Exclusion in Very Old Age The Impact of Three Critical Life Events

This paper focuses on relational exclusion (i.e. isolation and nonparticipation in social activities) in very old age. Based on a five-year study of an octogenarian cohort, the authors investigate the impact of three critical life events (deterioration of health, death of a close relative, entry into a nursing home) on relational life and social involvement. With advancing age, older people withdraw from some social activities, but their relationships with their family and friends remain stable. Life events have a stimulative effect on the support network (especially of family), and only the deterioration of health curbs social activity. This would seem to confirm the existence of a process of disengagement stemming more from the older people's functional or sensory disabilities than from an individual choice.


Introduction
The term 'exclusion' first appeared in France in the 1960's (for a review, see Paugam, 1996).At that time, exclusion was viewed primarily as the persistence of an earlier situation and as specific to certain population groups which had been unable to take advantage of the opportunities offered by developments in society.During the 1980's and 1990's a considerable change took place in the perception of this idea.Most scholars now consider that contemporary forms of exclusion result from the impact of changes in societal frameworks -the labour market, the institutions providing relational support to individuals (families in particular) or the institutions (such as churches or voluntary associations) integrating individuals into society.Castel (1995), for example, laid particular stress on changes in the system of economic production and exchange which undermine the status of persons in employment and the protections associated with that status.He also examined the changes in the structure and patterns of functioning of family and social ties and the emergence of a new cultural model.In the view of this French sociologist, these changes are fostering the development of a 'negative individualism' which tends to jeopardize relational patterns and to weaken the solidarities and institutions in which individuals are integrated.
More recently, the term 'exclusion' has entered the vocabulary of other European countries; it has been adopted in the official parlance of the European Union (where it has replaced the word 'poverty' as a key term in programmes on the socio-political agenda) and has provided a starting-point for a number of sociological studies (e.g., Byrne 1999; Hills et al. 2002;Woodward & Kohli 2001).
Advanced old age and aging appear to be of purely marginal interest in the multitude of writings and discussions of the last few years on the subject of exclusion.The purpose of the present article is specifically to shed some light on certain aspects of this little-known subject.In addition, it seems to the authors that there is a need to break away from the common view of old age as a life situation which by definition is one of withdrawal from, or existence at the fringe of, events in the outside world.We assume that, although forms of exclusion are encountered in later life, they are only the outcome of specific situations and processes and should be analyzed as such.In other words, individuals are not 'excluded' once and for all, but there are population groups in a state of frailty who are liable to become subject to marginalization processes (Schnapper 1996).
The multidimensional character of exclusion has been established in a number of contemporary writings (Leisering & Leibfried 1999;Paugam 1996).Attention must thus be directed not only towards material living conditions -economic and financial resources, housing, etc. -but also towards the forms of relational integration and social involvement.The problem of exclusion is distinct from that of poverty precisely because it comprises the question of social ties -that is to say, the manner in which an individual relates to other individuals and to society in general.The present study focuses on the relational aspect of exclusion, distinguishing between (1) isolation, characterized by the absence or rarity of interactions with members of the family or the network of friends and acquaintances and (2) non-participation in social activities, by which is meant activities conducted outside the home and involving the company of other persons, or at least a social environment.
On the basis of data obtained from a longitudinal study still under way, we initially studied how these dimensions evolve with advancing age.Do very old people become increasingly isolated and gradually withdraw?We then sought to determine the impact on isolation and nonparticipation of a certain number of disruptions peculiar to the aging process.To that end, we considered three critical life events: deterioration of health, death of a close relative, and, entry into a nursing home.To what extent do the disruptions which occur during the last stage of the life course induce exclusion?Are there differences deriving from the dimension of the exclusion under consideration or the type of life events studied?

Current State of Knowledge
Generally speaking, gerontological studies have demonstrated that aging is linked to a gradual disengagement and abandonment of a whole range of social activities (e.g., Bijnen et al. 1998).This withdrawal into the home may be accompanied by isolation but there is not necessarily a link between the two (Bickel & Lalive d'Epinay 2001).Often the absence of contacts -like loneliness -has been regarded as a problem associated with old age.However Wenger and her colleagues (1996), in a review of writings on the subject, affirmed that the phenomenon has often been overestimated.Several authors have shown that, all in all, older people are well looked after by relatives and acquaintances (Lalive d'Epinay et al. 2000; Wagner et al. 1999).Even so, there is agreement on the fact that the number of social contacts gradually decreases with advancing age (Rook 2000).This decrease is partly due to functional or social losses: spousal bereavement, illness or the death of other members of the network.
In any study on very old age, it is important to take into account the process of senescence within which the increasingly frail individual is gradually brought to accept a certain number of losses (relating to health, close relatives, the ability to act and to participate, etc.).These losses are not synonymous with exclusion but they may become key factors in the chain of biographical events leading to situations of exclusion, by disrupting the life continuity or by entailing the intervention of outside parties or recourse to compensatory measures.We will now examine some of these losses.

Deterioration of Health
Disability has often been associated with paucity of social interactions (Thompson & Heller 1990).However, a number of authors have demonstrated that poor health or disability does not necessarily lead to a decline in relational life or to isolation (Field et al. 1993;Simonsick et al. 1998).The family, when there is one, often strengthens its ties and rallies round its dependent or frail elder.The risk of losing ties of friendship appears relatively greater, inasmuch as the constraints imposed by poor health hinder respect for the norm of reciprocity, which plays an important regulatory role in friendly relationships (Johnson & Barer 1997).Isolation results less from desertion of the disabled person by his or her friends than from withdrawal by that person when the disability is a lasting one (Johnson & Barer 1992).Regarding social activities, several studies have shown that such activities dwindle with age and increasing physical frailty (e.g., Bickel & Lalive d'Epinay 2001;Cutler & Hendricks 2000); only attendance at religious services remains stable (Levin 1995).In particular, aging appears to lead to the abandonment of activities requiring physical effort, visual or auditory strain or absence from the home.
Most research has focused on functional disabilities, often overlooking the role of sensory losses, which are in fact closely related to functional abilities and everyday competence (e.g., Brennan et al. 2005).Vision and hearing impairments can disrupt interpersonal relations and limit social participation or leisure activities (Crews & Campbell 2004;Kelly 1995;Wallhagen et al. 2001).Inability to follow a group conversation is a serious impediment to social life that may lead to forms of self-exclusion.Sensory losses may also give rise to problems of balance and thus oblige older people to restrict their movements (Marsiske et al. 1999).

Death of a Close Relative
A great deal has been written about the process of bereavement and the causes of death, as well as spousal loss and its repercussions.On the other hand, there has been very little research into the consequences of the death of a child or a sibling.Regarding spousal bereavement, it is well established that the loss of the spouse increases the likelihood of early death (Bowling & Windsor 1995) and often gives rise to psychosomatic disorders (Lund et al. 1993).A death also seriously affects the composition of the relational 'convoy' (Antonucci & Akiyama 1987) of older people and diminishes the group of relatives and friends, and in particular the group of peers.However, spousal bereavement gives rise to widely differing results with regard to its impact on social integration.The loss of the spouse does not necessarily lead to isolation; widowers generally continue to be well supported (Ferraro et al. 1984), and contacts within the family tend to strengthen in the short term (Lalive d' Epinay et al. 2003).Occasionally, older people compensate for the loss of their spouse by increasing their participation in social activities or strengthening their activities or relationships with friends, who themselves have often been through the experience of spousal bereavement (Utz et al. 2002).But other studies show that isolation is more frequent among widowers and widows or reveal a decline in participation in social activities after the bereavement (cf.Wenger et al. 1996).The death of the spouse may also give rise to a gradual weakening of ties of friendship, especially where the deceased spouse was the main actor in those relationships (Rook 2000).

Entry Into a Nursing Home
Moving into a nursing home may be seen as a way of compensating for the weaknesses and losses of capacity occurring in advanced age or as a remedy when the handicap is no longer manageable within the home or when loneliness becomes difficult to bear.From the standpoint of common sense, entry into an institution often means marginalization, relegation and even segregation.In the pioneering work done by Goffman (1961), establishments for the aged are treated as 'total institutions', i.e. enclosed spaces, cut off from the outside world, in which there is a great lack of privacy -a social microcosm within which existence is experienced negatively in comparison with normal life and in which the individual is deprived of his roles.Is this a true picture?Doubt has been cast on it by a number of studies which propose a more qualified approach.The assertion that aged parents are 'dumped' into nursing homes and then neglected by their families has not been borne out.Older people are not cut off from their social networks once they are in an institution; their loved ones seem to feel it a duty to stick by their institutionalized relative (e.g., Pruchno et al. 1994;Yamamoto-Mitani et al. 2002).It is true that some nursing home residents are relationally marginalized.However, this is not because the nursing home causes isolation but rather because it admits individuals who have become isolated on account of events within their families or of their situation in life (Cavalli 2002; Kahana et al. 1984).Hence the advantage, when analyzing a transition (in this case into a nursing home) and its consequences, of studying it within a longer sequence of the life course.

Data
This article is based on data of the Swiss Interdisciplinary Longitudinal Study on the Oldest Old (SWILSOO), which scrutinizes health trajectories in very old age with the aim of identifying the individual and environ-mental factors and processes that are conducive to an older person's autonomy, physical and mental integrity, and participation in society.SWILSOO is conducted in two contrasting regions of Switzerland: the canton of Geneva -a culturally secular, urban area -and the central Valais region -an Alpine area of small towns and villages, Catholic in tradition.The panel was launched in 1994 and, at the present stage, involves two five-year cohorts (persons born 1910-1914, N=340, started in 1994; persons born 1915-1920, N=374, started in 1999).Each cohort was randomly selected among community-living persons aged 80-84 years at baseline, and stratified by region and gender.The information is gathered by faceto-face interviews based on a closed-end questionnaire.In cases where the elder is no longer able to take part in the interview personally, a proxy is used (for a more detailed presentation of SWILSOO, see Lalive d' Epinay et al. 2001).
Analyses are based on the first cohort, using data collected between 1994 and 1999 in the course of five survey waves carried out at intervals of 18, 12, 12 and 18 months.In view of the type of information we needed, we retained only those people who, at the first wave, answered the questionnaire themselves.The sample is thus made up of 295 individuals at baseline (1994), 157 (53%) of whom still participated in the fifth wave (1999), in person or through the intermediary of a close relative.The attrition of 46% comprised 23% deaths and 23% withdrawals and other departures from the study.Altogether, 1,085 interviews constituted the data for our analyses.At baseline, the mean age was 81.83 years (SD = 1.39).

Measures Exclusion
In studying the relational aspect of exclusion, a distinction is made here between: Isolation, which is evaluated by recording separately the frequency of visits received from family members or from friends and acquaintances.In both cases a scale is used ranging from 1 (no visits) to 5 (visits every day or nearly every day); and Non-participation, which is measured on the basis of a list of five so-cial activities (participation in parlour games, visits to cafés or restaurants, attendance at religious services, excursions or travel, and participation in local festivals or events).The regularity of active participation is measured, starting from a threshold of one activity at least once a month for the first three activities and of at least once a year for the last two.A scale of regular and active participation (ranging from 0 to 5) was constructed.

Critical Life Events
Three types of major disruption affecting older people were taken into account: Deterioration of health: the development of incapacity in three healthrelated areas was analyzed.The first two relate to functional health, i.e. the ability of the individual to accomplish the activities of daily living (ADL).The study covers five basic activities -toileting, eating and cutting up food, dressing and undressing, rising and going to bed, and moving around within the apartment (Katz et al. 1970) -and three measures of mobility -moving around outside, walking at least 200 metres, and going up and down stairs (Rosow & Breslau 1966).Sensory capacities are also studied: the first item is an evaluation of the ability to read a newspaper (vision); the other two items are measurements of the ability to participate in a face-to-face conversation or to follow a conversation in a group (hearing).It is considered that a disruption of health has taken place if the person concerned is no longer able to carry out alone one or more of the activities studied.
Death of a close relative: at each interview any deaths (of spouse, siblings or children) were recorded.
Entry into a nursing home: the last event studied was the act of removal to a long-stay establishment for the aged with medical facilities.

Analyses
Initially we focused on the changes which had taken place between the first and fifth waves of the study with respect to a certain number of characteristics of the sample and in relation to the indicators of isolation and social involvement.Subsequently, with a view to measuring the association of the three critical life events with our indicators of exclusion, we computed analyses of covariance (ANCOVA).This enabled us to compare the average scores of different groups on the two indicators of relational integration and on participation in social activities (in Wave n), taking into account the scores obtained on the same indicators during the previous wave (Wn-1).It should be specified here that the unit of analysis was the observations, not the individual.Thus for each individual we have several observations -the number depending on the number of waves in which the person concerned took part -which are classified in different groups according to whether or not, at a given moment (Wn), one of the life events was recorded.We carried out separate analyses for each disruption, each time comparing three groups: (1) persons who had suffered the loss under consideration; (2) persons who had suffered one of the other life events; (3) persons who had not suffered any life event.Naturally, an individual could suffer more than one life event between two waves.1

The Resources of the Oldest Old
Table 1 presents firstly some of the socio-demographic characteristics of our cohort of octogenarians at the start of the survey, and secondly the evolution of some of its resources between the first and fifth waves of the study.It must be emphasized that our sample of persons aged 80-84 is not representative of that age group.In the first place, the more or less equal proportions of men and women and the equal representativeness of the two regions is entirely an effect of the stratification of the sample; the high proportions of married persons is an indirect effect of the same factor, since it derives from the over-representation of men.Secondly, octo- genarians suffering from disabilities and forms of dependence are underrepresented on account of the elimination at the start of the study of persons living in nursing homes and persons unable to participate personally in the interviews.
While at the start of the study our sample was made up entirely of persons living in private dwellings, by the time of the fifth wave 8% of the persons interviewed were living in nursing homes (4% of the men and 12% of the women).It is worth noting also that, at the outset, only one man in five lived alone compared with three women in five.Developments in family networks confirmed that, the older a person, the more he or she becomes a survivor within his or her circle; an aged person is somebody who is faced with the loss of his close relatives and friends, and particularly his peers (cf.Lalive d' Epinay et al. 2003).Over the fiveyear period, nearly one out of five married persons had to cope with spousal bereavement, while the percentage of octogenarians without siblings increased by half.Although far more men than women still had their spouse at the start of the survey, women were more likely to lose their spouse during the period in question (32% compared with 13%).Moreover, within the five-year period the state of health of the survivors clearly deteriorated; in fact, the proportion of persons unable to perform by themselves at least one of the five basic ADL actually quadrupled.

Exclusion and Its Evolution in Advanced Old Age
The frequency of visits by the family and by friends and acquaintances did not vary significantly between the first and fifth waves of the study (see Table 2).Thus, notwithstanding the relative contraction of their social networks, very old people did not gradually become isolated.On the other hand, there was a clear decline in the level of their social activities; the proportion of very old people participating regularly in less than two activities almost doubled over five years (increasing from 23% in W1 to 40% in W5).
It should be pointed out that for the time being we have considered the evolution of the exclusion indicators in an aggregated fashion, i.e. as they affect the cohort as a whole.But it may be that these general changes we have commented on conceal individual trajectories departing from the norm: some individuals may withdraw and reduce their activities, while others may enjoy an enrichment of their social involvement.As can be seen from Figure 1, in the area of family visits nearly all (90%) of the older people who were receiving visits more than once a month at the start of the survey were still being visited often five years later.In only a very few cases did individuals experience a diminution of family life, while almost two-thirds of the persons who were isolated strengthened their links with members of their families.Regarding relations with friends, 40% of the individuals who were not being visited by friends in 1994, were by 1999 receiving at least one visit per month; for a similar proportion the trend was opposite.Finally, regarding sociability, just under one-third of the persons who had withdrawn into themselves increased their participation during the five-year period, while one-third of active persons took the opposite path.2The Impact of the Three Critical Life Events These initial findings confirm that one must beware of drawing a parallel between the trajectories of octogenarians and marginalization processes.
Having said that, it is reasonable to wonder whether certain life events associated with the final stages of the life course may not worsen a condition of frailty and lead to forms of exclusion.Over the five-year period, we recorded 159 cases of deterioration of health, and 107 deaths of close relatives, while 22 octogenarians moved into nursing homes.
Table 3 shows the effect of these three events on relational life and social involvement.Two general observations must be made: firstly, whatever the type of loss sustained, we did not observe any defections, either by families (which, on the contrary, showed a tendency to become more caring) or by friends; and secondly, the process of withdrawal from social activities only begins following the worsening of health.
Let us now examine the impact of the three critical life events in more detail.In comparison with those who did not suffer any disruption, persons who lost a close relative received significantly more visits from families (Fischer's LSD post-hoc test, p = .002)and friends (LSD post-hoc test, p = .003).Regarding social activities, the two groups were not sinifyone state to another are calculated on the basis of different figures.
Example of interpretation: among the 34 persons who had less than two regular activities in W1, the situation of 71% (n = 24) remained unchanged, while 29% (n = 10) changed categories and in W5 were participating more actively in social life.
cantly different using the LSD post-hoc test (p = .159).Our findings also highlighted a mobilization of the social network during the period following entry into a nursing home.Persons who had just moved received significantly more visits from family members than persons in the 'no disruption' group (LSD post-hoc test, p < .001).Today, institutionalization is often the consequence of a deterioration of health, and the persons involved are old-old people with a level of impairment such that they cannot live without the almost constant attendance of another person (Cavalli 2002).Even so, it should be noted that participation in social activities hardly declined following entry into a nursing home (LSD post- .48*** Note.Isolation: scales ranging from 1 (no visits) to 5 (every day or almost every day).Nonparticipation: between 0 and 5 regular activities.Means in the same row that do not share subscripts differ at p < .05. ** p < .01. *** p < .001.hoc test, p = .066). 3 Furthermore, LSD post-hoc tests did not detect any difference in the frequency of contacts between persons not having suffered any disruption and those whose health had declined (visits by family: p = .298;visits by friends: p = .223).Finally, among both men and women who suffered a decline in their functional or sensory abilities, the general level of activity was significantly lower than in the 'no disruption' group (LSD post-hoc test, p < .001).
At this stage of the analysis, we wish to explore in greater depth the consequences of a decline in health on social integration, distinguishing between the different types of incapacity (basic ADL, mobility ADL, and sensory disabilities).The results obtained (see Table 4) bear witness to the 3 One possible explanation may lie in the fact that most of these institutions regularly organize a number of social activities and that in addition the staff may encourage nursing home residents to take part.thus avoiding or delaying transfer to a nursing home.The study of the sensory dimension has likewise proved enlightening.Visual and hearing difficulties are highly debilitating and contribute to the abandonment of social activities and withdrawal into the home (LSD post-hoc test, p = .011);but here there was no external compensation in the form of increased frequency of visits by family and friends (LSD post-hoc tests: visits by family, p = .229;visits by friends, p = .859).

Discussion
In this article we have tried to shed some light on the problems of relational exclusion during the final stages of the life course.Based on SWIL-SOO data, we have pointed out that very old people cannot be systematically associated with the condition of exclusion, and demonstrated the need to distinguish between non-participation and isolation.We have established that each of these dimensions evolves in a different fashion.Although there is a decline in the practice of social activities, relational life remains stable.Our findings also revealed a measure of heterogeneity in individual trajectories.In particular, a significant proportion of older people whose relational lives or level of social activities were poor at the start of the survey escaped from those situations, which carried within them a risk of exclusion.These last findings corroborate the argument that exclusion is frequently a phase of life through which an individual passes rather than a stable condition (Schnapper 1996).An analysis of the impact of the three critical life events (deterioration of health, death of a close relative, entry into a nursing home) revealed some inflection in the mode of integration, with greater mobilization of the social network (particularly of the family) and at the same time a withdrawal from social activities, but only in the case of a declining state of health.
These findings encourage us to revisit the disengagement theory (Cumming & Henry 1961).It will be remembered that, in its original form, this theory treated the reduction of interactions with others and withdrawal from the public sphere as phenomena intrinsic to aging.It asserted that, starting from a certain stage of human development, the weakening of social involvement and the psychological detachment from social roles are preconditions for the successful aging sought by indi-viduals.In addition, it claimed that there is a congruence between the design of society and the destinies of individuals, with the former (through institutions such as retirement) responding to the requirements of biological-genetic evolution.This theory has attracted much criticism (see Hochschild 1975;Maddox 1964) and is hardly mentioned in current empirical studies (Achenbaum & Bengtson 1994).However, more recent writings have revived interest in the disengagement theory and at the same time reshaped it.Johnson and Barer (1992), on the basis of a study of 150 persons aged 85 or over, showed that almost one-half of the oldest-old interviewed compensated for the losses which marked the closing stages of their life course and maintained their commitments, while the other oldest-old tended selectively to modify their social world both socially and psychologically, and withdrew from some of their roles and activities.Seen in this perspective, disengagement may be understood as one possible pattern of adaptation in very old age.This leads us to investigate the factors which prompt older people to adopt that course, and in particular to enquire whether the withdrawal is intentional (as the original disengagement theory claimed) or, on the contrary, imposed.
From this standpoint we may speak of a tendency towards disengagement, without, however, conferring on it a character of generality or inevitability.If a process of withdrawal takes place, as seems to be the case, it would seem to result not so much from a deliberate choice by the individual concerned as from a deterioration of that person's health.In any case, if there is a decline in activity, the social network (led by the family) will step forward and fill the gap.Furthermore, our work has shown that a number of factors, both individual and environmental, come into play between (1) the increasing frailty of the individual and its attendant constraints, and (2) the fact of being (or not being) in a state of exclusion.
These findings point to the need for further research into differences that may exist between respondents, depending on such characteristics as their gender and region of residence.As noted above, the cohort members living alone were mostly women; we also know that the family network is wider in the Alpine region (Lalive d'Epinay et al. 2000).Does this have an impact in terms of relational exclusion?Is the decline in social activity common to both men and women, to both the urban and the rural area?Does the way in which family and friends rally round following certain life events vary by gender or by region?Initial discriminant analysis seems to suggest that trends are roughly the same for both genders and both regions.Some slight differences emerge solely in the context of death of a loved one, where men enjoy a greater wave of support from family and friends.At the present stage, however, our analysis is hampered by a lack of numbers.We shall return to the question once the full set of SWILSOO data becomes available.
Another possible development of our work would be a study of the linkages between frailty, exclusion and well-being.More specifically, this would consist of an empirical examination of the consequences, from the standpoint of the well-being of the individual, of the situations and processes of isolation and non-participation.The second aspect of the disengagement theory asserted that withdrawal was a condition for successful aging and for greater well-being of the individual; but that assertion has now been refuted by many scholars (cf. Hochschild 1975).Be that as it may, if frail older people do have contacts, they do not necessarily control them; they depend on the goodwill of those close to them.However, research has shown that it is precisely the feeling of being in control of one's social relationships and one's relationship with the environment that provides a source of well-being (e.g., Lang et al. 1997).In accordance with the metamodel of 'selective optimization with compensation' (Baltes & Baltes 1990; Baltes & Carstensen 1996), one interesting avenue of exploration would be to broaden the range of activities studied in order to determine the extent to which withdrawal from social activities is offset by a greater emphasis on certain other forms of activity.The argument here, in accordance with the theory, is that the well-being of the individual depends on this process of adaptation.Following this, one could seek to clarify the influence of the relational situation on the success of such a process.

Figure 1 .
Figure 1.Change between W1 and W5 in number of visits received by individuals and in their sociability activities (survivors in W5, n = 157)

Table 3 .
Adjusted means, standard errors and analysis of covariance.Results for the three dimensions of exclusion by disruptions