitivity. A wife is likely to share the joys and sorrows of intimate ones more vividly than her husband is, thus explaining why both supportive and negative spousal relationships affect the depressive symptoms of a wife more strongly than those experienced by her husband [51,[67][68][69].
Spousal network overlap, one of the marital relationship characteristics, was expected to be a significant moderator of spousal concordance in depressive symptoms for both husbands and wives. However, this association seemed significant only for husbands. A husband with a high proportion of spousal network overlap was less likely to also experience depressive symptoms along with his wife. For a wife, this interaction effect was not statistically significant, and only a direct consequence was associated; having a high proportion of spousal network overlap increased a wife's depressive symptoms. Our finding concerning the husband's depressive symptoms is that the presence of mutual networks with a wife buffers the psychological burden of a wife's depressive symptoms. This finding suggests that linkage with a wife's network can provide social support that can alleviate a husband's depressive symptoms when his wife has depressive symptoms. Previous studies have revealed the positive effects of couples' triadic relationships on spousal support and marital satisfaction [20,70,71]. If a respondent's spouse had more contact with the respondent's other network members, the respondent was more likely to view a spouse as a reliable source of support, open up to them, and discuss health issues with them [20]. Connecting with members of a spouse's network increased a spouse's empathic understanding and the ability to infer the spouse's needs [70]. Additionally, close network members enable common information flow through triadic relations, facilitating a sense of couplehood and enhancing marital quality [31]. In this context, we could confirm which mechanisms underlie the main finding that spousal concordance in depressive symptoms would decrease through spousal network overlap.
However, this buffering effect of spousal network overlap on depression concordance was significant for husbands but not for wives. Instead, spousal network overlap increased the wives' depressive symptoms and affected them directly. Traditional gender roles can help explain this gender difference: While gender roles are slow to change in Western society [72,73], South Korea has preserved the long-standing tradition of a patriarchal system. A wife is mainly involved in domestic life, child rearing, and caring for other family members [74]. We believe that the overlapping networks involved in a spousal relationship bring expectations about a wife's gender role into the triad and complete the triadic relationship by strengthening, diminishing, or modifying the wife's role. Therefore, the spousal network overlap effects were strongly gender specific, thereby lessening husbands' depressive symptoms while amplifying those of wives.

Limitations
Several limitations should be acknowledged. First, depressive symptoms were self-reported, and thus, these findings involve the risk of reporting bias and may not be generalizable to clinical levels of depression. It is also difficult to discriminate between depressive symptoms and other mental health problems. Second, our respondents were restricted to a traditional rural village in Korea. Approximately 67% of our respondents were working, and 88% were farmers [48]. In the Farm Household Economy Survey [75], 30% of Korean older adults were living in rural area; 47% of Korean rural older adults were working, and 75% engaged in agriculture, forestry, and fisheries. These differences indicate a small selection bias in the KSHAP. It would be inappropriate to generalize our findings to urban areas or countries without due consideration. Third, using longitudinal data, this study provides support for spousal relationships associated with spousal concordance in depressive symptoms. However, one's depressive symptoms provide evidence of the relevance of spouses' sociodemographic factors [76], behavior risks of alcohol and drug use [77,78], and physical health status [79][80][81]. We adjusted for covariates such as age, health-related comorbidities, and cognitive function in the SUR analysis. However, data for those factors pertained only to the respondents, and such data on their spouses were not included in the SUR analysis. Future research is needed to clarify whether those spouses' risk factors are longitudinally associated with spousal concordance in depressive symptoms. Fig. 2 The adjusted values of husbands' and wives' CES-D scores based on the interaction effects of spouses' CES-D and marital relationships from panel data, using seemingly unrelated regression (SUR) models. Footnotes: Analyses were conducted separately by marital relationships, including their supportive or negative characteristics and network overlap proportion, and were adjusted for age, number of social activities, education level, household income, number of comorbidities, alcohol consumption, and cognitive impairment. The results are presented in Supplementary Tables S1-S3

Conclusion
Our findings demonstrate that spousal depressive symptoms, supportive or negative marital relationships, and network overlap proportion are all associated with depressive symptoms among older adults. These findings suggest that approaches to helping older adults deal with mental well-being may need to occur at both the individual and couple levels. Our results indicate the presence of gender differences in the association of a marital relationship with depressive symptoms; spousal network overlap had buffering effects only for men. Although there is a possibility of reverse causation, gender-specific differences in the moderating effects of marital relationships on depression concordance still exist. These findings indicate the need to consider appropriate gender-specific strategies to support the mental well-being of older adults.