Life goes on: Influence of social network on mental health after late-life partnership dissolution

Jordi Gumà, Universitat Pompeu Fabra
Celia Fernandez-Carro, Universidad Nacional de Educación a Distancia

The literature on the relationship between partnership status and health points that living with a partner is associated with better health outcomes when measured in terms of mortality, morbidity or general well-being. Especially in mid-life and old age, the protective effect of a partner is explained by the construction of a wider social network which can be of help in case of necessity. Indeed, the features of individuals’ social networks (i.e. structure, size and subjective quality of social network) are associated with the mental dimension of health independent of marital status (Fiori et al., 2006; Litwin and Shiovitz-Ezra, 2011).

We aim at assessing the importance of the features of social networks in depression levels after the end of a union distinguishing between separatio/divorce and the death of partner. For this purpose we explore the association between depression and characteristics of social networks (extension, structure and level of satisfaction with the social network) in old age, comparing their protective effect over different types of union interruption; divorced and widow. Our hypothesis are: 1) those with an extended social network might show lower levels of depression after finishing their relationship; 2) there must be differences in the values of depression according to the reason that origins the end of the relationship, separation or loss of partner. We expect to find gender differences in our results according to different male and female probabilities of becoming divorced or widow as well as differences in social networks profiles.

Preliminary results from linear regression model with depression scale as a dependent variable and the interaction between "Time since the end of partnership happened" and "Size of the social network" confirm the protective effect of high social networks as well as significant differences by sex (women display higher values of depression).


Lifegoes on: Influence of social network on mental health after late-lifepartnership dissolution 

 

 

Introduction

The literature on the relationship betweenpartnership status and health points that living with a partner is associatedwith better health outcomes when measured in terms of mortality, morbidity orgeneral well-being (Martikainen et al., 2005; Liu andUmberson, 2008). Conversely, the end of arelationship for divorce or widowhood have been associated with negativeeffects on health, as exemplified by the increases in depressive symptoms amongmen and women who either experienced those events (Wade and Pevalin,2004; Hughes and Waite; 2009). Especially in mid-life and old age, theprotective effect of a partner is explained by the construction of a widersocial network which can be of help in case of necessity. Indeed, the featuresof individuals’ social networks (SN) (i.e. structure, size and subjectivequality of social network) are associated with the mental dimension of healthindependent of marital status (Fiori et al., 2006; Litwinand Shiovitz-Ezra, 2011).

This work aims move forward in thisdirection, examining the association between the depression level in middle-agedand older individuals who have experienced a union interruption;divorce/separation or widowhood, and the characteristics of their socialnetworks (structure, frequency of contacts, emotional closeness and level ofsatisfaction). Our hypotheses are: 1) We expect differences in the values ofdepression according to the reason that origins the end of the relationship;separation or loss of partner, 2) the protective effect of social networksvaries depending on the type of union interruption and the characteristics ofthe social network.

Data and Methods

We run linear regression models based on asubsample of cases from the sixth wave of the Survey of Health, Ageing andRetirement in Europe (SHARE). Although the original sample of SHARE is 66.196for 17 European countries (excluding Israel), our interest on those experiencedthe end of a union regardless of the fact they actually declare to are in a newunion or not, makes us to work with a subsample of 15,546 individuals (23,5% ofthe total sample).

Dependent variable

Depressive symptoms are measured using theEURO-D scale ranging from 0 to 12, higher values indicating more depressionsymptoms. Descriptive analysis of this variable shows a mean value of 3.6(standard deviation 2.7) for the analyzed subsample.

Variables of interest

·         Structure of SN: Different combinations ofFamily network (e.g., siblings, parents, children, and spouse), friendshipnetwork, and work-related network (e.g., coworkers and supervisors)

·         Satisfaction with SN: 0-10 Scale (0completely dissatisfied-10 completely satisfied)

•    Average emotional closeness to all membersof social network

·         Time since the end of partnershiphappened. Four categories: less than two years; from two to four years; fromfive to nine years; and ten years and over.

·         Average contact with all members of socialnetwork (daily; several times a week; about once a week; about every two weeks;about once a month; less than once a month; never)

All our analysis are controlled for age(age and age squared), gender, educational attainment (ISCED classification infour categories: no studies, low education –ISCED 1 and 2-, medium –ISCED 3 and4-, and high –ISCED 5 and 6-), physical limitations (limited or not limitedbased on GALI), clusters of countries (Northern –Sweden, Denmark and Estonia -,Central –Austria, Germany, France, Switzerland, Belgium and Luxembourg -,Eastern –Czech Republic, Poland, Slovenia and Croatia - and Southern –Spain,Italy, Greece and Portugal-).

Preliminary results

Figure 1 shows the predicted values by ourpreliminary model according to the two variables of interest. Those with asmall social net show higher values for the depression scale in all the periodsafter the end of the union, whereas those with a larger social network show thelowest values. Regarding the time after experiencing the end of the union, thegeneral decreasing trend becomes flat after 5 years and over for those with asmall social network while for those with a high social network still thedecrease is observed in all the periods.

Figure 1. Predicted margins of depressionscale by time since end of union and size of social network

Note: Controlling for age, gender,education and groups of countries.

Predicted values of depression accordingto the type of end of union are displayed in Figure 2. Although results fordivorced individuals are constantly lower than for widows, difference is smalland not statistically significant.

 

Figure 2. Predicted margins of depressionscale by time since end of union and type of end of union

 

Note: Controlling for age, gender,education and groups of countries.

Apart from the two figures above, itstands out the statistically significant coefficient for gender (female=1) inthe linear regression model (beta=0.93, p-value=0.000). This would confirm ourhypothesis of gender differences. Future analysis will be stratified by gender.

 

Presented in Session 1177: Health, Wellbeing, and Morbidity