Marital Transition and Risk of Stroke
How Living Arrangement and Employment Status Modify Associations
Background and Purpose—There have been consistent findings reported that marital transition (ie, change in marital status during a given time period) is associated with risk of cardiovascular disease; however, few studies have been conducted on stroke risk, particularly stroke subtypes. Moreover, no studies have examined the moderating effect of living arrangement and employment status on the association between marital transition and stroke risk.
Methods—We examined sex-specific associations between marital transition and stroke risk using data from Japan Public Health Center–based Prospective Study. We included 24 162 men and 25 626 women who were married at prebaseline (5 years before baseline). Marital transition was determined by marital status at baseline. Weighted hazard ratios of stroke risk were estimated by Cox proportional regression analysis with inverse probability of weighting using a propensity score.
Results—An increased risk of stroke, particularly hemorrhagic stroke, was observed among men and women with marital transition (ie, married to unmarried); weighted hazard ratios (95% confidence interval [CI]) for men and women were 1.26 (1.13–1.41) and 1.26 (1.09–1.45), respectively. Participants with marital transition and lived with children had increased stroke risk. Living with parents buffered the increased stroke risk owing to marital transition among men; however, no such effect was identified among women. Elevated stroke risk owing to marital transition was magnified among women if they were unemployed; weighted hazard ratio=2.98 (95% CI, 1.66–5.33).
Conclusions—Living arrangement and employment status modified the positive associations between marital transition and stroke risk, which differed by sex.
Marital status is viewed as an important determinant of health.1 Married people have been consistently reported to be healthier than unmarried people, with more profound effects among men.2–7 These associations can be attributed to both marital selection (ie, healthier people are more likely to be married) and marital protection (refers to the benefits of marital ties on health).8–11
Consistent findings have been reported from research investigating marital transition (ie, a change in marital status during a given time period).5 Men and women who experience marital transition by death or divorce have prospectively increased risk of associated onset and progression of cardiovascular disease.12–14 The hypothesized mechanisms underlying the association between marital transition and cardiovascular risk include preexisting health conditions,5 change to unhealthy profiles of health behavior,15–17 poorer psychological state2 stemming from the loss of financial stability,18,19 and reduced social support and social networks11,20 owing to loss of the spouse. However, few studies have been conducted to address stroke risk,17,21 particularly stroke subtypes,22 in this population. In addition, no such studies have been conducted in Asia.
Sociocultural differences could be a potential moderator for the health effect of marital transition.5 A prospective study of middle-aged men and women in Japan showed that marital dissolution significantly increased the risk of cardiovascular mortality among men, but no such impact was identified among women.6 Another prospective study of Japanese elderly adults showed no evidence of increased all-cause mortality risk among widowed men and women; in fact, decreased mortality risk was found among widowed women.7 These inconsistent results could be because the social roles of men and women in Japanese society are different, with strong gender role norms (ie, the male breadwinner model). Under such social norms, women are generally more likely to adopt the role of providing emotional support to their spouse. Therefore, widowed women might feel relieved of their duty to care for their spouse emotionally, which might in turn lower their mortality risk.
Marital transition often changes people’s living arrangement and economic situation. Living alone is a well-known health risk factor.23–26 In addition, a previous study conducted in Japan showed that who people live with was important for increased risk of coronary heart disease incidence and mortality.26 Thus, the magnitude of stroke risk could vary by living arrangement.
In the same way that marital transition affects economic status (ie, through loss of the breadwinner or reduced sources of household income), the impact on employment status must also be considered. However, to our knowledge, no studies have examined the moderating effect of living arrangement and employment status on the association between marital transition and stroke risk.
In this study, we sought to examine the sex-specific associations between marital transition and incidence of total stroke and stroke subtypes, as well as modification of the identified associations, according to living arrangement and employment status, in a large prospective cohort in Japan.
We used data from the Japan Public Health Center–based Prospective Study (JPHC Study) cohorts 1 and 2, a large population-based prospective study of 140 420 men and women aged 40 to 69 years.27 Cohorts 1 and 2 were initiated in 1990 and 1993, respectively, within 11 public health center areas throughout Japan. Two PHC areas in metropolitan Tokyo and Osaka were excluded from the present analysis (n=23 524) because no data on stroke incidence were available. Of the remaining 116 896 participants, 9 were excluded as ineligible.
A self-administered questionnaire was distributed to all registered participants in 1990 (for cohort 1) and 1993 (for cohort 2; response rate: 81.6%). A follow-up survey was conducted 5 years after the first survey (response rate: 84.9%). A total of 80 964 men and women responded to both questionnaires. For this study, we set the dates of the first survey as our prebaseline and those of the second survey as our baseline to determine marital transition between first and second surveys (Figure).
The JPHC study was approved by the institutional review boards of the National Cancer Center and Osaka University.
We limited our study population to those who were aged 45 to 64 years at baseline (n=65 921); we excluded 2892 people with no information on their living arrangement with spouse at either prebaseline or baseline. Of the remaining 63 029 participants, we included only those who lived with their spouse at prebaseline (n=52 525). We further excluded 2747 people with a history of cancer or cardiovascular diseases at baseline. The final study population included 24 162 men and 25 626 women.
Our main predictor was marital transition, determined by participant responses about cohabitation status with their spouse at baseline. Marital transition was categorized into 2 groups: (1) no: continuously lived with their spouse from prebaseline and (2) yes: stopped living with their spouse before baseline.
Age, residential area, occupation, and living arrangement at baseline were hypothesized confounding factors. We considered other baseline characteristics such as health behaviors, perceived psychological stress, life enjoyment, and history of disease, to be mediating factors for the association between marital transition and stroke risk because marital transition occurred before the baseline; thus, baseline characteristics were considered consequences of marital transition.
Cohabitation with ≥1 parent or child and employment status were our hypothesized moderating factors. On the basis of responses to living arrangement at baseline, we grouped participants into 2 groups for each factor: living with parent(s) (yes/no), and living with ≥1 child (yes/no). Employment status was identified by responses about participant occupation (employed/unemployed).
Confirmation of Stroke Incidence
The end points of this study were incidences of total stroke, hemorrhagic stroke, and ischemic stroke. Participants were followed from the time of the baseline survey until January 1, 2010 for cohort 1 and until January 1, 2011 for cohort 2. Residential status, including survival status, was confirmed annually through the residential registry of each area. Six percent of the respondents had moved from their original residential areas, and 0.2% of the respondents were treated as censored.
A total of 81 major hospitals were registered within the administrative districts of the JPHC cohort. Physicians blinded to patient lifestyle data reviewed the medical records at each hospital. Strokes were confirmed according to criteria of the National Survey of Stroke.28 Detailed confirmation methods have been described elsewhere.29
To address the possibility of selection bias, we applied inverse probability weighting (IPW) using a propensity score for the probability of having marital transition.30
Sex-specific propensity scores were calculated using multiple logistic regression as the probability of having marital transition, given all relevant factors at prebaseline including age, occupation, PHC area, living arrangement, overweight, smoking, ethanol intake, leisure time physical activity, perceived high psychological stress, life enjoyment, and history of hypertension, diabetes, or hypercholesterolaemia.
The standardized differences of covariates by marital transition before and after IPW using the propensity score were calculated (Table I in the online-only Data Supplement). All absolute standardized differences after IPW using the propensity score were <10%, which suggests that IPW using the propensity score created no significant differences in prebaseline characteristics according to marital transition at baseline.31
Sex-specific hazard ratios (HRs) and 95% confidence interval (CI) for total stroke, hemorrhagic stroke, and ischemic stroke according to marital transition were calculated using Cox proportional hazard regression analysis using IPW with the propensity score. We adjusted for our hypothesized potential confounding factors at baseline (model 2). Furthermore, we included the hypothesized mediating factors (model 3). We also compared stroke risk among categories created by the combination of marital transition and living arrangement with parent(s) at baseline, setting the reference group as participants without marital transition who did not live with their parent(s) at baseline. We performed a similar analysis for living arrangement with ≥1 child, setting the reference groups as participants without marital transition who did not live with any children at baseline, and for employment status, participants without marital transition who were employed at baseline.
A total of 1732 men and 3205 women had marital transition between prebaseline and baseline periods (Table 1). Men and women who had marital transition by the baseline were likely to have more health risks compared with those who had not had marital transition.
During the median follow-up period of 15.0 years, 2134 cases of stroke were documented (Table 2). An increased risk of total stroke was observed among men and women with marital transition; weighted multivariable HRs (95% CI) for men and women were 1.26 (1.13–1.41) and 1.26 (1.09–1.45), respectively (model 2). These relationships remained unchanged after further adjustment for possible mediating factors (model 3). The respective weighted multivariable HRs (95% CI) of hemorrhagic stroke and ischemic stroke for marital transition were 1.47 (1.23–1.76) and 1.20 (1.05–1.38) for men and 1.32 (1.08–1.61) and 1.15 (0.94–1.40) for women.
Men with marital transition who were employed at baseline showed significantly increased risk (HR, 1.31; 95% CI, 1.17–1.47; Table 3). In addition, marginally significant interaction was identified for living with a child (P=0.09); men with marital transition who lived with their children had the highest risk of developing stroke compared with other groups created by a combination of marital transition and living arrangement with a child (HR, 1.44; 95% CI, 1.23–1.70).
Among women, marginally significant association was identified with employment status (P=0.08). Women with marital transition who were unemployed had the highest stroke risk among groups created by a combination of marital transition and employment status (HR, 2.98; 95% CI, 1.76–5.07). In addition, although the interaction was not statistically significant, women with marital transition who lived with their parent(s) were at highest stroke risk among groups by a combination of marital transition and living arrangement with parent(s) (HR, 1.33; 95% CI, 1.03–1.72). Another analysis by living arrangement with ≥1 child identified that only the group of women with marital transition who lived with their child at baseline had increased stroke risk (HR, 1.45; 95% CI, 1.19–1.77).
In this large prospective analysis among a Japanese population, we found that men and women with marital transition in the 5 years before baseline were at higher risk of developing stroke compared with those who had been continuously married. These associations seemed to be more evident for hemorrhagic stroke and no significant difference was identified by sex. Subgroup analysis by living arrangement and employment status indicated that these factors modified the association between marital transition and stroke risk. Men and women who lost their spouse and lived with their child had an increased risk of stroke, whereas no such increased risk was identified among the other groups. However, living with parent(s) had different impacts by sex; living with parent(s) buffered the stroke risk after marital transition among men but exacerbated the stroke risk among women. For women, employment status was significantly associated with the effect of marital transition; women with marital transition who were unemployed had triple the risk of stroke incidence compared with those who had no marital transition and were employed.
Longitudinal studies on marital transition and stroke risk are limited. Prospective studies have identified increased stroke incidence after marital dissolution among both male and female adults in the United States17 and Sweden,21 and among elderly adults in Australia.22 Our results were consistent with those in that in our study, men and women with marital transition were more likely to develop stroke compared with their counterparts who had no change in marital status. Although there is a conflicting evidence about whether the magnitude of association between marital transition and stroke risk is higher for men17,21 or for women,22 we found no significant sex differences with respect to stroke risk.
One hypothesized mechanism underlying the association between marital transition and stroke risk is cardiovascular risk factors caused by marital transition. Our results indicated unhealthy profiles of behavior and psychological state among men and women with marital transition, consistent with previous evidence.2,15,16 We found a higher proportion of heavy drinkers and lower proportion of ex-smokers among men and women with marital transition than those without marital change. The average ethanol intake per week among drinkers was significantly higher in both men and women with marital transition than in those without. High perceived psychological stress and low-life enjoyment, which are risk factors for cardiovascular disease,32,33 were also more prevalent among men and women with marital transition. The change in psychosocial environment surrounding individuals after loss of the spouse, such as a lack of social support, a reduced social network, and poorer psychological state, could be responsible for their increased stroke risk. In addition, the identified greater impact of marital transition on hemorrhagic stroke compared with ischemic stroke suggests increased alcohol use after loss of the spouse could be a crucial explanation for the impact of marital transition on stroke, because heavy drinking is a stronger risk factor for hemorrhagic stroke than for ischemic stroke.34
Stroke risk after marital transition differed by participants’ living arrangement. Our results suggest that parental roles magnify the risk of stroke among both men and women who have lost their spouse. Socially and economically difficult situations surrounding single-parent status of either the mother or the father35 could contribute to a deteriorating health effect after loss of the spouse. However, living with one or both parents revealed different pictures for men and women. For men with marital transition, living with their parent(s) buffered the adverse health impact after loss of the spouse, but no such effect was identified among women. The effect of strong sex role norms in Japanese society, such as the male breadwinner model,36 could be one explanation for this discrepancy. Men who lost their spouse may be supported by parent(s) who take over their household duties, which could buffer an increased stroke risk after marital change. However, women with marital transition were likely to lose their financial stability owing to loss of the breadwinner in their household. However, it is unlikely that women would be financially supported by their elderly parent(s); rather, they might be obligated to support their parent(s) financially.
Employment status is another moderating factor for stroke risk. Among women, we found that the elevated stroke risk after marital transition was magnified in women who did not have a job. The tough economic conditions facing unemployed women could reinforce the tendency to neglect their health after loss of the spouse. The risk of stroke in unemployed women was interactively elevated by both loss of their partner and economic difficulties. Conversely, no clear picture emerged among men in this regard owing to the low proportion of unemployed men in this study.
This investigation is one of the few longitudinal studies to observe an association between marital transition and stroke risk by using IPW with a propensity score, and it is the first to do so in Japan. However, we note several limitations to this study. First, residual confounding could have occurred by measurement errors in our self-reported variables and unmeasured confounding variables (such as psychological stress). In particular, we included only occupation as an indicator of socioeconomic conditions. It is clearly necessary to consider including comprehensive measures of socioeconomic conditions in future studies. Second, misclassification bias of marital transition may exist; however, we considered that possibly the direction of bias would be toward null than overestimated. We considered the possibility of resuming living with their spouse for those with marital transition may be reduced as they become older. However, the possibility of losing their spouse among those without marital transition would occur more as time goes. Therefore, we think that the impact of misclassification may be limited. Third, although marital transition has 2 directions, we dealt with only marital transition from married to unmarried in this study. Fourth, the definition of marital transition deserves consideration. Marital transition was defined according to participants’ living arrangement with their spouse, thus it may have included people who lived apart from their spouse because of work. However, because the proportion of households with members who lived apart from the family owing to work has been estimated at 3.5% of the total private households in Japan in 1995,37 we think the effect of misclassification is small. Fifth, health impacts may have differed by marital transition because of separation, divorce, or death; however, we could not differentiate these owing to a lack of information about the reason for marital change. Finally, generalizability of our study results could be limited. Although our study population consisted of people living in various areas throughout Japan, it may not be nationally representative. Notably, we did not include metropolitan areas. In addition, it could be limited for our findings to be generalized to other societies because social factors such as marital transition are closely associated with its sex in each society. However, our findings indicated the significance of gender perspective about stroke risk, and, in this point, we think our results are of use to the other societies.
In our study, marital transition was associated with a higher risk of stroke, in particular hemorrhagic stroke, among both men and women. Although our results suggested no significant sex difference in stroke risk after marital transition, the impacts of living arrangement and employment situation differed materially by sex. Living with child(ren) could be a factor that increases stroke risk of marital transition among both sexes. Living with parent(s) buffered the increased stroke risk after marital change among men, but amplified it among women. Elevated stroke risk after loss of their spouse was magnified only in women who did not have a job.
Sources of Funding
This work was supported by National Cancer Center Research and Development Fund (23-A-31[toku] and 26-A-2; since 2011) and a Grant-in-Aid for Cancer Research from the Ministry of Health, Labour and Welfare of Japan (from 1989 to 2010). This work was also supported by Japan Society for the Promotion of Science KAKENHI Grant Numbers 26460744 and 26253043.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.011926/-/DC1.
- Received October 29, 2015.
- Revision received February 2, 2016.
- Accepted February 2, 2016.
- © 2016 American Heart Association, Inc.
- Ebrahim S,
- Wannamethee G,
- McCallum A,
- Walker M,
- Shaper AG.
- Koskenvuo M,
- Kaprio J,
- Romo M,
- Langinvainio H.
- Lee S,
- Cho E,
- Grodstein F,
- Kawachi I,
- Hu FB,
- Colditz GA.
- Eng PM,
- Kawachi I,
- Fitzmaurice G,
- Rimm EB.
- Rutledge T,
- Linke SE,
- Olson MB,
- Francis J,
- Johnson BD,
- Bittner V,
- et al
- Simons LA,
- McCallum J,
- Friedlander Y,
- Simons J.
- Ikeda A,
- Iso H,
- Kawachi I,
- Yamagishi K,
- Inoue M,
- Tsugane S,
- JPHC Study Group
- Tsugane S,
- Sawada N.
- Iso H,
- Baba S,
- Mannami T,
- Sasaki S,
- Okada K,
- Konishi M,
- et al.,
- JPHC Study Group
- Iso H,
- Date C,
- Yamamoto A,
- Toyoshima H,
- Tanabe N,
- Kikuchi S,
- et al
- Shirai K,
- Iso H,
- Ohira T,
- Ikeda A,
- Noda H,
- Honjo K,
- et al.,
- Japan Public Health Center-Based Study Group
- Iso H,
- Kitamura A,
- Shimamoto T,
- Sankai T,
- Naito Y,
- Sato S,
- et al
- Shirahase S.
- Brinton M.
- 37.↵Statistic Bureau of Ministry of Internal Affairs and Communications, Japan. Household with married-single-person. 1995 population census of Japan results of special tabulation on living with parents. http://www.stat.go.jp/english/data/kokusei/1995/12.htm. September 14, 2000. Accessed October 15, 2015.