Copyright 1997 byThe Cerontological Society of America

The CerontologistVol. 37, No. 5, 666-674

The appropriateness of the stress-coping and life transitions models for late life experienceswas assessed by examining psychosocial correlates of spouse caregiving in association with

widowhood. Data were from the Health Status of Older People (HSOP) survey of 1,000people aged 65 and over living in the community in Melbourne, Australia. Caregiving was

found to be a usual precursor to widowhood and may have prepared older people forwidowhood. However, these two life experiences had different consequences. While

widowhood appears to be the more severe transition, caring for a spouse often seriouslythreatens the meaningfulness of life.

Key Words: Life transitions, Health behaviors, Psychological well-being, Sense of coherence

Health and Well-Being of Spouse Caregiversand the Widowed1

Yvonne D. Wells, PhD2 and Hal L. Kendig, PhD:

Providing care for a spouse and widowhood aremajor experiences of later life. Many studies con-clude that caregiving is stressful, particularly forspouses (e.g., Caserta, Lund, Wright, & Redburn,1987; Gilhooly, 1984). There is a well-established lit-erature on widowhood and its specific and globalconsequences (e.g., Avis, Brambilla, Vass, & McKin-lay, 1991). Caregiving now dominates the geronto-logical literature, while there has been little recentinterest in widowhood. Few studies have comparedthe two transitions or explored the effects of priorcaregiving on the experience of widowhood.

This article contrasts the capacity of two models— the stress-coping model and the life transitionsapproach — to account for similarities and differ-ences in caring for a spouse and being widowed.Both of these later life experiences were examinedin a survey broadly representative of older peopleliving in the community in an Australian city. Thecomprehensive study avoided a focus on negativeoutcomes and included specific health behaviors aswell as global measures of health and well-being.

Literature Review

Variants of the stress-coping model have pro-vided a useful basis for understanding the effects ofundesirable life events. In the ABCX version of thismodel (see Biegel, Sales, & Schulz, 1991), an exter-nal stressor (A) interacts with the family's ability tocope with the event (B) and with the family's per-

1This research was supported by funding from the Victorian Health Pro-motion Foundation and the Australian Research Council and was under-taken while the first author was funded by an award from the NationalHealth and Medical Research Council. This article is a substantially revised

. version of a paper presented at the 1996 National Conference of the Aus-tralian Association of Gerontology, Hobart, Tasmania, May 1996.

2Lincoln Gerontology Centre, School of Public Health, La Trobe Uni-versity, Bundoora, Australia 3083. Address correspondence to Dr. YvonneWells.

ceptions of the stressor (C). Health crises and men-tal health problems (X) may result. The DoubleABCX model is similar to its predecessor but allowsfor a "pil ing up" of stressors over time (Rankin,Haut, & Keefover, 1992). The effects of stress may betime-lagged (Biegel et alv 1991, p. 9).

Stress-coping studies have viewed stress as uni-tary and cumulative, and have focused primarily onglobal outcomes, such as physical illness and de-pression. Relatively little attention has been paid tofavorable life events, except as potential furthersources of stress, or to positive outcomes such aslife satisfaction. Further, specific behavioral out-comes, such as changes in physical and social activ-ity, deserve attention and could well differ for differ-ent stressful events. For example, caring for a spouserestricts life styles (Aneshensel, Pearlin, & Schuler,1993), whereas widowhood may represent a releasefrom obligations (Aldersberg & Thorne, 1990).

Caregiving research has relied heavily on versionsof the stress-coping model (Biegel et al., 1991). Incontrast, although widowhood is rated the moststressful of all life events by Holmes and Rahe(1967), research on widowhood has relied more onconcepts such as transition and adjustment. Transi-tions are permanent, sudden life changes that havemajor impact and initiate a period of dislocationand adjustment at both the individual and networklevel. Major transitions, such as retirement and wid-owhood, challenge important assumptions peoplehold about the world and their place in it (McCal-lum, 1986).

Global outcomes associated with caregiving in-clude ill health, depression, and anxiety (see reviewby Schulz, O'Brien, Bookwala, & Fleissner, 1995).Caregiving research on specific outcomes has exam-ined the impact on employment and social life (e.g.,Schulz & Williamson, 1993). Little attention has beenpaid to other possible impacts, such as the capacity

666 The Gerontologist

to engage in a healthy life style, even though healthbehaviors could mediate between caregiving stres-sors and global outcomes. Connell (1994) reportedthat caring for a person with dementia restricts pre-ventive health behaviors, but other studies havefound no such link (Schulz et al., 1995). Although al-cohol overuse is implicated in elder abuse (Pille-mer, 1986), few links have been established be-tween caregiving and other health behaviors.

The conclusion that caregiving is universallystressful can be questioned for three main reasons.First, most studies are based on selective samples ofcaregivers who seek help from service providers orself-help groups (Barer & Johnson, 1990). Caregiverswho seek help are likely to have the greatest needand to report negative consequences of caregivingdisproportionately (MaloneBeach & Zarit, 1991;Matthews, 1988). Second, research on family carehas focused heavily on caring for someone with de-mentia, which is more stressful than caring forsomeone with a physical disability (Birkel, 1987; Mo-hide, Torrance, Streiner, Pringle, & Gilbert, 1988) oran intellectual disability (Whittick, 1988). Third, moststudies of caregiving cue participants to focus onthe most negative aspects of their experiences.Some caregivers, however, report that the experi-ence makes them feel closer to their dependent rel-ative (Hinrichsen, Hernandez, & Pollack, 1992; Wil-son, 1990). Caregivers may gain satisfaction andself-esteem from care provision (O'Bryant, Straw, &Meddaugh, 1990; Wright, 1991).

Almost all research on widowhood in later life hasfocused on women (Bengtson, Rosenthal, & Burton,1990). The death of a spouse may trigger changes indaily routines, such as those associated with foodpreparation and consumption (Rosenbloom & Whit-tington, 1993). Widowhood brings about changes insocial participation (Lopata, 1979). Loneliness is thebiggest problem reported by both widows and wid-owers (Lund, Caserta, & Dimond, 1986).

Widowed men and women report lower life satis-faction and poorer psychological well-being thantheir married counterparts and they are more likelyto die (Arbuckle & de Vries, 1995; Bennett & Mor-gan, 1992; Parkes, 1986). However, widowhood canbe a positive transition for some women (Alders-berg & Thorne, 1990). Widowed people often ex-press greater self-efficacy than married adults (Ar-buckle & de Vries, 1995), and most widows andwidowers adjust successfully in time (McCallum,1986).

Studies of entry into and exit from caregiving arelimited almost entirely to adult children of olderpeople (e.g., Dwyer, Henretta, Coward, & Barton,1992; Walker & Pratt, 1991). Yet, older spouses arelikely to have a different caregiving history. Themost common conclusion of spouse caregivingis institutionalization of the care recipient (e.g.,Pruchno, Michaels, & Potashnik, 1990), which hasbeen likened to "quasi-widowhood" (Rosenthal &Dawson, 1991). Widowhood is the second majorway by which spouse caregiving ends (Pruchno etal., 1990), but little attention has been given to the

caregiver-to-widowhood transition, although Bassand Bowman (1990) found that a greater caregivingburden predicts a greater bereavement strain.

Conversely, few widowhood studies have takenprior caregiving experience into account. Suddenbereavement is a risk factor for more severe be-reavement reactions (Hill, Thompson, & Gallagher,1988; Smith, 1978). Widows who anticipate the deathof their spouse may engage in anticipatory grievingthat minimizes regrets and prevents "unfinishedbusiness" (Lopata, 1986); however, older peoplewhose spouses suffer long illnesses can be vulnera-ble in widowhood (Gerber, Rusalem, Hannon, Bat-tin, & Arkin, 1975). Caring for a spouse, then, may ei-ther provide time to prepare for widowhood orserve as a cumulative stressor, depending on theduration of the caregiving experience.

The life transitions model provides a view thatcan take into account the relationship betweencaregiving and widowhood. Spouse caregiving canbe viewed as a transition that may precede widow-hood and influence the meaning of widowhood.Further, the life transitions model permits the inter-pretation of caregiving as being a source of satisfac-tion as well as stress, and of widowhood as a releaseas well as a loss.

In summary, the consequences of spouse caregiv-ing in relation to widowhood are not well under-stood at present. The stress-coping model and thelife transitions model provide alternative perspec-tives on these major late-life experiences.

MethodThis study aimed to test the stress-coping and life

transitions models as competing predictors of theconsequences of spouse caregiving and widow-hood. It examined specific as well as global conse-quences, including health behaviors, strain, andglobal well-being. Hypotheses were tested in thefollowing three areas.

1. Consequences of spouse caregiving. Thestress-coping model predicts that current and for-mer spouse caregivers will report more negativeoutcomes (specific and global) than people whohave not been spouse caregivers. In contrast, thelife transitions model predicts that both positiveand negative outcomes may be associated withcaregiving experience.

2. Consequences of widowhood. The stress-cop-ing model predicts that widows and widowers willreport more negative outcomes (specific and global)than people who are married. In contrast, the lifetransitions model suggests that widowhood may en-tail both positive and negative consequences.

3. Comparison of widowhood and spouse care-giving experience.

(a) More negative consequences will be associ-ated with widowhood than with spouse caregiving,because widowhood is likely to require more ad-justment than becoming a spouse caregiver.

(b) The stress-coping model predicts that theconsequences of the two experiences will be simi-

Vol. 37, No. 5,1997 667

lar, while the life transitions approach permits themto entail different consequences.

(c) The stress-coping model predicts that the twostressors, widowhood and spouse caregiving, willoperate cumulatively, whereas the life transitionsapproach permits them to operate interactively.

SampleParticipants were drawn from the Health Status of

Older People survey of 1,000 people aged 65 yearsand older living in the community (see Kendig et al.,1996). The survey was conducted in 1994 by the Lin-coln Gerontology Centre in collaboration with theNational Ageing Research Institute. An area proba-bility sample was designed by the Australian Bureauof Statistics. The eligible population consisted ofresidents of private dwellings in the Melbourne Sta-tistical Division, thus excluding older people livingin hospitals, nursing homes, and hostels. Individu-als were also ineligible if they did not speak basicEnglish, were cognitively impaired, severely ill, ordeaf. The response rate was 70% of the eligible pop-ulation. In addition to personal interviews in thehome, questionnaires were left for self-completionand return by mail. Eighty-five percent of surveyparticipants returned a usable questionnaire, andthe current study includes only this subsample.Comparisons with comparable community surveysindicate that the HSOP sample comprised slightlyhigher than expected proportions of healthy andmarried people (Kendig et al., 1996).

Four subgroups were identified for the purposesof the analyses. Croup 1 comprised married peoplewho had not been spouse caregivers (married non-caregivers), while Group 2 included married peoplewho were currently providing spousal care (marriedcurrent caregivers). Widows and widowers were in-cluded in the study if they had been widowed forfour years or less. Group 3 comprised widowed peo-ple who had not been spouse caregivers (widowednever caregivers), while widowed people who for-merly were spouse caregivers formed Group 4 (wid-owed former caregivers). People who were marriedbut not living with their spouse were omitted. It wasnot possible to determine how many widowed peo-ple had formerly institutionalized their spouse.

The size, gender balance, and age distribution ofeach group are presented in Table 1. As expected,married current caregivers were older than marriednoncaregivers. Widowed people were older than

married people and included a higher proportion ofwomen. Men were more equally representedamong the married current caregivers (Group 2)than is usual in studies of caregiving.

MeasuresMarried current and widowed former caregivers

replied "yes" to the item, "Since the age of 60 haveyou had the main responsibility in caring for some-one who has a long-term illness, disability, or otherproblem?" People who had been widowed wereidentified by an item on marital status, with a fol-low-up question on duration of widowhood.

Health, life style, and well-being variables wereselected to provide a comprehensive assessment ofthe health status of older people (Teshuva, Stan-islavsky, & Kendig, 1994). Following are the specificand global measures used in the analyses (seeKendig et al., 1996, for specification of the surveyquestions). Internal reliability for scales quotedbelow includes all survey participants and employsCronbach's alpha:

Specific Activity Measures. — Participants ratedtheir current way of life on a 3-point scale from veryhealthy to not so healthy. Level of physical and socialactivity compared with others of the same age wereboth rated from 1 (more active) to 3 (less active).

Specific Health Practices. — Eating patterns wererated from 1 (very healthy) to 3 (not so healthy) andappetite during the past month was rated from 1(very good) to 4 (poor). Smoking was designated yesor no, and alcohol consumption was rated from 1(more than once a day) to 7 (never). Trouble fallingasleep, waking at night, and feeling rested in themorning were all rated on 4-point scales from mostof the time to never. Use of psychotropics wasscored yes or no for each of four domains: anal-gesics, antidepressants, tranquilizers, and hypnotics.

Global Measures of Well-Being. — These wereidentified primarily from previous caregiving re-search. Participants rated their physical health from1 (excellent) to 5 (poor). Feeling lonely and boredwere each rated from 1 (never) to 5 (very fre-quently). Strain was measured using a single item,"How often do you feel that you are under so muchstrain that your health is likely to suffer?," ratedfrom 1 (never) to 5 (very frequently).

Croup

1234

Table 1.

Designation

Married noncaregiverMarried current caregiverWidowed never caregiverWidowed former caregiver

Size, Gender Balance, and Age Distribution of Groups (%)

Men

65-74

46338

21

75+

2016176

Women

65-74

26335052

75+

9192521

%

100100100100

Total

n

456432433

Note: Row percentages may not sum to 100 due to rounding.

668 The Gerontologist

Lawton's Affect Balance Scale. — This was used tomeasure positive and negative affect (Lawton, Kle-ban, Dean, Rajogopal, & Parmelee, 1992). Each sub-scale includes five items, and internal reliability was.74 for both. Depression was measured using thePsychogeriatric Assessment Scales depression sub-scale (Jorm & Mckinnon, 1994). This subscale com-prises 12 items; internal reliability was .67. Becausescores on the depression scale were highly posi-tively skewed, they were treated with a log transfor-mation; participants were also categorized as likely(score of 5 or above) or not likely to be depressed.

Life Satisfaction. — This was measured by sum-ming ratings over eight domains (health, financial sit-uation, friendships, marriage, family life, neighbor-hood, ability to handle problems, and life in general),each from 1 (extremely satisfied) to 5 (very dissatis-fied). Items were selected from the domain life satis-faction measure (Campbell, Converse, & Rodgers,1976) on the basis of their likely salience for olderAustralians. Internal reliability was .87. Interviewersalso rated participants' quality of life (5-point scalefrom highly meaningful to impoverished).

Self-efficacy. — This was measured using scalesby Seeman, Rodin, and Albert (1993). The 3-item in-terpersonal efficacy subscale had an internal relia-bility of .79, while the 4-item instrumental efficacysubscale had an internal reliability of .84. Anto-novsky's (1984) Sense of Coherence subscales mea-sured respondents' view of life's Comprehensibility(5 items, alpha = .65), Manageability (4 items, alpha =.61), and Meaningfulness (4 items, alpha = .57).

Analysis

Analyses assessed relationships between key lifeexperiences (widowhood and spouse caregiving)and outcomes (specific and global), employing a fac-torial design. Factor 1 differentiated those withoutspouse caregiving experience from both marriedcurrent caregivers and widowed former caregivers.Therefore, variables associated with this factor indi-cated enduring consequences of spouse caregivingor selection effects in becoming a spouse caregiver.Factor 2 differentiated widowed from married peo-ple. Interactions between factors could indicate ei-ther that consequences associated with spouse care-giving ceased (or were exacerbated) on widowhoodor that the consequences of being widowed differedaccording to prior caregiving experience.

The chi-square statistic was employed with cate-gorical dependent variables. Categories on someitems were combined to meet requirements for ex-pected frequencies. To control for age group andgender, chi-square analyses were weighted so thatall groups corresponded with the size and propor-tions in age and gender represented by the marriedcurrent caregivers (Group 2). This group was cho-sen as the reference group for weighting because(a) men and women were almost equally repre-sented; (b) the proportions in the young-old and

old-old groups reflected proportions for the wholesample (65% young-old in Group 2; 64% young-oldfor the whole sample); and (c) the number of partic-ipants (43) was sufficient for the analyses withoutunduly inflating their power. Hence, for these anal-yses the number of widowed participants was in-creased (from 57 to 86), but the number of currentlymarried people was substantially reduced (from 499to 86).

Interaction effects (Factor 1 X Factor 2) were as-sessed using logit analysis, which is a method oftesting relationships between three or more cate-gorical variables. The logit model starts with all ofthe one-, two-, and three-way associations and theneliminates as many of them as possible while main-taining an adequate fit between expected and ob-served cell frequencies (Tabachnick & Fidell, 1989).

Interval-level dependent variables were analyzedusing partial correlation, controlling for gender andage. (Regression analysis was chosen rather thananalysis of variance because the research designwas nonexperimental; Keppel & Zedeck, 1989.)Small numbers of outliers were deleted. Interactioneffects (Factor 1 x Factor 2) were assessed using par-tial correlation, controlling for gender, age, Factor 1,and Factor 2.

Results

Percentages and mean scores cited below areweighted for gender and age.

Outcomes Associated With Spouse CaregivingIn terms of specific outcomes, Table 2 shows that

people with spouse caregiving experience weremore likely to take tranquilizers (18% vs 7%). How-ever, married current and widowed former care-givers were only one sixth as likely to smoke as themarried noncaregivers and widowed never care-givers (3% vs 19%). Specific outcomes not associ-ated with spouse caregiving included healthy lifestyle, diet, sleep, and levels of both physical and so-cial activity.

Associations between global outcomes and care-giving experience (see Table 3) were also mixed.Married current and widowed former caregiversrated their health as better than the noncaregivinggroups (married noncaregivers and widowed nevercaregivers): 22% of the current and former care-givers reported being in excellent health, as op-posed to only 12% of the noncaregiving groups.However, the married current and widowed formercaregivers were more likely than the noncaregivinggroups to report feeling so much strain that theirhealth was likely to suffer - 45% vs 25%. Currentand former caregivers had lower mean scores for allthree Sense of Coherence subscales than the non-caregiving groups. Adjusted means were: on Com-prehensibility, 25.88 (SD = 5.49) for current andformer caregivers vs 27.35 (SD = 5.44) for noncare-giving groups; on Manageability, 20.89 (SD = 5.12) vs22.01 (SD = 4.43), respectively; and on Meaningful-

Vol. 37, No. 5,1997 669

Table 2. Correlates of Spouse Caregiving Experience: Specific Outcomes

Set of Outcomes

Activity measures

Health practices

Medication use

Measure

Healthy life stylePhysical activitySocial activityAppetiteHealthy dietDrinks alcoholSmokesFalling asleepWaking at nightFeels restedAnalgesicsAntidepressantsTranquilizersHypnotics

Statistic3

X2

X2

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

Value

1.983.180.693.530.123.01

10.96***2.861.300.320.322.185.49*3.16

df

12221313311111

Outcome

Better

Worse

"Chi-square analyses are weighted to take account of gender and age group (n = 172).*p<.05;***p<.001.

Table 3. Correlates of Spouse Caregiving Experience: Global Outcomes

Set of Outcomes

Social integration

Physical healthMental health

Quality of life

Self-efficacy

Sense of coherence

Measure

Feels boredFeels lonelySelf-rated healthFeels strainDepression (scale)Depression (category)Positive affectNegative affectLife satisfactionObserved qualityInterpersonalInstrumentalComprehensibilityManageabilityMeaningfulness

Statistic3

x2

x2

x2

x2

r

x2

rrr

x2

rrrrr

Value

3.064.569.66*7.16*.02.31

-.06-.01

.042.85

.04

.02-.11*-.10*-.09*

df

2432

5521

550552531

3513515520526528

Outcome

BetterWorse

WorseWorseWorse

"Chi-square analyses are weighted to take account of gender and age group (n = 172), while partial correlation coefficients (r) controlfor age and gender.

*p < .05.

ness, 21.77 (SD = 4.39) vs 23.02 (SD = 4.18), respec-tively. Global outcomes not associated with Factor 1included social integration, self-efficacy, affect, andquality of life.

The first hypothesis generated by the stress-copingmodel, that spouse caregiving would entail negativeoutcomes, was partly supported. Positive outcomesassociated with caregiving experience were moreconsistent with the life transitions model.

Outcomes Associated With Widowhood

Relative to married people, people who had beenwidowed within the previous four years showedpoorer appetite (32% vs 10% with fair-to-poor ap-petite), more smoking (17% vs 5%), and higher useof hypnotics (18% vs 7%) (Table 4). Specific out-comes not associated with widowhood included ac-tivity levels, alcohol use, and sleep patterns.

With respect to global outcomes (see Table 5),widows and widowers relative to married peopleshowed more loneliness (33% vs 2% frequently or

very frequently lonely), more boredom (19% vs 8%frequently or very frequently bored), and poorerhealth (37% vs 58% excellent or very good health).Widowed people also reported more symptoms ofdepression than married people (Ms 1.77 and 1.20,respectively; SDs 1.92 and 1.68, respectively) andless positive affect (Ms 19.37 and 20.55; SDs 2.91 and2.65). However, widows and widowers reported lessstrain than married people (72% vs 56% never feltstrain). Quality of life, self-efficacy, and sense of co-herence were not associated with widowhood. Thesecond hypothesis generated by the stress-copingmodel, that negative outcomes would be associatedwith widowhood, was supported overall.

Comparison of Widowhood and SpouseCaregiving Experience

Spouse caregiving experience and widowhoodwere not independent. On the one hand, currentspouse caregivers were relatively rare, comprisingonly 4% of the HSOP sample and 9% of those who

670 The Gerontologist

Table 4. Correlates of Widowhood: Specific Outcomes

Set

Activity measures

Health practices

Medication use

Measure

Healthy life stylePhysical activitySocial activityAppetiteHealthy dietDrinks alcoholSmokesFalling asleepWaking at nightFeels restedAnalgesicsAntidepressantsTranquilizersHypnotics

Statistic3

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

x2

Value

0.211.682.43

16.28***0.912.386.37*5.174.951.512.741.430.535.35*

df

12221313311111

Outcome

Worse

Worse

Worse

'Chi-square analyses are weighted to take account of gender and age group (n = 172).*p<.05;***p<.001.

Table 5. Correlates of Widowhood: Global Outcomes

Set

Social integration

Physical healthMental health

Quality of life

Self-efficacy

Sense of coherence

Measure

Feels boredFeels lonelySelf-rated healthFeels strainDepression (scale)Depression (category)Positive affectNegative affectLife satisfactionObserved qualityInterpersonalInstrumentalComprehensibilityManageabilityMeaningfulness

Statistic"

X2

X2

X2

X2

r

x2

rrr

x2

rrrrr

Value

6.98*57.81***7.85*6.04*

.11**

.82-.14***-.01

.083.68-.03-.01-.06

.03-.04

df

2432

5521

550551531

3515511520526528

Direction

WorseWorseWorseBetterWorse

Worse

•Chi-square analyses are weighted to take account of gender and age group (n = 172), while partial correlation coefficients (r) controlfor age and gender.

*p<.05;**p<.01;***p<.001.

were married. On the other hand, 58% of the wid-owed people were former spouse caregivers.

(a) Hypothesis: More negative consequences willbe associated with widowhood than with spousecaregiving.

This hypothesis was supported. A simple count ofsignificant positive and negative associations re-vealed that widowhood appeared to have some-what worse consequences for people than spousecaregiving. Negative consequences of widowhoodwere comprehensive and included health behav-iors, social integration, physical health, and mentalhealth. Negative consequences of spouse caregiv-ing were confined to strain and sense of coherence.

(b) Similarity of spouse caregiving and widow-hood.

The consequences of widowhood and spousecaregiving were not similar. Only three variableswere associated with both of these factors, and for

these variables the relationships between stressorsand outcomes were in opposite directions: Currentand former spouse caregivers reported lower ratesof smoking and better health despite more strainthan noncaregivers, whereas widowed people re-ported higher rates of smoking and worse healthbut less strain than married people. These resultsare more consistent with the life transitions modelthan with the stress-coping model.

(c) Spouse caregiving and widowhood as cumula-tive or interactive.

Interaction effects between Factor 1 and Factor 2were examined for two reasons: to detect furtherdifferences between groups and to examine nonad-ditivity of factors. Such nonadditivity would providesupport for the life transitions model rather thanthe stress-coping model. Interaction effects weredetected only for "trouble falling asleep," the cate-gorical depression variable, and the probability oftaking tranquilizers. Widowed former caregivers

Vol. 37, No. 5,1997 671

had the most trouble falling asleep of any group:39% had trouble falling asleep most of the time,compared with 14% to 18% for the other threegroups. The widowed former caregivers were alsomost likely to be taking tranquilizers (21%, com-pared with 0% to 8% for the other groups). High useof tranquilizers in widowed former caregivers ac-counted for the bivariate relationship betweenspouse caregiving experience and tranquilizer use:Married current caregivers were no more likelythan any other group to use these drugs.

Widowed people with no caregiving experiencewere more likely than any other group to be classi-fied as depressed: 18%, compared with 12% for mar-ried current caregivers, 7% for widowed formercaregivers, and 5% for married noncaregivers.Hence, widowed people were about twice as likelyto be classified as depressed if they had no spousecaregiving experience. This supports a buffering ef-fect of spouse caregiving experience, as predictedby the life transitions model.

Discussion

This study compared the stress-coping and lifetransitions models in predicting the consequencesof two life events — widowhood and spouse care-giving. The results show that it is useful to examinespouse caregiving and widowhood together. Spousecaregiving is a transition that is closely linked withwidowhood: More than half of the widowed peoplewere former spouse caregivers, and both life expe-riences had enduring consequences.

The general stress model does not apply equallyto spouse caregiving and widowhood because theoutcomes of these life events were different andsometimes contrary. Some outcomes showed inter-active rather than additive effects. In regard to de-pression, former spouse caregivers appeared to beprovided with a measure of protection against thestress of widowhood. That result supports thenotion that people who are forewarned of theirspouse's death may adjust better to widowhood(e.g., Lundin, 1984). Conversely, in regard to sleep-ing problems and the use of tranquilizers, apparentdifficulties of widowed people were exacerbated byprior caregiving experience.

The outcomes associated with widowhood werestrongly negative and pervasive across aspects ofhealth and well-being. These results closely parallelthose of other widowhood studies (e.g., Brecken-ridge, Gallagher, Thompson, & Peterson, 1986) andconform well to the general stress model. They arealso consistent with the life transitions model, be-cause adjustment is likely to continue up to fouryears after widowhood (Lund, Caserta, & Dimond,1986; McCallum, 1986). For life events whose conse-quences are largely negative, the stress-coping andlife transitions models are not mutually exclusive.

In contrast, outcomes associated with spousecaregiving experience were fewer than expectedfrom the caregiving literature and not invariablynegative. For example, current and former spouse

caregivers were more likely to report good healthand less likely to smoke than people withoutspouse caregiving experience. Negative outcomesassociated with spouse caregiving were largely lim-ited to sense of coherence, and caregiving ap-peared to have no negative consequences forhealth behaviors, health, affect, or quality of life.

These findings suggest that spouse caregiving isnot best conceptualized simply as a stressor formost people. Rather, spouse caregiving may pro-vide a complex challenge, giving a sense of compe-tence and satisfaction that may compensate for thelosses experienced (O'Bryant et al., 1990; Wright,1991). Our results support Matthews' (1988) criti-cism that too pessimistic a picture of caregiving hasbeen perpetuated. It is overly simplistic to stereo-type older caregivers as necessarily hidden victimsof their spouse's illness.

This cross-sectional study does not, of course,suggest that caregiving improves health or helpspeople give up smoking (although the responsibil-ity of looking after a spouse may motivate people tomaintain health habits). It is more plausible thathealthy people are more likely to take on spousecaregiving in the first place. Alternatively, a personwho provides care may not view him- or herself as acaregiver if he also has poor health and each part-ner relies on support from the other.

Despite a generally encouraging picture, therewere signs that spouse caregivers endured moredifficulties than their counterparts without caregiv-ing experience. They reported higher levels ofstrain and were more likely to use sedatives. Fur-ther, both current and former spouse caregivers ex-perienced life as less meaningful, manageable, andcomprehensible than people who had never beencaregivers. The sense of coherence (Antonovsky,1984) experienced by spousal caregivers appears tobe diminished. This loss of faith in the order andpurpose of life persists after the death of the de-pendent spouse.

We might speculate as to why providing care for aspouse has persistent effects on existential beliefs,while widowhood impacts on affective state. Al-though widowhood is a potentially traumatic transi-tion, it is an expected event for people over the ageof 65 (Lopata, 1986; McCallum, 1986). Conversely,becoming a spouse caregiver, although a less trau-matic transition, may not be anticipated by mostpeople. There is some evidence that middle-agedmen are less likely to anticipate spouse caregivingthan women, and are more idealistic in their expec-tations (Wells, 1993). In contrast to having a spousedie, the realities of spousal caregiving may pro-foundly violate the expectations people have oftheir retirement years. More research on the mean-ing of spouse caregiving is required to extend thefew studies now available (Motenko, 1988; Vinick,1984). Preparation and counseling for spousal care-giving should include a focus on the meaning andpurpose of life.

Findings on medication use are of some concern.Widows, especially those who had not provided

672 The Gerontologist

care for their spouse before their bereavement, re-ported a high level of depressive symptoms, butthey were no more likely than still-married peopleto be receiving antidepressant medication. Widowswere, however, relatively likely to be taking hyp-notics; widows with spouse caregiving experiencewere the group most likely to be taking tranquiliz-ers. These apparent contradictions between depres-sive symptoms and prescription of medication meritfurther study. Health promotion for older peopleneeds to consider the heightened risk of inappro-priate medication use among widows.

The cross-sectional design of the study has limita-tions in distinguishing consequences from selec-tion effects. This is not a major issue for the widow-hood findings because low well-being is unlikely toincrease the risk of widowhood. Some findings as-sociated with spouse caregiving, however, may beinterpreted as selection effects because people withfewer disabilities are more likely to take on theseresponsibilities. Further, even in a study of 1,000people, there were too few present or past spousecaregivers to examine potentially important genderdifferences in the consequences of caregiving (seemeta-analyses by Miller & Cafasso, 1992) or widow-hood. For example, women usually expect to out-live their husbands.

One might question why these findings are sosimilar to those in the widowhood literature butstarkly different from those in the caregiving litera-ture. The explanation may lie mainly in the predom-inance of caregiving studies that focus on the diffi-culties of people who seek help. That emphasisbrings recognition to a genuine social problem andguides the provision of support. However, there is apaucity of population-based studies that also exam-ine positive aspects of caregiving and include repre-sentative samples of people at different points intheir caregiving history.

This study has shown the importance of carefullyconsidering the appropriateness of the stress-cop-ing and life transition models in terms of specificlife experiences and specific consequences. Care-giving and widowhood are usual experiences thathave different but interrelated consequences in oldage. While both experiences can require difficultadjustments, it is important to emphasize that a sat-isfactory level of well-being is achieved eventuallyby most older people.

References

Aldersberg, M., & Thorne, S. (1990). Emerging from the chrysalis: Olderwidows in transition. Journal of Gerontological Nursing, 76,4-8.

Aneshensel, C. S., Pearlin, L I., & Schuler, R. H. (1993). Stress, role captivityand the cessation of caregiving. Journal of Health and Social Behavior,34, 54-70.

Antonovsky, A. (1984). The sense of coherence as a determinant of health.In J. D. Matarazzo, S. M. Weiss, J. A. Herd, N. E. Miller, & S. M. Weiss(Eds.), Behavioral health: A handbook of health enhancement and dis-ease prevention (pp. 114-129). New York: Wiley.

Arbuckle, N. W., & de Vries, B. (1995). Long-term effects of later lifespousal and parental bereavement on personal functioning. TheCerontologist, 35, 637-647.

Avis, N. E., Brambilla, D. J., Vass, K., & McKinlay, J. B. (1991). Effect of wid-owhood on health: A prospective analysis from the MassachusettsWomen's Health Study. Social Science and Medicine, 33,1063-1070.

Barer, B. M., & Johnson, C. L. (1990). A critique of the caregiving literature.The Cerontologist, 30, 26-29.

Bass, D. M., & Bowman, K. (1990). Transition from caregiving to bereave-ment: The relationship of care-related strain and adjustment to death.The Gerontologist, 30, 35-42.

Bengtson, V., Rosenthal, C, & Burton, L. (1990). Families and aging: Diver-sity and heterogeneity. In R. H. Binstock & L. K. George (Eds.), Hand-book of aging and the social sciences (3rd ed., pp. 263-287). San Diego,CA: Academic Press.

Bennett, K. M., & Morgan, K. (1992). Health, social functioning, and maritalstatus: Stability and change among elderly recently widowed women.International Journal of Geriatric Psychiatry, 7, 813-817.

Biegel, D. E., Sales, E., & Schulz, R. (1991). Family caregiving in chronic ill-ness. Newbury Park, CA: Sage.

Birkel, R. C. (1987). Toward a social ecology of the home-care household.Psychology and Aging, 2, 294-301.

Breckenridge, J. N., Gallagher, D., Thompson, L W., & Peterson, J. (1986).Characteristic depressive symptoms of bereaved elders. Journal ofGerontology, 41,163-168.

Campbell, A., Converse, P. E., & Rodgers, W. L. (1976). The quality of Ameri-can life. New York: Russell Sage Foundation.

Caserta, M. S., Lund, D. A., Wright, S. D., & Redburn, D. E. (1987). Care-givers of dementia patients: The utilization of community services. TheGerontologist, 27, 209-214.

Connell, C. M. (1994). Impact of spouse caregiving on health behaviorsand physical and mental health status. American Journal of Alzheimer'sCare and Related Disorders and Research, 9, 26-36.

Dwyer, J. W., Henretta, J. C, Coward, R. T., & Barton, A. J. (1992). Changesin the helping behavior of adult children as caregivers. Research onAging, 14,351-375.

Gerber, I., Rusalem, R., Hannon, N., Battin, D., & Arkin, A. (1975). Anticipa-tory grief and aged widows and widowers. Journal of Gerontology, 30,225-229.

Gilhooly, M. L. M. (1984). The impact of care-giving on care-givers: Factorsassociated with the psychological well-being of people supporting adementing relative in the community. British Journal of Medical Psy-chology, 57, 33-44.

Hill, C. D., Thompson, L. W., & Gallagher, D. (1988). Role of anticipatory be-reavement in older women's adjustment to widowhood. The Gerontol-ogist, 28, 792-796.

Hinrichsen, G. A., Hernandez, N. A., & Pollack, S. (1992). Difficulties and re-wards in family care of the depressed older adult. The Gerontologist,32, 486-492.

Holmes, T. H., & Rahe, R. H. (1967). The social readjustment rating scale.Journal of Psychosomatic Research, 11,213-218.

Jorm, A. F., & Mckinnon, A. (1994). Psychogeriatric Assessment Scales:Users' guide and materials. Canberra, Australia: Anutech Pty Ltd.

Kendig, H., Helme, T., Teshuva, K., Osborne, D., Flicker, L., & Browning, C.(1996). Health Status of Older People project: Data from a survey of thehealth and lifestyles of older Australians. Melbourne, Australia: Victo-rian Health Promotion Foundation, Lincoln Gerontology Centre, andNational Ageing Research Institute.

Keppel, G., & Zedeck, S. (1989). Data analysis for research designs. NewYork: Freeman.

Lawton, M. P., Kleban, M. H., Dean, J., Rajagopal, D., & Parmelee, P. A.(1992). The factorial generality of brief positive and negative affectmeasures. Journal of Gerontology: Psychological Sciences, 47,P228-P237.

Lopata, H. Z. (1979). Women as widows: Support systems. New York: Else-vier.

Lopata, H. Z. (1986). Time in anticipated future and events in memory.American Behavioral Scientist, 29, 695-709.

Lund, D. A., Caserta, M. S., & Dimond, M. F. (1986). Gender differencesthrough two years of bereavement among the elderly. The Gerontolo-gist, 26, 314-320.

Lundin, T. (1984). Long-term outcome of bereavement. British Journal ofPsychiatry, 145,424-428.

MaloneBeach, E. E., & Zarit, S. H. (1991). Current research issues in caregiv-ing to the elderly. International Journal of Aging and Human Develop-ment 32,103-114.

Matthews, S. H. (1988). The burdens of parent care: A critical evaluation ofrecent findings. Journal of Aging Studies, 2,157-165.

McCallum, J. (1986). Retirement and widowhood transitions. In H. L.Kendig (Ed.), Ageing and families: A social networks perspective (pp.129-148). Sydney, Australia: Allen & Unwin.

Miller, B., & Cafasso, L. (1992). Gender differences in caregiving: Fact or ar-tifact? The Gerontologist, 32,498-507.

Mohide, E. A., Torrance, G. W., Streiner, D. L, Pringle, D. M., & Gilbert, R.(1988). Measuring the wellbeing of family caregivers using the timetrade-off technique. Journal of Clinical Epidemiology, 41, 475-482.

Motenko, A. K (1988). Respite care and pride in caregiving: The experienceof six older men caring for their disabled wives. In S. Reinhart & G. D.Rowles (Eds.), Qualitative gerontology (pp. 104-127). New York:Springer.

O'Bryant, S. L, Straw, L. B., & Meddaugh, D. I. (1990). Contributions of thecaregiving role to women's development. Sex Roles, 23, 645-658.

Vol. 37, No. 5,1997 673

Parkes, C. P. (1986). Bereavement: Studies of grief in adult life. Har-mondsworth, England: Penguin.

Pillemer, K. A. (1986). Risk factors in elder abuse: Results from a case-con-trol study. In K. A. Pillemer & R. S. Wolf (Eds.), Elder abuse: Conflict inthe family. Dover, MA: Auburn House.

Pruchno, R. A., Michaels, J. E., & Potashnik, S. L. (1990). Predictors of institu-tionalization among Alzheimer disease victims with caregivingspouses. Journal of Gerontology: Social Sciences, 45, S259-S266.

Rankin, E. D., Haut, M. W., & Keefover, R. W. (1992). Clinical assessment offamily caregivers in dementia. The Gerontologist, 32,813-821.

Rosenbloom, C. A., & Whittington, F. J. (1993). Effects of bereavement oneating behaviors and nutrient intakes in elderly widowed persons.Journal of Gerontology: Social Sciences, 48, S223-S229.

Rosenthal, C. A., & Dawson, P. (1991). Wives of institutionalized elderlymen: The first stage of the transition to quasi-widowhood. Journal ofAging and Health, 3, 315-334.

Schulz, R., O'Brien, A. T., Bookwala, J., & Fleissner, K. (1995). Psychiatricand physical morbidity effects of dementia caregiving: Prevalence, cor-relates, and causes. The Gerontologist, 35, 771-791.

Schulz, R., & Williamson, C. M. (1993). Psychosocial and behavioral dimen-sions of physical frailty. Journal of Gerontology, 48 (Special Issue), 39-43.

Seeman, T. E., Rodin, J., & Albert, M. (1993). Self-efficacy and cognitive per-formance in high-functioning older individuals. Journal of Aging andHealth, 5,455-474.

Smith, W. J. (1978). Etiology of depression in a sample of elderly widows: Aresearch report. Journal of Geriatric Psychiatry, 11, 81-83.

Tabachnick, B. C , & Fidell, L. S. (1989). Using multivariate statistics (2nded.). New York: Harper Collins.

Teshuva, K., Stanislavsky, Y., & Kendig, H. (1994). Towards healthy aging.Melbourne, Australia: Collins Dove.

Vinick, B. H. (1984). Elderly men as caregivers of wives. Journal of GeriatricPsychiatry, 17, 61-68.

Walker, A. J., & Pratt, C. C. (1991). Daughters' help to mothers: Intergenera-tional aid versus caregiving. Journal of Marriage and the Family, 53,3-12.

Wells, Y. (1993). Anticipating caregiving: Differences between middle-agedmen and women. Unpublished master's thesis, La Trobe University,Melbourne, Australia.

Whittick, J. E. (1988). Dementia and mental handicap: Emotional distress incarers. British Journal of Clinical Psychology, 27,167-172.

Wilson, V. (1990). The consequences of elderly wives caring for disabledhusbands: Implications for practice. Social Work, 35, 417-421.

Wright, L K. (1991). The impact of Alzheimer's disease on the marital rela-tionship. The Gerontologist, 31, 224-237.

Received September 11,1996Accepted March 27,1997

Springer Publishing CompanyAssessing the Health Status of Older AdultsElena Andresen, PhD, Barbara Rothenberg, MPA,MA and James G. Zimmer, MD, EditorsThis volume reviews widely used methods for measuring thehealth status of older adults and addresses the assets and limitationsof the most useful instruments and procedures. Topics coveredinclude measurement of disease severity and comorbidity,cognitive or mental status screening, and depression screening.1997 288pp 0-8261-9780-9 hard $38.95prepub.

Gerontological Social WorkTheory into Practiceflene L. Nathanson, DSW, BCD and Terry T. Tirrito, DSWThe authors develop a conceptual framework for task-centered,social work practice with older adults. Their integrative modelclearly illustrates how practitioners and students shouldincorporate social work practices in various settings includingpsychiatric, health, religious, legal, occupational, and more.1997 176pp 0-8261-9890-2 $37.95 (outside US $42.80)

A National Agenda for Geriatric EducationWhite PapersSusan M. Klein, DNSC, RN, editorWritten by experts in their fields, this collection of papers,provides a continuous quality improvement approach to geriatricpreparation of the health care workforce. Contained within it arethe educational seeds for a new generation of health care providersserving an aging nation.1997 232pp 0-8261-9970-4 soft $41.95 (outside US $47.80)

Elder Abuse and Neglect, 2nd Ed.Causes, Diagnosis, and Intervention StrategiesMary Joy Quinn, RN, MAand Susan K. Tomita, MSW, PhDIn this new edition of their classicvolume, the authors present their elderabuse diagnosis and interventionmodel. They also include informationon legal interventions with suggestionson how the practitioner should act inthe courtroom, give testimony,document findings, and prepare forlegal involvement with the criminaljustice system.Springer Series on Social Work1997 384pp 0-8261-5121-3 $48.95 (outside US $53.90)

Risk Management in Long-Term CareA Quick Reference GuideAndrew David Weinberg, MD, FACPA practical, effective, and thorough risk management tool,this book helps health professionals address common problemareas in order to avoid litigation. The volume also featuresvaluable information on how to respond to legal claims.Indispensable to administrators, nurses, and physicians, aswell as students of health administration.1998 176 (est) 0-8261-9940-2 $33.95 prepub.

536 Broadway, New York, NY 10012-3955 • (212) 431-4370 • Fax (212) 941-7842

674 The Gerontologist