Content uploaded by Stefano Cavalli
Author content
All content in this area was uploaded by Stefano Cavalli on May 03, 2015
Content may be subject to copyright.
OMEGA, Vol. 60(4) 301-325, 2009-2010
BEREAVEMENT IN VERY OLD AGE:
IMPACT ON HEALTH AND RELATIONSHIPS
OF THE LOSS OF A SPOUSE, A CHILD, A SIBLING,
ORACLOSEFRIEND*
CHRISTIAN J. LALIVE d’EPINAY, PH.D.
STEFANO CAVALLI, PH.D.
LUC A. GUILLET, MA
University of Geneva, Switzerland
ABSTRACT
This article deals with the following two questions: In very old age, which are
the main sources of bereavement? And what are the consequences of such
losses on health and on relationships? The findings are based on the complete
set of data compiled in the course of the Swiss Interdisciplinary Longitudinal
Study on the Oldest Old (SWILSOO), which provided a 10-year follow-up of
a first cohort (1994-2004) and a 5-year follow-up of a second (1999-2004).
The data revealed that, in very old age, the great majority of the dear ones who
died were either siblings or close friends. Taken as a whole, the bereaved
suffered a marked and lasting increase in depressive symptoms, together with
a short-term deterioration in their functional status; those bereft of a spouse or
*The data on which this article is based were drawn from the Swiss Interdisciplinary
Longitudinal Study on the Oldest Old (SWILSOO), a study conducted at the Center for Inter-
disciplinary Gerontology of the University of Geneva, Switzerland. This research program
was supported by grants from the Swiss National Science Foundation (Principal investigator
Prof. Christian J. Lalive d’Epinay).
301
Ó2010, Baywood Publishing Co., Inc.
doi: 10.2190/OM.60.4.a
http://baywood.com
a child saw their functional status worsen and exhibited enduring depressive
symptoms but they also benefited from support in the form of increased
interaction; those bereft of siblings only suffered from a mild, short-term
deterioration in functional status; those who had lost a close friend suffered
a very significant increase in depressive symptoms. In the medium term,
most of these effects disappeared, lending weight to the claim that the
survivors manage to cope with the misfortunes of life.
DEATH AND ITS IMPACT IN LATER LIFE
The further a person advances into old age, the more he or she becomes a
survivor—i.e., someone whose close relatives and friends have disappeared one
by one. Having already snatched the person’s ascendants, death now claims the
spouse, decimates siblings and friends, and increasingly strikes at later gener-
ations. In the circle of family and friends, there are soon as many dead as living.
Spousal bereavement is, however, more typical among the young-old than the
old-old. Beyond the age of 80, women greatly outnumber men and most of
them are already widows. The men, who are usually married, are still highly
likely to die before their spouse. In Switzerland, according to the 1990 census
(Swiss Federal Statistical Office, 2000), there are about twice as many women
as men in the 80-and-over age bracket. Only 36% of the women still have their
spouse, compared with 64% of the men. Advanced old age thus seems to be
punctuated by the loss of contemporaries among siblings and friends, people
with whom the survivor had ties forged over years of sharing the ups and downs
of life, activities, memories, and emotions. At the same time, death now seems
to strike later generations more frequently than before. Moss, Lesher, and Moss
(1986) found that 10% of persons had lost a child when they were over 60, and
the risk grows with age.
Impact on Health
A good deal has been written about the process of bereavement, the causes of
mortality, and the advent and repercussions of widowhood. The literature is scant,
however, on the reactions of older people to the death of a child, a sibling, or a
friend (see Fitzpatrick, 1998).
Spousal Bereavement
The excess mortality of the widowed, especially in the year following bereave-
ment, is now well established (Bowling & Windsor, 1995; Carey, 1979-1980;
Thierry, 1999, 2000), confirming the existence of a “broken heart” syndrome
(Parkes, 1972; see also Stroebe & Stroebe, 1993). Spousal bereavement often
affects health also: according to Tudiver, Hilditch, Permaul, and McKendree
(1992), roughly one person in four incurs problems of physical or psychological
302 / d’EPINAY, CAVALLI, AND GUILLET
health in the first year of widowhood. Nevertheless, as far as the long-term
consequences of widowhood on health are concerned, the findings of research
do not all concord. Widowhood does not seem to be associated with wor-
sening physical health (Bennett, 1997; Lalive d’Epinay, Bickel, Maystre, &
Vollenwyder, 2000). From the psychosomatic point of view, however, most
researchers agree that, compared to their married contemporaries, the widowed
display a greater tendency toward depressive symptoms, even after the mourning
period (Delbès & Gaymu, 2000; Gallagher, Breckenridge, Thompson, & Peterson,
1983; Umberson, Wortman, & Kessler, 1992). However, the widowed do not all
react in the same way. Applying a cluster analysis to a sample of bereaved spouses,
Ott, Lueger, Kelber, and Prigerson (2007) concluded that about half of them
experienced “common grief” (a depressive mourning period), one-third showed
quick resilience, and less than one out of five suffered from chronic depression.
The loss of a spouse, undoubtedly a major event, is not the only instance in
which a person with whom one has close affective ties and memories of shared
times of life, activities, and emotions is taken away. Setting aside the loss of an
ascendant, which is an unusual event in very old age, let us consider the death
of contemporaries (among siblings and friends) and offspring.
Loss of a Sibling
Although the death of a sibling may be seen as a normal and accepted event
during advanced stages of life, work done by Hays et al. on 3,173 bereaved elders
showed that it has as great an impact on functional and cognitive status as the
loss of a spouse (Hays, Gold, & Pieper, 1997). Some researchers underline the
adverse consequences of this disruptive event (e.g., Brubaker, 1985), but others
suggest that the impact varies according to the affective relation between the
survivor and his/her sibling (Cicirelli, 1995, 2001; Moss et al., 1986; Moss &
Moss, 1989a, 1989b; Moss, Moss, & Hansson, 2001).
Loss of a Child
Death is expected to respect the generational order. However, the more one
advances in old age, the greater the risk of losing an offspring, especially for older
women with male descendants (Moss et al., 1986, 2001). Among the very few
studies on parental bereavement in advanced old age, that of de Vries, Davis,
Wortman, and Lehman (1997) based on longitudinal data, detected a strong
and lasting increase in depression (see also Arbuckle & de Vries, 1995; Rogers
& Reich, 1988).
Loss of a Friend
Among the few empirical studies on this subject, that of de Vries, Lehman, and
Arbuckle (1995) concluded that those bereft of a friend remained in better
BEREAVEMENT IN VERY OLD AGE / 303
health than the control group of non-bereaved. Hays et al. compared the elders
bereft of a friend with those bereft of a sibling. The former had a lower ADL score
and better self-rated health, but also a higher level of depressive symptoms
(Hays et al., 1997). More qualitative studies, for their part, pointed up the adverse
consequences of the death of a friend (de Vries & Johnson, 2002; Fitzpatrick,
1998; Roberto & Stanis, 1994).
Two methodological remarks are necessary here. First, as Hansson, Hayslip,
and Stroebe (2007) noted, many of the studies failed to consider pre-bereavement
health status. Second, is the recourse to a non-bereaved group of elders as control
adequate? Consider for a moment the study by Hays, Gold, and Pieper (1997).
Based on a longitudinal survey, the authors compared changes in the health
status of four groups of individuals: bereft of a spouse, bereft of a brother or sister,
bereft of a friend, and a non-bereaved control group. As seen above, the com-
parison between the three groups of bereaved elders revealed that it is the loss of
a brother or sister that has the greatest impact on the survivor’s health. But at the
same time, they detected no significant difference between any of the bereaved
groups and the non-bereaved group; all the groups, including the last-mentioned,
suffered from declining health over the period in question, the decline in func-
tional status and self-rated health even being steeper in the control group than in
the bereaved groups! According to the authors, these findings may be accounted
for by the heterogeneous composition of the non-bereaved group. Following this
remark, in our research design we split the non-bereaved elders into two groups:
one consisting of those mentioning that a negative change had occurred in their life
(usually related to health), and the other of those reporting no important change.
The result was that the “non-bereaved loss group” had a higher level of physical
ailments and functional impairment than the other two (“bereaved group” and “no
change group”), and a higher level of depressive symptoms than the non-bereaved
no-change group (Lalive d’Epinay. Cavalli, & Spini, 2003). The results show that,
as suspected by Hays et al. (1997), events other than bereavement impact the life
of the very old, some being even more disruptive.
Impact on Relationships
Regarding the impact that the death of a loved one has on relational life, we
note some measure of agreement that, with advancing age, the number of social
relationships dwindles gradually (see Rook, 2000). But does this mean that
relational life wanes? Spousal bereavement, for example, does not necessarily
lead to social isolation; on the contrary, the surviving spouse is usually not short
of company (Ferraro, Mutran, & Barresi, 1984). Some older people make up
for the disappearance of their spouse by strengthening their relationships with
friends and their social participation (Gallagher & Gerstel, 1993). Conversely,
other studies conclude that isolation is more acute among widows and widowers
(Wenger, Davies, Shatahmasebi, & Scott, 1996) and that, despite the family’s
304 / d’EPINAY, CAVALLI, AND GUILLET
solicitude, the widowed suffer more from loneliness than those who are married
(Delbès & Gaymu, 2000; Wenger et al., 1996). Childless couples are particularly
prone to isolation when one of the spouses dies (Johnson, 1994; Johnson &
Catalano, 1981).
From a theoretical standpoint, it is important to make a distinction between
family and friends. Ties of blood and marriage engender a network in which all
the members are related to one another. The death of a close relative is an event
for the whole family, prompting a collective ritual and entailing relational adjust-
ments throughout the network. The friendship network, on the other hand, is
premised on elective affinities and shared interests that bind an individual to each
friend, which in no way prejudges the relations that the friends may have among
themselves (Allan, 1979). Based on reciprocal choice, the network of friends
consists principally of contemporaries and, in the case of the elderly, of long-time
friends; it therefore atrophies drastically in later life, especially as both the
opportunities and the desire to make new friends are said to dwindle (Matthews,
1986). The last-mentioned traits are, however, disputed (Johnson & Barer, 1997).
Some authors have pointed out that, in contrast to the loss of a spouse or close
relative, the loss of a friend is not an occasion for any particular recognition or
ritual and no specific reaction is organized by those close to the friend in question,
which makes coping particularly difficult (Doka, 1989; Lalive d’Epinay et al.,
2003; Sklar & Hartley, 1990); they checked that the loss of a close relative
(taken as a composite group including spouse, sibling, and child) is followed by
an increase in interactions with the family network, and that a similar increase
is not observed in the social network of an elder bereft of a friend. But, even
assuming that family ties strengthen following the loss of a close relative, is this
equally true whatever the kindred relationship? The surviving sibling is often
given little family support in mourning (Cicirelli, 2001), but the evidence provided
by research is not convincing.
Gender
A recent study entitled The feminization of bereavement among community-
dwelling older adults (Williams, Baker, Allman, & Roseman, 2006) showed that
older women (aged 65 and over) are more likely than men to be confronted with
a bereavement. This “feminization of bereavement” was already well established
in the case of spousal bereavement but the authors claim that it also holds true
with reference to the loss of another relative or a friend. A general population
poll carried out in Japan has confirmed this finding and shown further that it
applies to all age groups, from 10 to over 80 (Shimai, 2004). Both studies arrive at
the conclusion that the impact of bereavement on physical and mental health is
greater among women than among men.
Other studies are less unequivocal. Most research on spousal bereavement
has shown that a widower’s health is more affected than that of a widow (e.g.,
BEREAVEMENT IN VERY OLD AGE / 305
Lee, Willets, & Seccombe, 1998; Matthews, 1991), but this conclusion is not
unanimous. Lalive d’Epinay et al. (2000) detected no gender difference, while
Gallagher et al. (1983) found that women encountered more problems of mental
health than men.
Various explanations are given for the higher rates of mortality and morbidity
(at least in the short-term) of men following spousal bereavement. According to
the paradigm focusing on disruptive events and their regulation, the more an event
is foreseeable the more it can be anticipated and the better it can be surmounted
(Lowenthal, Thurner, & Chiriboga, 1975); from that standpoint, spousal bereave-
ment is an eventuality that women can cope with better than men (Lee et al.,
1998). Another theory (Aldwin, 1990) dwells on the repercussions of a death
on the survivor’s social network. The woman, rather than the man, is known to be
the keeper of the network; her relational life is richer and, if she loses her spouse,
she does not find herself isolated (Coenen-Huther, Kellerhals, & Von Allmen,
1994; Matthews, 1991). Moreover, the status of widow, unlike that of widower, is
something that is well established, calling to mind the existence of widows’ groups
which are always ready to receive and support new members. This is another
reason for the enhanced risk of solitude among men after spousal bereavement.
A third approach centers on the reordering of daily life; up until the most
recent cohorts, at least, elderly widows have been better armed than widowers
to deal with the chores of everyday life. The combination of these three perspec-
tives (normality and anticipation of the event, network and role, reordering of
daily life) adds credence to the assertion that the level of stress associated with
spousal bereavement is higher among men (Pearlin, Menaghan, Lieberman, &
Mullan, 1981).
Men have the consolation of at least one positive factor, however, and that is
their better chance of remarrying (Thierry, 1999) or more generally of finding
a new partner. A qualitative study carried out in the canton of Geneva first with
widows and then with widowers, all in their 70s, revealed that nearly all the
widowers had found a companion or constant friend, without necessarily con-
templating marriage or even cohabitation (Groupe “SOL” de l’Université du 3e
âge de Genève, 1996, pp. 186-187).
At this point, five remarks must be made. First of all, with some quoted
exceptions, the empirical studies on the loss of a close relative (other than a
spouse) or friend are of a qualitative nature and/or are based on small samples.
Second, they deal with the elderly without distinction. However, according to one
hypothesis of the life course theory, the likelihood of a given event occurring, and
the repercussions of the event, differ depending on the period of life when the
event occurs (Elder, 1998). Johnson and Barer (1997) point out, for example,
that one characteristic of the very old is their ability to cope with tragic events.
Third, most of the studies examine the impact of bereavement a few months or
a year after the event. Psychosomatic and relational effects may, however, come
to light much later in the adjustment process. As Fitzpatrick (1998) suggests for
306 / d’EPINAY, CAVALLI, AND GUILLET
morbidity, the analytical plan should be designed to cover both the short- and the
medium-term. Fourth, the research design must be able to take into account
the pre-bereavement health status of the elder. Lastly, the non-bereaved taken
as a whole do not constitute an adequate control group.
WORKING HYPOTHESES
Based on a review of the literature, which showed that on many topics
the results of research were contradictory, we propose the following working
hypotheses.
Differential Incidence of the Loss of a Dear One
Hypothesis 1: In advanced old age, the source of most bereavements is
a sibling or a friend, rather than the spouse or a child.
Impact on Health
There are three dimensions to health: physical ailments, functional health, and
depressive symptoms (see next section).
General Hypotheses
Hypothesis 2: The loss of a close relative or friend affects negatively the
health (whatever its dimension) of the elder.
Hypothesis 3: The greatest negative impact of the loss of a dear one is
on the psychological dimension of health.
Differential Hypotheses (According to the
Relationship to the Person Mourned)
Hypothesis 4: The death of a spouse or a child has a stronger and more
lasting impact on health (physical, functional and depressive symptoms)
than the death of a sibling or friend.
Impact of Relationships
Hypothesis 5: Close relatives: considering the systemic nature of the
family network and the presence of a bereavement ritual, the loss of
a close relative is followed by a strengthening of family interactions
(strongest when the person mourned is the spouse or a child).
Hypothesis 6: Friends: considering the egocentric organization of the
friend’s network and the lack of a bereavement ritual, the loss of a close
friend can be expected to impoverish the social life of the survivors.
BEREAVEMENT IN VERY OLD AGE / 307
Gender
We choose here to test the validity for older women of the contention formu-
lated by Williams et al. (2006) on the “feminization of bereavement.”
Hypothesis 7: Older women are more at risk than men to suffer losses
among their loved ones.
Hypothesis 8: The death of a loved one has a greater adverse effect on
older women than on older men.
With the exception of the first proposition, for which there is ample evidence,
these are exploratory hypotheses which will serve as a guide to the analysis of
data and interpretation of results.
METHODS
Sample and Respondents
This study is based on data compiled under the Swiss Interdisciplinary Longi-
tudinal Study on the Oldest Old (SWILSOO). It scrutinizes health trajectories
in advanced old age with the aim of identifying the individual and environmental
factors and processes that are conducive to an older person’s autonomy, physical
and mental integrity, and participation in community life; conversely, it also aims
at pinpointing the factors and processes that impair or diminish that person’s
autonomy, integrity, and ability to participate. The SWILSOO data were gathered
in two contrasting regions of Switzerland: the Canton of Geneva (a culturally
secular, urban area) and the central part of the Canton of Valais (a Catholic and
Alpine area of small towns and villages). The panel was launched in 1994 with a
first cohort of 340 persons born 1910-1914, and augmented in 1999 by a second
cohort of 377 elders born 1915-1919. The two samples were randomly selected
among community-living elders aged 80 to 84, and stratified by region and gender.
The information was collected by means of face-to-face interviews with a mostly
closed-ended questionnaire. When an elder was not able to take part personally
in the interview, a proxy was used. Nine survey waves were carried out with the
first cohort and five with the second, at intervals of 18, 12, 12, and 18 months
inside each 5-year period. The fieldwork was completed in 2004, after 10 years.
Among the first cohort, 59 members were still participating in 2004 (dropout
rate from 1994 to 2004: 51.7% deceased, 30.9% other), and among the second
cohort 173 (dropout rate from 1999 to 2004: 21.2% deceased and 32.9% others).
For a more detailed presentation of SWILSOO; see Lalive d’Epinay, Guilley,
Guillet, and Spini (2008).
In view of the type of information needed for this article, we retained only
the questionnaires filled in with the older people themselves (and not with a
308 / d’EPINAY, CAVALLI, AND GUILLET
proxy). This meant a total of 2,496 questionnaires (1,317 from the first cohort,
1,179 from the second).
Measures
Records of Deaths
Spouse, sibling, child—The completed questionnaire provided an exact record
of the civil status of the participant and of the number of siblings and children
at each wave. At baseline, 69% had at least one sibling, 79% had at least one
child. Forty-nine percent of participants were living with their spouse, but here
the gender gap was huge: 71% of the men had their spouse at home, but only
27% of the women. From the second wave onward, the recording of deaths was
checked against the information gathered at baseline.
Friend—At baseline, 74% declared they had a close friend in their network.
The recording of deaths of friends started at the second wave of interviews with
each cohort by way of an open question asking whether changes had occurred
in the individual’s circle of friends since the previous round.
Health
Physical ailments—This question listed a number of possible sources of
physical suffering (see Lalive d’Epinay, Christe, Coenen-Huther, Hagmann,
Jeanneret, Junod, et al., 1983, pp. 134-137). Eleven items were proposed: lower
limbs, upper limbs, head/face, back, heart, respiratory organs, stomach/abdomen,
genital/urinary organs, chest, fever, other. For each part of the body, the respon-
dent reported the degree of his or her suffering on a 3-point scale: no suffering at
all, some suffering, great suffering. This measure was similar to what Kane, Bell,
Riegler, Wilson, and Kane (1983) used for measuring physical ailments, with the
difference that they evaluated the frequency of pain. Only the last answer (“great
suffering”) was taken into account across the 11 possible parts of the body and
we were thus using here a scale ranging from 0 (no acute suffering) to 11 (11 parts
of the body with acute pain). Sample mean at W1 (SD) = 0.76 (1.11).
Functional health —The person’s ability to perform unaided the essential acts
of everyday living (Katz, Downs, Cash, & Grotz, 1970; Lawton & Brody, 1969)
was assessed. Included were five basic activities—toileting, eating and cutting
up food, dressing and undressing, rising and going to bed, and moving around
within the apartment—and three actions involving mobility—moving around
outside, walking at least 200 meters, and going up and down stairs. The overall
score, ranging from 0 to 16, was the sum of the responses calculated as follows:
can do easily (0), can do but with difficulty (1), unable to do alone (2). Sample
mean at W1 (SD) = 1.16 (2.51).
BEREAVEMENT IN VERY OLD AGE / 309
Depressive symptoms (psychological health)—Depressive symptoms were
recorded by means of the Self-Assessing Depression Scale (SADS; see Wang,
Treul, & Alverno, 1975), as adapted by Lalive d’Epinay et al. (1983, pp. 127-129).
Replies of “never” (0), “rarely” (1), “often” (2), or “always” (3) were recorded for
13 items: feels tired, has trouble sleeping, feels sad, feels lonely, breaks into tears,
feels worried, feels irritable, feels anxious, lacks confidence in the future, lacks
appetite, feels bored, lacks self-confidence, has no pleasure in doing things. This
scale ranged from 0 (no symptoms) to 39. Sample mean at W1 (SD) = 7.85 (5.58).
Relations with Family and Friends
The interactive dimension of family and social life was taken in account,
separately, based on a record of the frequency—“almost never” (0), “at least
once a year” (1), “once a month” (2), “once a week” (3), “almost daily” (4)—
of visits received, visits made and telephone conversations; range of 0 (no
contacts) to 12; sample mean at W1 (SD) = 7.05 (2.62) for family relations
and 4.93 (2.92) for social relations.
Description of Analysis
Timespan
Two temporal measures of the impact of bereavement were introduced,
short-term with a two-wave interval, and medium-term with a three-wave interval.
To be precise, for the short-term, the change in the dependent variable relating to
a death occurring between two successive waves (i.e., occurring after Wn-1 and
registered at Wn) was measured at Wntaking into account the value of the
dependent variable established at Wn-1. For the medium-term, the death was
recorded in the same way (at Wn) but the change in the dependent variable
was measured during the following wave, at Wn+1 (with reference to the value
of the dependent variable at Wn-1).
Control Group: Categorization of the Non-Bereaved Elders
If we assume that there is an adverse effect on health, for example, a decline
among bereaved subjects has to be observed but it is also necessary to demonstrate
that the decline is attributable to the bereavement and nor to some other kind
of event. This was the stumbling-block encountered by Hays et al. (1997) in
the above-mentioned study; in comparing the bereaved with the non-bereaved,
the authors observed a similar change in health in the two groups. But they
did not venture to conclude from this result that the death of a close relative
had no specific impact on the survivor’s health because, they said, of the hetero-
geneous composition of the control group. As a matter of fact, the latter certainly
comprised elders who had experienced other kind of losses, most often directly
related to their health.
310 / d’EPINAY, CAVALLI, AND GUILLET
Although Hays et al. did not offer any solution, one could be sought by
subdividing the non-bereaved into subgroups and forming only homogeneous
categories. To that end, we took advantage of a question aimed at studying the
elders’ perception of significant changes in their own life. As from the second
wave of our survey, the interview started with the question, “Since our last visit,
have there been any important changes in your life?” Affirmative answers were
classified according to the life domain impinged upon by the change and the
valence (positive, negative, and neutral).
Table 1 displays the classification based on the self-reported changes (or no
change) among the non-bereaved sequences (i.e., two-wave intervals in which
no deaths occurred). It can be seen that 64% of elders reported that no important
change had occurred in their life since the previous interview; 19% indicated
health problems, 4% a positive change (improved health, a wedding or birth in
the family, etc.). The remaining 13% were classified as “neutral” (when a change
was reported but its valence was unclear) or “other” (when the elder reported
two or more changes different in nature or contradictory in valence, or in a few
cases negative changes that were not health-related).
At that stage, we decided to disregard the categories “other,” for its lack of
homogeneity, and “positive change,” because of its relatively small number of
sequences (n= 62). The “no change” category served as control group. Comparing
the bereaved with the “no change” group provided a measure of the effect
attributable to bereavement over the short- and medium terms. A supplementary
short-term comparison was done between the bereaved and the “health trouble”
group, in order to highlight the differential impact of each kind of event.
Analyses
Multilevel analyses were applied in comparing the bereaved group with the
“no change” and “health trouble change” groups on the four dependent variables
BEREAVEMENT IN VERY OLD AGE / 311
Table 1. Sequences of Bereavement and Non-Bereavement.
Distribution of Non-Bereavement Sequences by Type of Self-Reported Change
n%%
Non-bereavement sequences
Health trouble changes
Positive changes
Neutral and other
No change
Total
Bereavement sequences
284
62
195
953
1,494
350
19
4
13
64
100
15
3
11
52
81
19
Total 1,844 100
(physical ailments, functional health, well-being, and interactive dimension).
Multilevel models were developed for the purpose of analyzing data with multi-
level sets (e.g., repeated measures for one individual). We were thus able to use
repeated measures where both the number of interviews per participant and the
time interval between interviews varied. Furthermore, multilevel analyses can
tolerate an incomplete data set because they use all available data instead of
restricting the analysis to individuals who participated in all nine waves, thus
limiting the selectivity effects (Bryk & Raudenbush, 1992).
A model was tested for each of the five dependent variables. Age (centered
on its grand mean), the three categories of change (see below), and the measure
of health, well-being, or activities at the previous wave (Wn-1) were included
as level-1, time-varying predictors; gender as a level-2, time-invariant
predictor. Also examined were the possible interaction effects between gender
and the three categories. Results were reported with a robust estimation of
the standard errors and the effects were tested by the method of restricted
maximum likelihood using HLM version 6.0 (Raudenbush, Bryk, Cheong, &
Congdon, 2004).
RESULTS
Death Ever-Present
Four hundred and sixty deaths of close relatives (spouses, children, or siblings)
and friends were registered (see Table 2). At the end of the first 5-year interval,
more than half of the survivors had lost a close relative, a dear friend, or both.
At the end of the 10-year follow-up, seven out of ten survivors (first cohort) had
experienced one or more bereavements. Close to half of the losses referred to
friends (n= 224) and four out of ten to siblings (n= 183), the loss of a sibling
being by far the most frequent type of loss in the family. Due to the small number
of the losses of a spouse (n= 34) or a child (n= 19), we had to keep those two
categories merged.
The disparity between the total of 460 bereavements in Table 2 and the 350
bereavement sequences in Table 1 is due to the sequences with more than one
death registered (in 82 questionnaires two deaths which had occurred since last
wave were registered, in 14 cases three deaths). On the subject of multiple deaths,
it is worth noting that, over the 5- or 10-year period, only one elder lost two
children, about 50 lost more than one sibling, and 100 lost several friends (ten
elders mentioned the loss of between five and eight friends).
Table 2 shows the distribution of the deceased according to their relationship
with the bereaved and Table 3 the bereaved elders by gender and according to their
relationship with the deceased. Although the mean duration of participation
in the study was about half a year longer among women, the same proportion of
both genders suffered losses, with a higher mean number of deaths reported by
312 / d’EPINAY, CAVALLI, AND GUILLET
BEREAVEMENT IN VERY OLD AGE / 313
Table 3. Bereaved Elders by Gender and by Relationship
of the Deceased
Gender Men Women Total
Death nM%a%bnM%a%bnM%a%b
Spouse
Child
Sibling
Friend
18
12
51
59
1.0
1.1
1.5
2.5
6
4
16
18
8
5
23
—
16
6
73
42
1.0
1.0
1.5
1.9
5
2
23
13
18
2
33
—
34
18
124
101
1.0
1.1
1.5
2.2
5
3
19
16
11
4
28
—
Total bereaved
elders
111 2.3 34 — 112 1.9 35 — 223 2.2 34 —
Participants at
baseline
Mean duration
of participation
(years)
329
5.5
323
5.9
652
5.7
aPercentage of total participants.
bPercentage of total number of those who had such a relative (or a close friend) at
baseline.
Table 2. Distribution of the Deceased
by Nature of Kinship Tie
n%
Spouse
Child
Sibling
Friend
34
19
183
224
7
4
40
49
Total 460 100
men. This last difference resulted from the fact that men reported more losses of
friends than women. However, taking into account only those who had a spouse
or siblings, older women were more at risk of losing their spouse or siblings.
Bereavement and Health
A comparison between the bereaved (as a whole) and the “no change” group
showed that the loss had a strong short-term impact on well-being (depressive
symptoms) and also a marked effect on functional health (Table 4). The latter
impact disappeared with time while the former waned but remained significant.
In relation to the “health trouble” group, it appeared that the two kinds of
disruptive event had a similar impact on well-being. It is interesting to note that
the subjective report of health trouble was well corroborated by the measures
of physical ailments both in the short- and in the medium-term, and of functional
health in the short-term.
When the bereaved were grouped according to their relationship with the
deceased, differences appeared (Tables 5 through 7). In none of the three groups
did the loss have a significant (at p£0.05) short-term effect on either the physical
314 / d’EPINAY, CAVALLI, AND GUILLET
Table 4. Loss of a Dear One (Close Relative or Friend):
Effect on Health (Physical, Functional, and Psychological).
Short-Term Comparison with the “No Change” and “Health Trouble”
Groups; Medium-Term Comparison with the “No Change” Group;
Multilevel Analyses
Bereaved Versus
no change
Versus
health trouble change
nof sequences 953 284
Dep. variables Coef. p-Value Coef. p-Value
Short-term
Physical health
Functional health
Depressive symptoms
Medium-term
Physical health
Functional health
Depressive symptoms
0.08
0.29
1.15
–0.01
0.02
0.76
0.293
0.044
<0.001
0.967
0.918
0.021
–0.26
–0.92
0.14
—
—
—
0.009
<0.001
0.672
—
—
—
Note: Age and sex are controlled. Physical ailments: scale ranging from 0 (no acute
suffering) to 11 (11 parts of the body with acute pain). Functional health: scale ranging from 0
(no disability or difficulty) to 16 (8 disabilities). Depressive symptoms: scale ranging from
0 (no depressive symptoms) to 39.
or the functional dimension of health. The level of significance of the relation
with functional health remained at under 0.10 for those bereft of a close relative
(spouse or child: 0.077, cf. Table 5; siblings: 0.085, cf. Table 6), but not for those
mourning a friend. Additional analyses merging all those bereft of a close relative
showed that for this group bereavement had a significant impact on functional
health, but not for those who had lost a close friend.
The results also revealed a significant increase in depressive symptoms among
those who had lost their spouse or a child (Table 5) and among those bereft of
friends (Table 7), but not among those bereft of siblings. In the medium-term,
no effect on health could be detected, the only exception being a residual impact
on depressive symptoms for elders who had lost either their spouse or a child
(Table 5, p= 0.067).
BEREAVEMENT IN VERY OLD AGE / 315
Table 5. Loss of the Spouse or a Child: Effect on Health
(Physical, Functional, and Psychological) and Family Interactions.
Short-Term Comparison with the “No Change” and “Health Trouble”
Groups; Medium-Term Comparison with the “No Change” Group;
Multilevel Analyses
Bereft of the spouse
or a child
Versus
no change
Versus
health trouble change
nof sequences 953 284
Dep. variables Coef. p-Value Coef. p-Value
Short-term
Health
Physical health
Functional health
Depressive symptoms
Family interactions
Medium-term
Health
Physical health
Functional health
Depressive symptoms
Family interactions
–0.08
0.46
0.23
0.83
0.10
0.40
1.24
–0.24
0.783
0.077
0.002
0.003
0.718
0.241
0.067
0.424
–0.46
–0.84
1.24
0.92
—
—
—
0.123
0.010
0.091
0.003
—
—
—
Note: Age and sex are controlled. Physical ailments: scale ranging from 0 (no acute
suffering) to 11 (11 parts of the body with acute pain). Functional health: scale ranging from
0 (no disability or difficulty) to 16 (8 disabilities). Depressive symptoms: scale ranging
from 0 (no depressive symptoms) to 39. Family interactions: scale ranging from 0 (no
contact) to 12.
In short, depressive symptoms of elders mourning their spouse or a child
deteriorated with lasting effect. The loss also appeared to impact their functional
health. No depressive symptoms was observed among those bereft of a sibling,
taken as a group, but a slight, transitory impact on their functional health was
noted. Those elders mourning a friend were not affected in their physical or
functional health but they did display a sharp increase in depressive symptoms,
which seemed to be regulated over time.
Bereavement and Relations with Family and Friends
We enquired into the relationship between the loss of a close relative and family
life, and between that of a friend and extra-familial life. Those bereft of their
316 / d’EPINAY, CAVALLI, AND GUILLET
Table 6. Loss of a Sibling: Effect on Health
(Physical, Functional, and Psychological) and Family Interactions.
Short-Term Comparison with the “No Change” and “Health Trouble”
Groups; Medium-Term Comparison with the “No Change” Group;
Multilevel Analyses
Bereft of a sibling Versus
no change
Versus
health trouble change
nof sequences 953 284
Dep. variables Coef. p-Value Coef. p-Value
Short-term
Health
Physical health
Functional health
Depressive symptoms
Family interactions
Medium-term
Health
Physical health
Functional health
Depressive symptoms
Family interactions
0.14
0.35
0.40
0.25
0.09
0.20
–0.44
–0.23
0.217
0.085
0.250
0.195
0.518
0.936
0.250
0.252
–0.23
–0.95
–0.66
0.35
—
—
—
0.073
0.001
0.108
0.146
—
—
—
Note: Age and sex are controlled. Physical ailments: scale ranging from 0 (no acute
suffering) to 11 (11 parts of the body with acute pain). Functional health: scale ranging from
0 (no disability or difficulty) to 16 (8 disabilities). Depressive symptoms: scale ranging
from 0 (no depressive symptoms) to 39. Family interactions: scale ranging from 0 (no
contact) to 12.
spouse or of a child experienced a short-term strengthening of their exchanges
with the family (Table 5), seemingly as a consequence of the very close rela-
tionship between the surviving elder and the deceased. This did not occur when
the deceased was a sibling (Table 6), and no similar support from their social
network was forthcoming for those bereft of a friend (Table 7). The effect seems
specific to bereavement as a disruptive event; the increase in family exchanges
was significantly higher in this instance than for elders suffering from a serious
health problem (Table 5).
A different result appears for the bereaved siblings and friends. Among them,
no change was observed in the frequency of the exchanges with the family
(Table 6), respectively in the frequency of their social contacts (Table 7).
BEREAVEMENT IN VERY OLD AGE / 317
Table 7. Loss of a Friend: Effect on Health
(Physical, Functional, and Psychological) and Friends Interactions.
Short-Term Comparison with the “No Change” and “Health Trouble”
Groups; Medium-Term Comparison with the “No Change” Group;
Multilevel Analyses
Bereft of a friend Versus
no change
Versus
health trouble change
nof sequences 953 284
Dep. variables Coef. p-Value Coef. p-Value
Short-term
Health
Physical health
Functional health
Depressive symptoms
Friends interactions
Medium-term
Health
Physical health
Functional health
Depressive symptoms
Friends interactions
0.13
0.18
1.25
–0.24
0.13
–0.30
0.61
0.41
0.266
0.209
<0.001
0.285
0.285
0.068
0.185
0.090
–0.24
–1.10
0.04
–0.02
—
—
—
0.065
<0.001
0.916
0.930
—
—
—
Note: Age and sex are controlled. Physical ailments: scale ranging from 0 (no acute
suffering) to 11 (11 parts of the body with acute pain). Functional health: scale ranging from
0 (no disability or difficulty) to 16 (8 disabilities). Depressive symptoms: scale ranging
from 0 (no depressive symptoms) to 39. Friends interactions: scale ranging from 0 (no
contact) to 12.
CONCLUSIONS
Reconsidering the Hypotheses
Sources of bereavement (Hyp. 1): Results confirmed that, in very old age,
the two main sources of bereavement are siblings and friends. The fact that as
many as three out of four deaths among close relatives (spouse, children, siblings)
are of siblings may come as a surprise.
Effect on health – general (Hyp. 2-3): We hypothesized an impact of bereave-
ment on health in general, the impact being stronger in the psychological
dimension than in the physical or functional dimensions. All in all, the results
validate the second part of the hypothesis, but only partially the first part. The
death of a loved one was associated with a pronounced short-term increase in
depressive symptoms, continuing for some into the medium-term. A short-term
impact was noted on functional status but none on physical health, the evolution
of which among the bereaved group was no different from among the “no
change” group.
Effect on health – differential (Hyp. 4): We hypothesized that the loss of a
spouse or a child would be the more damaging for the health of the survivors.
This was partly confirmed by the results. Only the loss of a very close relative had
an impact on the functional status of the mourning spouse or parent. But in the
short term, the consequences for the depressive symptoms of the bereaved were
greater for a death of a friend than for a death of a spouse or a child.
Death of a close relative and family life (Hyp. 5): Our data confirmed that
the loss of a spouse or a child is followed by an increase in interaction between
the bereaved elder and his/her family members. This phenomenon takes place
during the mourning period and fades away in the medium-term. No such regu-
lation was noted for those mourning siblings.
Death of a close friend and social life (Hyp. 6): No supportive reaction from
the friendship circle was noted for those bereft of a friend but the data did not
substantiate the idea of a decrease in friend interaction, either in the short-term or
in the medium-term, where in fact a slight increase was noted.
The feminization of bereavement thesis (Hyp. 7): Our findings confirmed
that, in the course of very old age, losing a spouse is no longer a common
occurrence, but also that the few remaining married women still remain more
at risk of spousal bereavement than married men. As for the likelihood of
losing other dear ones, the risk seemed to be equally divided between women and
men, although differences were observed depending on the relationship with the
deceased, women more often mourning siblings and men more often friends.
Gender was introduced as a control variable in every multilevel analysis but
no significant interaction was detected. All the negative consequences of bereave-
ment we measured in our research were shared equally by the two genders.
Our data did not corroborate the “feminization of bereavement” thesis.
318 / d’EPINAY, CAVALLI, AND GUILLET
Discussion
Bearing in mind the doubts expressed by Hays et al. (1997) concerning the
adequacy of a non-bereaved group as control because of the heterogeneity created
by the range of possible events impinging on an elder’s life, we decided to split
the non-bereaved into more homogeneous groups. Among the disruptive events
occurring during very old age, health problems were the most frequent with
the loss of a close one. We decided to keep two groups of non-bereaved: (i) those
who declared a recent important health problem; and (ii) those reporting that
they had not experienced any important change in their life recently, with the
latter as control group.
With this methodological design, and thanks to the high number of observa-
tions, it was possible to carry out a differential analysis of the specific implica-
tions of three categories of bereavement (bereft of spouse or child, bereft of
sibling, bereft of friend) on health and family or friends interactions.
First, our results did not produce any evidence in support of the mainstream
thesis on the repercussions of bereavement on physical health. No relation to
physical health was detected, and the modest impact (of relatively weak statistical
significance) on functional health applied only to those bereft of a close relative.
Second, our results confirmed that the most severe consequences were on
psychological health, but they were not omnipresent. The increase in depressive
symptoms was high when the loss was that of the spouse, a child, or even a friend
but was not significant for the loss of a sibling. The absence of depressive
symptoms when a sibling dies is surprising and does not tally with the results
of other studies (e.g., Hays et al., 1997). A possible explanation may lie in the
assertion that among siblings the intensity of the grief is a function of the
quality—the affective dimension—of the relationship between the survivor and
the deceased (Cicirelli, 1995; Moss et al., 2001). A similar type of reasoning
could explain the intensity of the grief associated with the death of a close
friend. As a matter of fact (see Measures section above), in this study, all deaths
in the close family circle were registered whereas, due to the elective nature of
friendship, the death of a friend was mentioned only when the elder answered
the question about “recent important changes” in his/her circle of friends and
acquaintances. Consequently, all disappearances within the family circle were
recorded whatever the affective relationship with the deceased person, while
only friends whose death mattered for the survivors were registered.
Third, our results showed that bereavement is a good tracer of the differences in
structure and in functioning of each network, the systemic nature of the family
network with its resilience, contrasting with the egocentric nature of the friends’
network with its rigidity. The family seems to offer elders more alternatives for
their affective needs than the narrow network of remaining friends. When the
deceased had been a member of the intimate family circle, the relatives rally
around the bereft mourner. Confirming Doka (1989) and Skar and Hartley (1990),
BEREAVEMENT IN VERY OLD AGE / 319
we found that those who have lost a friend do not receive any special support
from their social network and remain alone with their grief, a fact that may
explain their distress.
Fourth, in the short-term, mourning the death of a close relative or a friend
has a noticeable but differential impact on elders’ health and relational life.
In the medium-term, this very disruptive event seems to be regulated; health
and family or social life go back to normal—“normal” meaning not different
from the situation displayed by the “no change” group. However, there is one
exception: elders mourning their spouse or a child seem to suffer from a lasting
decline in morale.
While most research on bereavement dwells on its disruptive effects, espe-
cially on health, Dutton and Zisook underline people’s ability to cope with it:
“Accumulating evidence suggests that resilience to grief in the face of bereave-
ment is the norm, rather than an exception” (2005, p. 877). The focus of that study
was not old age, but the resilience of even very old persons in coping with the
various misfortunes that afflict old age has been highlighted, for example, by
Johnson and Barer (1997) and by Pearlin (1994). Our study brings fresh evidence
of the strong impact that the loss of a loved one can have on the depressive
symptoms of elders, but also of their potential to cope.
On the subject of gender, we failed to find any evidence of a “feminization of
bereavement” as applied to advanced old age. Obviously, one consequence of the
feminization of the population with advancing age is that women, living longer,
have to cope with more bereavement and more frequently end up as survivors.
But elders of both genders of the same very old age seem to be treated by death
in an even-handed—but tough—way.
The methodological strength of our research is that we were able to disentangle
the “non-bereaved” groups, which allowed us to trace the consequences of bereave-
ment and to identify the distinctive patterns of consequences according to the
deceased person’s relationship to the survivor: spouse or child, sibling, friend.
One limitation of the study is that, for statistical reasons (scantiness of data), we
merged the bereaved spouses and parents. Additional multilevel analyses on
elders bereft of a spouse and those bereft of a child, taken separately and compared
with the “no change” group, showed that the two groups experienced the same
pattern of consequences: an increase in depressive symptoms (widowed: coef. =
2.72, p= 0.005; bereft of a child: coef. = 1.54, p= 0.094) and also an increase
in contacts with the family (widowed: coef. = 0.86, p= 0.017; bereft of a child:
coef. = 0.77, p= 0.043), but more abundant data would be necessary for con-
firmation of this result.
The SWILSOO program centers on the life trajectories of the oldest old, who
now constitute the fastest-growing age group in Western countries and at the
same time the group suffering from the greatest lack of social research. Possible
comparisons with other studies are hence few. A second limitation is that our
study focused on the morbidity, not the mortality, ensuing from bereavement. One
320 / d’EPINAY, CAVALLI, AND GUILLET
of our conclusions was that the oldest old have proved to be very resilient
to bereavement; it would be more accurate to say that the survivors have proved to
be so. Suggestions for future work could be: (a) to obtain a more complete picture
by expanding the list of deaths in the family (including in-laws and grandchildren),
to make a comparative analysis of the impact of bereavement by type of kinship
tie with the deceased person (which would be worthwhile but would require
large samples); (b) to introduce the variable of subjective closeness (affective
dimension) into the model; (c) bearing in mind that the flexibility of the family
system stems mainly from its multigenerational nature, to study the situation
where a person bereft of a sibling has no offspring (20% of our elders belonged
to a one-generation family); (d) to better understand the personal experience of
death of friends in old age, by improving the measures or by developing quali-
tative research; (e) to extend samples to other periods of the life course with a
view to formulating generalizations and age-specific conclusions.
FINAL NOTE
This article is an upgraded version of a previous one published in this journal
(Lalive d’Epinay, Cavalli, & Spini, 2003). The 2003 article was based on the
observation of SWILSOO’s first octogenarian cohort over a period of 5 years
(1994-1999); participants numbered 340 and interviews gathered 1,045. This
article is based on all the data of the completed SWILSOO research program,
with a 10-year follow-up of the first cohort (1994-2004) and a 5-year follow-up
of the second (1999-2004). The number of participants at baseline was 717
(twice as many), and of interviews 2,496. Consequently, we were able to dif-
ferentiate the impact of bereavement according to the nature of the tie between
the bereft elder and the deceased: spouse or child, sibling or close friend.
ACKNOWLEDGMENTS
The authors wish to thank Dario Spini, co-author of the 2003 article, Ian
Hamilton for his editorial assistance, and two anonymous reviewers for their
helpful comments on the manuscript.
REFERENCES
Aldwin, C. (1990). The Elders Life Stress Inventory (ELSI): Egocentric and nonegocentric
stress. In M. A. P. Stephens, S. E. Hobfoll, J. H. Crowther, & D. L. Tennenbaum (Eds.),
Stress and coping in late life families (pp. 49-69). New York: Hemisphere.
Allan, G. (1979). A sociology of friendship and kinship. London: George Allen & Unwin.
Arbuckle, N. W., & de Vries, B. (1995). The long-term effects of later life spousal and
parental bereavement on personal functioning. The Gerontologist, 35, 637-647.
BEREAVEMENT IN VERY OLD AGE / 321
Bennett, K. M. (1997). Widowhood in elderly women: The medium- and long-term effects
on mental and physical health. Mortality, 2, 137-148.
Bowling, A., & Windsor, J. (1995). Death after widow(er)hood: An analysis of mortality
rates up to 13 years after bereavement. Omega: Journal of Death and Dying, 31, 35-49.
Brubaker, T. (1985). Later life families. Beverly Hills: Sage.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and
data analysis methods. Newbury Park, CA: Sage.
Carey, R. G. (1979-1980). Weathering widowhood: Problems and adjustment of the
widowed during the first year. Omega: Journal of Death and Dying, 10(2), 163-174.
Cicirelli, V. G. (1995). Sibling relationships across the life span. New York: Plenum
Press.
Cicirelli, V. G. (2001). Sibling relationships. In G. L. Maddox (Ed.), Encyclopedia of aging
(3rd ed., pp. 928-930). New York: Springer.
Coenen-Huther, J., Kellerhals, J., & Von Allmen, M. (1994). Les réseaux de solidarité dans
la famille [Solidarity networks in the family]. Lausanne: Réalités Sociales.
Delbès, C., & Gaymu, J. (2000). Du veuvage à l’isolement [From widowhood to isolation].
Gérontologie et Société, 95, 11-26.
de Vries, B., Davis, C. G., Wortman, C. B., & Lehman, D. R. (1997). Long-term psycho-
logical and somatic consequences of later life parental bereavement. Omega: Journal
of Death and Dying, 35, 97-117.
de Vries, B., & Johnson, C. L. (2002). The death of friends in later life. In R. A. Settersten,
& T. J. Owens (Eds.), Advances in life course research, Vol. 7: New frontiers in
socialization (pp. 299-324). Oxford: Elsevier.
de Vries, B., Lehman, A. J., & Arbuckle, N. W. (1995, November). Reactions to the death
of a close friend in later life. Paper presented at the annual meeting of the Geron-
tological Society of America, Los Angeles.
Doka, K. J. (1989). Disenfranchised grief: Recognizing hidden sorrow. New York:
Lexington Books.
Dutton, Y., & Zisook, S. (2005). Adaptation to bereavement. Death Studies, 29(10),
877-903.
Elder, G. H. J. (1998). The life course and human development. In R. M. Lerner (Ed.),
Handbook of child psychology, Vol. 1: Theoretical models of human development
(pp. 939-991). New York: Wiley.
Ferraro, K. F., Mutran, E., & Barresi, C. M. (1984). Widowhood, health, and friendship
support in later life. Journal of Health and Social Behavior, 25, 245-259.
Fitzpatrick, T. R. (1998). Bereavement events among elderly men: The effects of stress
and health. Journal of Applied Gerontology, 17(2), 204-228.
Gallagher, D., Breckenridge, J., Thompson, L. W., & Peterson, J. A. (1983). Effects of
bereavement on indicators of mental health in elderly widows and widowers. Journal
of Gerontology, 38, 565-571.
Gallagher, S. K., & Gerstel, N. (1993). Kinkeeping and friend keeping among older
women: the effect of marriage. The Gerontologist, 33, 675-681.
Groupe “SOL” de l’Université du 3e âge de Genève. (1996). Vivre sans elle. Le veuvage au
masculine [To live without it. The widowhood of men]. Genève: Georg.
Hansson, R. O., Hayslip, B., & Stroebe, M. S. (2007). Grief and bereavement In
J. A. Blackburn, & C. N. Dulmus (Eds.), Handbook of gerontology: Evidence-based
approaches to theory, practice, and policy. Hoboken, NJ: John Wiley & Sons.
322 / d’EPINAY, CAVALLI, AND GUILLET
Hays, J. C., Gold, D. T., & Pieper, C. F. (1997). Sibling bereavement in late life. Omega:
Journal of Death and Dying, 35(1), 25-42.
Johnson, C. L. (1994). Differential expectations and realities: Race, socioeconomic status
and health of the oldest old. International Journal of Aging and Human Development,
38(1), 13-27.
Johnson, C. L., & Barer, B. M. (1997). Life beyond 85 years. The aura of survivorship.
New York: Springer.
Johnson, C. L., & Catalano, D. J. (1981). Childless elderly and their family support. The
Gerontologist, 21(6), 610-618.
Kane, R. L., Bell, R., Riegler, S., Wilson, A., & Kane, R. A. (1983). Assessing the outcomes
of nursing home patients. Journal of Gerontology, 38(4), 385-393.
Katz, S., Downs, T. D., Cash, H. R., & Grotz, R. C. (1970). Progress in development of
the index of ADL. The Gerontologist, 10(1), 20-30.
Lalive d’Epinay, C. J., Bickel, J.-F., Maystre, C., & Vollenwyder, N. (2000). Vieillesses
au fil du temps. 1979-1994: Une révolution tranquille [Old age across the years.
1979-1994: A quiet revolution]. Lausanne: Réalités Sociales.
Lalive d’Epinay, C. J., Cavalli, S., & Spini, D. (2003). The death of a loved one: Impact on
health and relationships in very old age. Omega: Journal of Death and Dying, 47(3),
265-284.
Lalive d’Epinay, C. J., Christe, E., Coenen-Huther, J., Hagmann, H.-M., Jeanneret, O.,
Junod, J.-P., et al. (1983). Vieillesses: Situations, itinéraires et modes de vie des
personnes âgées aujourd’hui [Old ages. Life conditions, life trajectories and life
styles]. Saint-Saphorin: Georgi.
Lalive d’Epinay, C. J., Guilley, E., Guillet, L. A., & Spini, D. (2008). The Swiss Inter-
disciplinary Longitudinal Study on the Oldest Old: Design and population. In
E. Guilley & C. J. Lalive d’Epinay (Eds.), The closing chapters of long lives: Results
from the 10-year SWILSOO study on the oldest old (pp. 9-26). New York: Nova Science.
Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Self-maintaining
and instrumental activities of daily living. The Gerontologist, 9, 179-186.
Lee, G. R., Willets, M. C., & Seccombe, K. (1998). Widowhood and depression: Gender
differences. . Research on Aging, 20, 611-630.
Lowenthal, M. F., Thurner, M., & Chiriboga, D. A. (1975). Four stages of life. San
Francisco, CA: Jossey-Bass.
Matthews, M. A. (1991). Widowhood in later life. Toronto and Vancouver: Butterworths.
Matthews, S. H. (1986). Friendship in old age: Biography and circumstances. In V. W.
Marshall (Ed.), Later life: The social psychology of aging (pp. 233-266). Beverly Hills,
CA: Sage.
Moss, M. S., Lesher, E. L., & Moss, S. Z. (1986). Impact of the death of an adult child on
elderly parents: Some observations. Omega: Journal of Death and Dying, 17(3),
209-218.
Moss, M. S., & Moss, S. Z. (1989a). Death of the very old. In K. J. Doka (Ed.),
Disenfranchised grief: Recognizing hidden sorrow. Lexington: Lexington Books.
Moss, M. S., & Moss, S. Z. (1989b). The impact of the death of an elderly sibling: Some
considerations of a normative loss. American Behavioral Scientist, 33(1), 94-106.
Moss, M. S., Moss, S. Z., & Hansson, R. O. (2001). Bereavement and old age. In M. S.
Stroebe, R. O. Hansson, W. Stroebe, & H. Schut (Eds.), Handbook of bereavement
research. Washington: American Psychological Association.
BEREAVEMENT IN VERY OLD AGE / 323
Ott, C. H., Lueger, R., Kelber, S., & Prigerson, H. (2007). Spousal bereavement in older
adults: Common, resilient and chronic grief with defining characteristics. Journal
of Mental and Nervous Diseases, 195(4), 332-341.
Parkes, C. M. (1972). Bereavement: Studies of grief in the adult life. New York: Inter-
national Universities Press.
Pearlin, L. I. (1994). The study of the oldest-old: Some promises and puzzles. International
Journal of Aging and Human Development, 38(1), 91-98.
Pearlin, L. I., Menaghan, E. G., Lieberman, M. A., & Mullan, J. T. (1981). The stress
process. Journal of Health and Social Behavior, 22, 337-356.
Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., & Congdon, R. T., Jr. (2004). HLM 6:
Hierarchical linear and nonlinear modelling (6th ed.). Lincolnwood, IL: Scientific
Software International.
Roberto, K. A., & Stanis, P. I. (1994). Reactions of older women to the death of their
close friends. Omega: Journal of Death and Dying, 29(1), 17-27.
Rogers, M. P., & Reich, P. (1988). On the health consequences of bereavement. New
England Journal of Medicine, 319, 510-512.
Rook, K. S. (2000). The evolution of social relationships in later adulthood. In S. H. Qualls,
& N. Abeles (Eds.), Psychology and the aging revolution. How we adapt to longer
life (pp. 173-191). Washington, DC: American Psychological Association.
Shimai, S. (2004). Bereavement experience in the general population: incidence, con-
sequences, and coping in a national sample of Japan. Omega: Journal of Death and
Dying, 48(2), 137-147.
Sklar, F., & Hartley, S. F. (1990). Close friends as survivors: Bereavement patterns in a
hidden population. Omega: Journal of Death and Dying, 21, 103-112.
Stroebe, M. S., & Stroebe, W. (1993). The mortality of bereavement: A review. In
M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement
(pp. 175-195). New York: Cambridge University Press.
Swiss Federal Statistical Office. (2000). The 2000 population census. Bern: Swiss
Federal Statistical Office.
Thierry, X. (1999). Risques de mortalité et de surmortalité au cours des dix premières
années de veuvage [Risks of mortality and excess mortality in the first ten years of
widowhood]. Population, 54(2), 177-204.
Thierry, X. (2000). Mortel veuvage. Risques de mortalité et causes médicales des décès aux
divers moments du veuvage [Mortal widowhood. Mortality risk and medical causes
of death at the various stages of widowhood]. Gérontologie et Société, 95, 27-45.
Tudiver, F., Hilditch, J., Permaul, J. A., & McKendree, D. J. (1992). Does mutual help
facilitate newly bereaved widowers? Evaluation and the Health Professions, 15,
147-162.
Umberson, D., Wortman, C. B., & Kessler, R. C. (1992). Widowhood and depression:
Explaining long-term gender differences in vulnerability. Journal of Health and Social
Behavior, 33, 10-24.
Wang, R. I., Treul, S., & Alverno, L. (1975). A brief self-assessing depression scale.
Journal of Clinical Pharmacology, 15(2-3), 163-167.
Wenger, G. C., Davies, R., Shatahmasebi, S., & Scott, A. (1996). Social isolation and
loneliness in old age: Review and model refinement. Ageing and Society, 16(3),
333-358.
324 / d’EPINAY, CAVALLI, AND GUILLET
Williams, B. R., Baker, P. S., Allman, R. M., & Roseman, J. M. (2006). The feminization
of bereavement among community-dwelling older adults. Journal of Women and
Aging, 18(3), 3-18.
Direct reprint requests to:
Stefano Cavalli
Center for Interdisciplinary Gerontology
Route de Drize 7
Site de Battelle
CH-1227 Carouge-Geneva
Switzerland
e-mail: stefano.cavalli@unige.ch
BEREAVEMENT IN VERY OLD AGE / 325