Cross-sectional and longitudinal associations between quality of rst marriage and subjective health assessment in older Chinese residents

Background: First marriage was vital for a common person in all-life, and there were concerns that marital quality had relationship with self-reported health and quality of life (SRH and SRQoL), health change, and sleep quality. This study aimed to examine longitudinal associations between subjective health assessment and quality of rst marriage to characterize the stability and directionality of the trajectory of marriage- health over time. Methods: Data were from the Chinese Longitudinal Healthy Longevity Survey. Chinese elders completed surveys across 3 waves (2008/2009, 2011/2012, and 2014). Using autoregressive cross-lagged models, bidirectional relationships between SRH, SRQoL, health change, sleep quality, and quality of rst marriage over time were examined. Results: Cross-sectional analysis conrmed the signicant associations between SRH, SRQoL, health change, sleep quality, and quality of rst marriage. Autoregressive linear models of SRH, SRQoL, health change, sleep quality, and quality of rst marriage were conrmed. Cross-lagged relationship between SRQoL and SRH, between SRQoL and sleep quality, between SRQoL and health change, between SRH and sleep quality, between quality of rst marriage and SRQoL, and between sleep quality and health change were conrmed. Conclusions: Subjective health assessment was associated with future subjective health assessment across 3 longitudinal waves. Quality of rst marriage might be inuenced by SRQoL among older Chinese. Future research needs to examine inuencing psychological mechanism of the cross-lagged relationships.

Relationship between marriage and health were often analyzed with longitudinal data. For example, longitudinal associations of marital quality and marital dissolution [10], longitudinal associations between depressive symptoms and marital processes [11], longitudinal associations between marital quality and sleep quality in older adulthood [12], and longitudinal associations between alcohol consumption and negative marital quality [13] were investigated. Several longitudinal studies observed bidirectional association between poor marital quality and depression [14], bidirectional association between marital problems and marital dissatisfaction [15], bidirectional associations between changes in insomnia and changes in marital quality [16], and bidirectional relationships between marital and sleep problems [17]. Another study suggested both marital strain and marital strength have potent effects on biology and health [18]. Using longitudinal data, previous studies have examined how narcissism [19], change in cognitive limitations [20], workloads [21], and dyadic coping [22] predicted the trajectory of marital quality over time. Regarding the link between marital quality and health, a study with a national longitudinal survey show that marital strain accelerates the typical decline in SRH at older ages [23].
Several studies empirically evaluated and validated the effectiveness of intervention for couples. Regarding marital functioning, there were some signi cant partner effects [24]. A supportive spouse might buffer stress-related autonomic processes linking low socioeconomic status to risk for cardiovascular disease [25]. For example, providing support to couples might improve marital functioning and an opportunity for relational growth during end-stage cancer [26]. On the basis of marital health model, marriage checkup could promote marital health by marital interventions [27]. A cross-disease review evaluated couple-oriented interventions for chronic physical illness [28]. Additionally, a current meta-analysis suggested a relationship checkup to improve couples' marital functioning up to six-month follow-up [29]. Likewise, couple-based interventions could be feasible, acceptable, and e cient [30] and be initiated early during pregnancy [31]. But, effective interventions could not possibly operate well in China. Before liberation and on the early Days of New China, early marriage was common in Chinese society and might have an important role on late-life marital instability. Furthermore, a current study reported that the majority of the health behaviors of elderly individuals in China were not healthy [32].
To better understand the stability and directionality of these associations over time, we used a autoregressive cross-lagged models to examine the direct and reciprocal relationships between marital and health variables using a longitudinal survey. For example, we can simultaneously model autoregressive effects (ie, relationships between early a quality of rst marriage predicting the same quality of rst marriage at a future wave; or relationships between early a subjective health assessment predicting the same subjective health assessment variable at a future wave) and cross-lagged effects (ie, indirect effects of earlier marital variables on future subjective health assessment variables, or indirect effects of earlier subjective health assessment variables on future quality of rst marriage variables). If the cross-lagged relationship was unidirectional over multiple waves such that SRH and SRQoL predicts quality of rst marriage but not vice versa, this would further support the notion that quality of rst marriage were risk factors for subjective health assessment.
Thus, this study rst depicted the sample characteristics. Second, this study explored associations between quality of rst marriage and self-reported subjective health assessment in 2008/2009 wave, 2011/2012 wave, and 2014 wave. Third, this study explored longitudinal mediation effects between quality of rst marriage and subjective health assessment. In the end, the key statistical outcomes were analyzed.

Method
Sample Data for this study were from the Chinese Longitudinal Healthy Longevity Survey Second, In order to explore the association between socioeconomic factors, SRH, SRQoL, and quality of rst marriage in 2008/2009 wave, 2011/2012 wave, and 2014 wave with multiple logistic regressions, quality of rst marriage were recoded as good (=0) and poor (so so, bad =1). Simultaneously, age (less than 79=0, more than 80 =1), gender (female=0, male=1), ethnicity (ethnic minority=0, Han majority=1), and number of people living with (less than 2=0, 2 and above=1) were categorized.
Third, in order to explore longitudinal associations of quality of rst marriage with the four health variables, quality of rst marriage with SRH, SRQoL, health change, sleep quality was still re ected by original categorical options. According to combination formula, quality of rst marriage could be combined with single health variable, double variables, three variables, and four variables. Thus, fteen models could be obtained. Path models were used to simultaneously estimate the following: (1)  Analyses were run by using Stata (version 14.0).

Participants
In . This was in line with average age difference of rst-marriage couples in the 2000s that the "older husband and younger wife" was the main mode of marriage age matching [33]. Regarding rst marriage pattern, China remained as a universal-marriage society despite a steady rise of the age at rst marriage [34]. Obviously, there were signi cant gender differences in age group, number of people living with, SRH, health change, quality of rst marriage, and sleep quality in  Note: ***, ** and * indicated 0.01, 0.05 and 0.10 significance level, respectively.  Note: ***, ** and * indicated 0.01, 0.05 and 0.10 significance level, respectively.

Longitudinal associations
Results in gures 1a-1o from the path models indicated signi cant autoregressive effects of quality of rst marriage trajectories. Thus, quality of rst marriage in 2008/2009 wave predicted quality of rst marriage in 2011/2012 wave. In turn, quality of rst marriage in 2011/2012 wave predicted quality of rst marriage in 2014 wave. The same results of SRH, SRQoL, health change, and sleep quality could be found in models 1 to 15. CFA results on model t for 15 models were presented in Table 3. Here, all 15 models were acceptable. There were signi cant cross-lagged relationship between SRQoL and SRH in models 5, 11, and 15 cross-lagged relationship between SRQoL and sleep quality in models 6 and 13, cross-lagged relationship between SRQoL and health change in model 7, cross-lagged relationship between SRH and sleep quality in models 8, 11, 14, and 15, cross-lagged relationship between quality of rst marriage and SRQoL in model 13, and cross-lagged relationship between sleep quality and health change in model 13, respectively. There were signi cant correlations among quality of rst marriage, SRH, SRQoL, health change, and sleep quality in 2008/2009 wave.

Discussion
This study explored reciprocal relationships between subjective health assessment and quality of rst marriage among an elder sample of couples in autoregressive cross-lagged models using 3 waves of matched longitudinal data. In the logistic regression, signi cant associations between SRH, SRQoL, health change, sleep quality, and quality of rst marriage were documented. In the structural equation models, this study did nd signi cant bidirectional associations between quality of rst marriage and SRQoL between 2011/2012 wave and 2014 wave. Likewise, the reciprocal relationships between SRQoL and SRH, between SRQoL and sleep quality, between SRQoL and health change, between SRH and sleep quality were reported.
Explicably, the marriage-health relationship could not be supported successfully in this study. First, Chinese marriage process was different from that in western countries. A discrete-time event history analysis indicated that there was nonlinear relationship between age at rst marriage and marital stability in a China's settings [35]. Another cause may be cohort differences in the relative impact of marital dissolution on physical health [36]. Second, marital trajectories have speci c marital quality. The trajectory of marital quality over the life course had three periods of decline, stagnation, and decline in Iran [37]. The third explanation might come from spousal interactions. A study supported gender differences in the genetic and environmental in uences on different aspects of marital quality in the United States [38]. Regarding associations between marital quality and negative experienced well-being in later life, marital appraisals played a complex role in shaping negative emotions among older adults [39]. There might be marital dynamics of gender difference in well-being in later life. For example, the association between husband's marital quality and life satisfaction might be in uenced by marital quality of his wife [40].
The results that marital and psychological functioning could not predict each other could be explained on the basis of the previous studies. Subjective assessment was not the biological and clinical outcomes. Part of the participants was married in adolescence. Thus, age at rst marriage might be related to the risk of developing cardiovascular diseases and cancer in later life [41]. Regarding married men, changes in longitudinal marital relationship quality appears associated with associations with a range of CVD risk factors [42]. Psychologically, variations in the marital quality might affect cardiovascular health [43,44].
Empirically, marital quality was related to metabolic syndrome through its relationship to depressive symptoms for men and women [45].

Limitations
Due to statistical failure, this study did not re ect the spousal interactions in the longitudinal associations. For example, a study uses data from the 1992 Health and Retirement Study show that women reported lower marital happiness, marital interaction, and marital power than do men in later life [46]. From a life course perspective, a study showed that the strains of marital dissolution undermine the self-assessed health of men but not women [47]. Without psychological data, this study did not re ect the psychological processes in the longitudinal associations. Another study reported integrating psychological processes and physical health into change in marital quality, model of marriagehealth links were constructed [48].

Conclusion
The present study con rmed cross-sectional associations between SRH, SRQoL, health change, sleep quality, and quality of rst marriage among older Chinese adults. Autoregressive relationships among SRH, SRQoL, health change, sleep quality, and quality of rst marriage were also con rmed. But, the most cross-lagged relationships were not acceptable in China's settings. Possibly, the psychological mechanisms of the classic relationships need be analyzed further.

Consent for publication (for human subjects)
Not applicable.

Availability of data and material
The datasets analysed during the current study are available in the http://opendata.pku.edu.cn/.

Competing Interests
The authors declared no potential con ict of interest with respect to the research, authorship and/or publication of this article.
Authors' contributions MG designed the study, performed the descriptive and cross-sectional analysis, and completed the original version. HG conducted the statistical analysis of the longitudinal associations.